Financially Interlinked Business Groups

Maitreesh GhatakUniversity of Chicago

Raja KaliSam M Walton College of Business

University of Arkansas Fayetteville AR 72701RKaliWaltonuarkedu

Financial interlinkage in the form of cross-holding of equity and debtbetween rms characterizes business groups in many countries We sug-gest that such nancial interlinkage can be viewed as a way to solve creditrationing caused by asymmetric information If rms possess better informa-tion about each other than a bank then business groups can be a mechanismto induce rms to sort on the basis of this information Banks can offer amenu of contracts that vary in the extent of nancial interlinkage to inducerms to self-select on the basis of the equilibrium composition of the businessgroups they can form

1 Introduction

Business groups feature prominently in the industrial organizationof many countries both developed and developing Reecting thewidespread prevalence of these organizations is the diversity in theway they are dened1 The ties that bind group rms range fromadministrative and nancial linkages to those grounded in family eth-nicity society religion and region While it is the diversity in thesefactors that leads to the subtle external differences between businessgroups in different countries a common internal thread is the inter-linkage in equity and debt among the rms that constitute the groupIt is this kind of nancial interlinkage that is the focus of this paperOur objective here is to present a theory of business groups based onthe cross-holding of equity and debt

We thank two anonymous referees for detailed comments that signicantly improvedthe paper as well as Sugato Bhattacharya Enrico Perotti Tarun Khanna and par-ticipants at the 2000 Financial Market Development for Emerging and TransitionEconomies Conference at LBS for helpful comments We are grateful to the James HPenick endowment for nancial support of this research The usual disclaimer applies

1 The emphasis is on ldquo an intermediate level of binding mdash excluding on the onehand a set of rms bound merely by short-term strategic alliances and on the other aset of rms legally consolidated into a single onerdquo See Granovetter (1994)

copy 2001 Massachusetts Institute of TechnologyJournal of Economics amp Management Strategy Volume 10 Number 4 Winter 2001 591ndash619

592 Journal of Economics amp Management Strategy

The existence of nancial interlinkages between the constituentrms of a business group has been widely noted Previous expla-nations of these interlinkages have focused on the role played bycross-shareholding in either providing risk sharing (see for exampleGoto 1982 Brioschi et al 1989 Nakatani 1984 Kali 1999b) soften-ing intensity of competition between rms in imperfect product mar-kets (see Clayton and Jorgensen 2000) or in mitigating moral-hazardproblems within the group (see Aoki 1982 Berglof and Perotti 1994)

What has been overlooked is the cross-holding of external debtthat often accompanies the cross-holding of equity among the rmsthat constitute the business group Sometimes referred to as cross-payment guarantees or mutual debt guarantees these imply that if amember rm is on the verge of defaulting on an external loan theother group rms will each pay a fraction of the defaulting rmrsquosexternal debt provided they are in a position to do so

Cross-guarantees of this kind are prevalent within businessgroups in several emerging economies In their study of the nanc-ing constraints of Korean chaebols Shin and Park (1999) emphasizethe role played by intragroup cross-guarantees in supporting exter-nal bank lending2 Such practices are also prevalent within Chinesebusiness groups Keisterrsquos (2000) extensive study of Chinese busi-ness groups describes the importance of mutual debt relationshipsin times of nancial adversity and how they ease credit constraintsin the absence of a well-developed nancial market A recent econo-metric study by Khanna and Yafeh (1999) documents the importanceof intragroup loans as a mechanism by which group rms assist eachother in times of nancial distress in India Castenedarsquos (1998) studyof Mexican business groups also notes the existence of loan guaranteesamong member rms of the business group Mutual debt guaranteesplay an important role in the theory we develop here in addition tothe cross-holding of equity

Our theory suggests that business groups that are nanciallyinterlinked through cross-shareholding and cross-guarantee of loanscan be viewed as a way to obviate credit rationing caused by asym-metric information If all rms possess better information about thetypes of some other rms than the bank then nancially interlinkedbusiness groups can be a mechanism to induce rms to sort on the

2 For example according to the Bank of (South) Korea in 1991 cross-paymentguarantees by the top ve chaebol (13 companies) amounted to 199 trillion won andby the top 30 chaebol (76 companies) 383 trillion won (1 US dollar 1113 Koreanwon June 2000) Samsungrsquos three core companies ranked rst with combined paymentguarantees of 58 trillion won followed by Daewoorsquos core companies with 54 trillionwon See Business Korea 9 (9) 22 1992

Financially Interlinked Business Groups 593

basis of this information Consequently banks can offer a menu ofcontracts that vary in the extent of nancial interlinkage to inducerms to self-select on the basis of the equilibrium composition ofthe business groups they can form By accessing information rmsmay possess about each other nancial interlinkage among businessgroups can improve efciency in the credit market3

The starting point for our theory is the well-known lemon prob-lem that arises in credit markets with asymmetric informationbetween borrowers and lenders (see Stiglitz and Weiss 1981 De Mezaand Webb 1987) In the presence of adverse selection in the creditmarket the equilibrium allocation with standard debt contracts maybe inefcient mdash deserving projects may not get funded [the underin-vestment problem of Stiglitz and Weiss (1981)] or undeserving projectsmay get funded [the overinvestment problem of De Meza and Webb(1987)] But what if rms which are better equipped than outsidelenders to judge project riskiness are allowed to have cross-holdingof equity and debt guarantees thus forming business groups char-acterized by nancial interlinkage In this event we show that byoffering contracts that involve nancial interlinkage along with stan-dard debt contracts lenders can induce borrowers to form groups thatdisplay assortative matching and self-select among these contracts Inparticular in the presence of mutual debt guarantees high-risk bor-rowers will not be able to induce low-risk borrowers to associate withthem even if side payments are allowed The intuition is simple whileany type of rm will prefer to have a low-risk rm in its group thevalue from having a low-risk member is strictly higher for low-riskrms since they are themselves less likely to default and hence morelikely to have to pay the debt guarantee amount for a defaulting mem-ber Business groups are therefore formed of rms with similar char-acteristics This idea is expressed more formally as Proposition 1 ofthe model

We then examine the properties of these business groupsCorollary 1 expresses the nding that the isocost curves of businessgroups formed as a result of the positive assortative matching displaythe single-crossing property In particular for a given reduction in theinterest rate a low-risk rm will be willing to offer a higher debt guar-antee on the loans of its member rms since it has low-risk rms inits group Consequently an outside lender can use this property to

3 It is important to emphasize that our focus is on reciprocal shareholding anddebt guarantees that are the pattern among horizontal groups This is distinct froma unidirectional chain of shareholding as in a pyramidal structure There is an exten-sive literature on pyramidal structures See for example Wolfenzon (1999) Bebchuket al (1998) and La Porta et al (1999)

594 Journal of Economics amp Management Strategy

offer a menu of contracts to sort the business groups by quality ofinvestment projects Low-risk business groups will pay lower interestrates but engage in a higher degree of cross-holding than businessgroups with higher-risk projects

Starting from a situation where under standard debt contractsprojects with negative social surplus are borrowing by offering appro-priate nancially interlinked contracts lenders can exclude theseprojects from the credit market Conversely if rms with low-riskprojects do not nd it worthwhile to nance their projects when facedwith a standard pooled debt contract they can be attracted back intothe market using nancially interlinked contracts thereby enhancingefciency Proposition 2 expresses this idea

Mutual debt guarantees are the driving force behind the resultsmentioned above We extend our basic model in two different ways toexplain mutual equity cross-holdings one based on mutual monitor-ing among group members and the other based on risk sharing Withthese extensions we provide explanations for equity cross-holdingsand show at the same time that the assortative matching and screen-ing results in our basic model due to mutual debt guarantees continueto hold

Our paper has links to several different literatures It contributesto the understanding of business groups a literature that has beenrecently reinvigorated by the interest in the industrial organizationof developing and transition countries (see Ghemawat and Khanna1998 Khanna and Palepu 1998a Kali 1999b) Within this literature ourpaper is closest in spirit to Berglof and Perottirsquos (1994) analysis of theJapanese nancial keiretsu Their model focuses on the role of cross-shareholding among rms within a group to provide the incentive tomonitor as well as the means through reciprocal voting rights andcoalition-enforced threats of removal To guard against the possibilityof collusive behavior by the coalition as whole (namely no managerexerts effort and all vote their rmsrsquo cross-holdings to protect currentmanagerial appointment) the role of external debt from the groupbank is emphasized Poor protability results in nancial distress andthe control of the rm is shifted to the main lender moving away frommutual governance by cross-shareholders Unlike that paper the mainfocus of our paper is cross-holding of debt and how it can resolveadverse-selection problems However the extension of our baselinemodel in Section 23 is similar to their explanation of cross-holding ofequity although the formal models are different Their paper analyzesthe problem of ex post moral hazard and how it can be alleviated in arepeated-game setup through cross-holdings of equity which throughreciprocal exchange of voting rights allows rms to punish a manager

Financially Interlinked Business Groups 595

who shirks We focus on ex ante moral hazard in a static model andshow how cross-holdings of equity are a way to directly induce themanager of a rm to monitor the manager of another rm

A recent paper by Clayton and Jorgensen (2000) provides a dif-ferent explanation for cross-holding of equity than the ones offeredby Berglof and Perotti and our paper They show that in the presenceof Cournot quantity competition in the product market cross-equityholdings can lead to higher joint prots by inducing each rm to inter-nalize the effect of its quantity decision on the prots of the other rmOur result on assortative matching can also be compared with a recentpaper by Matsusaka (2000) that models the diversication decision ofrms as a search problem by which rms seek businesses that aregood matches for their capabilities The main difference between thesetwo models is that sorting (between rms) in our model is inducednot by any technological features but by a contractual feature

Our paper is also related to the literature on the problem ofadverse selection in credit markets and the role of collateral andmutual loan guarantees in alleviating this problem In the economichistory literature analyzing the banking insurance experience in theUS during the antebellum period and the 1920s Calomiris (1990) ndsevidence of mutual-liability-based bank insurance schemes beingmore successful than others In the banking literature Bester (1987)shows that collateralization of loans can ameliorate adverse-selectionproblems by screening borrowers by riskiness of project In partic-ular safe borrowers will be willing to offer greater collateral thanrisky borrowers for a given reduction in the interest rate becausethey expect to repay (and hence not lose the collateral) more often Ifborrowers are too poor to offer collateral then the problem of adverseselection can lead to inefciencies in the allocation of credit Inspiredby the successful experience of the Grameen Bank of Bangladesh inlending to poor villagers without any collateral by asking borrowersto form self-selected groups and making the group members jointlyliable for each otherrsquos loans several recent papers in the developmenteconomics literature have examined how this mechanism can solvevarious informational problems [see Morduch (1999) for a recent sur-vey] In particular joint liability can be used as an instrument to sortborrowers according to the riskiness of their projects (see for exampleGhatak 2000)

While our paper is similar in spirit to this literature there areseveral differences First joint liability in the context of microlend-ing takes the form of denying future credit to the entire group ifany member defaults which is quite different from cross-holding ormutual guarantee of debt Second there is no cross-holding of equity

596 Journal of Economics amp Management Strategy

among borrowers within a group and we demonstrate here that assor-tative matching and sorting is robust to the introduction of equityinterlinkage

Our paper demonstrates that nancial interlinkage throughequity and debt cross-holding can be efciency-improving in somecircumstances This is in contrast to the minority-expropriation viewof nancial interlinkage put forward in the literature on pyramidalstructures (such as La Porta et al 1999) However it is important tonote that the environment we are concerned with involves reciprocalshareholding and debt guarantees while pyramids are unidirectionalchains of shareholding4 A recent paper by Wolfenzon (1999) developsa theory of pyramidal ownership and its implications for extractionand rm value5

We have organized the remainder of the paper in the followingway Section 2 presents the basic model and considers various exten-sions of it Section 3 explores empirical implications of the model andconcludes

2 The Model

We develop a simple static model of adverse selection in the creditmarket similar to Stiglitz and Weiss (1981) and De Meza and Webb(1987) The economy consists of a continuum of risk-neutralentrepreneurs normalized to unity Each entrepreneur is the ownerof a blueprint for an investment project that requires a capital outlayfor the purchase of productive assets These assets can then be com-bined with entrepreneurial labor to produce a return on the invest-ment There is no moral hazard and entrepreneurs supply labor tothe project inelastically Once the capital is in place and the requiredunit of labor is put in projects yield either a high or a low return Werefer to these outcomes as success (S) and failure (F) respectively Thereare two exogenously given types of entrepreneurs characterized bythe probabilities of success of their projects pr and ps where

0 lt pr lt ps lt 1

Henceforth we will refer to them as risky and safe entrepreneurs Riskyand safe entrepreneurs exist in proportions h and 1 h in the popu-

4 Several recent studies (see Khanna 2000) cast doubt on economistsrsquo equation ofgroups and pyramids

5 We also focus solely on nancial interlinkage as a solution to asymmetric infor-mation problems in credit but there can be of course other economic functions thatbusiness groups perform In situations with imperfect markets for labor and capitalbusiness groups are able to act as surrogate labor markets and venture capitalists SeeKhanna and Palepu (1998a) for more detailed arguments Further theoretical work willtry to account for these functions explicitly

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

592 Journal of Economics amp Management Strategy

The existence of nancial interlinkages between the constituentrms of a business group has been widely noted Previous expla-nations of these interlinkages have focused on the role played bycross-shareholding in either providing risk sharing (see for exampleGoto 1982 Brioschi et al 1989 Nakatani 1984 Kali 1999b) soften-ing intensity of competition between rms in imperfect product mar-kets (see Clayton and Jorgensen 2000) or in mitigating moral-hazardproblems within the group (see Aoki 1982 Berglof and Perotti 1994)

What has been overlooked is the cross-holding of external debtthat often accompanies the cross-holding of equity among the rmsthat constitute the business group Sometimes referred to as cross-payment guarantees or mutual debt guarantees these imply that if amember rm is on the verge of defaulting on an external loan theother group rms will each pay a fraction of the defaulting rmrsquosexternal debt provided they are in a position to do so

Cross-guarantees of this kind are prevalent within businessgroups in several emerging economies In their study of the nanc-ing constraints of Korean chaebols Shin and Park (1999) emphasizethe role played by intragroup cross-guarantees in supporting exter-nal bank lending2 Such practices are also prevalent within Chinesebusiness groups Keisterrsquos (2000) extensive study of Chinese busi-ness groups describes the importance of mutual debt relationshipsin times of nancial adversity and how they ease credit constraintsin the absence of a well-developed nancial market A recent econo-metric study by Khanna and Yafeh (1999) documents the importanceof intragroup loans as a mechanism by which group rms assist eachother in times of nancial distress in India Castenedarsquos (1998) studyof Mexican business groups also notes the existence of loan guaranteesamong member rms of the business group Mutual debt guaranteesplay an important role in the theory we develop here in addition tothe cross-holding of equity

Our theory suggests that business groups that are nanciallyinterlinked through cross-shareholding and cross-guarantee of loanscan be viewed as a way to obviate credit rationing caused by asym-metric information If all rms possess better information about thetypes of some other rms than the bank then nancially interlinkedbusiness groups can be a mechanism to induce rms to sort on the

2 For example according to the Bank of (South) Korea in 1991 cross-paymentguarantees by the top ve chaebol (13 companies) amounted to 199 trillion won andby the top 30 chaebol (76 companies) 383 trillion won (1 US dollar 1113 Koreanwon June 2000) Samsungrsquos three core companies ranked rst with combined paymentguarantees of 58 trillion won followed by Daewoorsquos core companies with 54 trillionwon See Business Korea 9 (9) 22 1992

Financially Interlinked Business Groups 593

basis of this information Consequently banks can offer a menu ofcontracts that vary in the extent of nancial interlinkage to inducerms to self-select on the basis of the equilibrium composition ofthe business groups they can form By accessing information rmsmay possess about each other nancial interlinkage among businessgroups can improve efciency in the credit market3

The starting point for our theory is the well-known lemon prob-lem that arises in credit markets with asymmetric informationbetween borrowers and lenders (see Stiglitz and Weiss 1981 De Mezaand Webb 1987) In the presence of adverse selection in the creditmarket the equilibrium allocation with standard debt contracts maybe inefcient mdash deserving projects may not get funded [the underin-vestment problem of Stiglitz and Weiss (1981)] or undeserving projectsmay get funded [the overinvestment problem of De Meza and Webb(1987)] But what if rms which are better equipped than outsidelenders to judge project riskiness are allowed to have cross-holdingof equity and debt guarantees thus forming business groups char-acterized by nancial interlinkage In this event we show that byoffering contracts that involve nancial interlinkage along with stan-dard debt contracts lenders can induce borrowers to form groups thatdisplay assortative matching and self-select among these contracts Inparticular in the presence of mutual debt guarantees high-risk bor-rowers will not be able to induce low-risk borrowers to associate withthem even if side payments are allowed The intuition is simple whileany type of rm will prefer to have a low-risk rm in its group thevalue from having a low-risk member is strictly higher for low-riskrms since they are themselves less likely to default and hence morelikely to have to pay the debt guarantee amount for a defaulting mem-ber Business groups are therefore formed of rms with similar char-acteristics This idea is expressed more formally as Proposition 1 ofthe model

We then examine the properties of these business groupsCorollary 1 expresses the nding that the isocost curves of businessgroups formed as a result of the positive assortative matching displaythe single-crossing property In particular for a given reduction in theinterest rate a low-risk rm will be willing to offer a higher debt guar-antee on the loans of its member rms since it has low-risk rms inits group Consequently an outside lender can use this property to

3 It is important to emphasize that our focus is on reciprocal shareholding anddebt guarantees that are the pattern among horizontal groups This is distinct froma unidirectional chain of shareholding as in a pyramidal structure There is an exten-sive literature on pyramidal structures See for example Wolfenzon (1999) Bebchuket al (1998) and La Porta et al (1999)

594 Journal of Economics amp Management Strategy

offer a menu of contracts to sort the business groups by quality ofinvestment projects Low-risk business groups will pay lower interestrates but engage in a higher degree of cross-holding than businessgroups with higher-risk projects

Starting from a situation where under standard debt contractsprojects with negative social surplus are borrowing by offering appro-priate nancially interlinked contracts lenders can exclude theseprojects from the credit market Conversely if rms with low-riskprojects do not nd it worthwhile to nance their projects when facedwith a standard pooled debt contract they can be attracted back intothe market using nancially interlinked contracts thereby enhancingefciency Proposition 2 expresses this idea

Mutual debt guarantees are the driving force behind the resultsmentioned above We extend our basic model in two different ways toexplain mutual equity cross-holdings one based on mutual monitor-ing among group members and the other based on risk sharing Withthese extensions we provide explanations for equity cross-holdingsand show at the same time that the assortative matching and screen-ing results in our basic model due to mutual debt guarantees continueto hold

Our paper has links to several different literatures It contributesto the understanding of business groups a literature that has beenrecently reinvigorated by the interest in the industrial organizationof developing and transition countries (see Ghemawat and Khanna1998 Khanna and Palepu 1998a Kali 1999b) Within this literature ourpaper is closest in spirit to Berglof and Perottirsquos (1994) analysis of theJapanese nancial keiretsu Their model focuses on the role of cross-shareholding among rms within a group to provide the incentive tomonitor as well as the means through reciprocal voting rights andcoalition-enforced threats of removal To guard against the possibilityof collusive behavior by the coalition as whole (namely no managerexerts effort and all vote their rmsrsquo cross-holdings to protect currentmanagerial appointment) the role of external debt from the groupbank is emphasized Poor protability results in nancial distress andthe control of the rm is shifted to the main lender moving away frommutual governance by cross-shareholders Unlike that paper the mainfocus of our paper is cross-holding of debt and how it can resolveadverse-selection problems However the extension of our baselinemodel in Section 23 is similar to their explanation of cross-holding ofequity although the formal models are different Their paper analyzesthe problem of ex post moral hazard and how it can be alleviated in arepeated-game setup through cross-holdings of equity which throughreciprocal exchange of voting rights allows rms to punish a manager

Financially Interlinked Business Groups 595

who shirks We focus on ex ante moral hazard in a static model andshow how cross-holdings of equity are a way to directly induce themanager of a rm to monitor the manager of another rm

A recent paper by Clayton and Jorgensen (2000) provides a dif-ferent explanation for cross-holding of equity than the ones offeredby Berglof and Perotti and our paper They show that in the presenceof Cournot quantity competition in the product market cross-equityholdings can lead to higher joint prots by inducing each rm to inter-nalize the effect of its quantity decision on the prots of the other rmOur result on assortative matching can also be compared with a recentpaper by Matsusaka (2000) that models the diversication decision ofrms as a search problem by which rms seek businesses that aregood matches for their capabilities The main difference between thesetwo models is that sorting (between rms) in our model is inducednot by any technological features but by a contractual feature

Our paper is also related to the literature on the problem ofadverse selection in credit markets and the role of collateral andmutual loan guarantees in alleviating this problem In the economichistory literature analyzing the banking insurance experience in theUS during the antebellum period and the 1920s Calomiris (1990) ndsevidence of mutual-liability-based bank insurance schemes beingmore successful than others In the banking literature Bester (1987)shows that collateralization of loans can ameliorate adverse-selectionproblems by screening borrowers by riskiness of project In partic-ular safe borrowers will be willing to offer greater collateral thanrisky borrowers for a given reduction in the interest rate becausethey expect to repay (and hence not lose the collateral) more often Ifborrowers are too poor to offer collateral then the problem of adverseselection can lead to inefciencies in the allocation of credit Inspiredby the successful experience of the Grameen Bank of Bangladesh inlending to poor villagers without any collateral by asking borrowersto form self-selected groups and making the group members jointlyliable for each otherrsquos loans several recent papers in the developmenteconomics literature have examined how this mechanism can solvevarious informational problems [see Morduch (1999) for a recent sur-vey] In particular joint liability can be used as an instrument to sortborrowers according to the riskiness of their projects (see for exampleGhatak 2000)

While our paper is similar in spirit to this literature there areseveral differences First joint liability in the context of microlend-ing takes the form of denying future credit to the entire group ifany member defaults which is quite different from cross-holding ormutual guarantee of debt Second there is no cross-holding of equity

596 Journal of Economics amp Management Strategy

among borrowers within a group and we demonstrate here that assor-tative matching and sorting is robust to the introduction of equityinterlinkage

Our paper demonstrates that nancial interlinkage throughequity and debt cross-holding can be efciency-improving in somecircumstances This is in contrast to the minority-expropriation viewof nancial interlinkage put forward in the literature on pyramidalstructures (such as La Porta et al 1999) However it is important tonote that the environment we are concerned with involves reciprocalshareholding and debt guarantees while pyramids are unidirectionalchains of shareholding4 A recent paper by Wolfenzon (1999) developsa theory of pyramidal ownership and its implications for extractionand rm value5

We have organized the remainder of the paper in the followingway Section 2 presents the basic model and considers various exten-sions of it Section 3 explores empirical implications of the model andconcludes

2 The Model

We develop a simple static model of adverse selection in the creditmarket similar to Stiglitz and Weiss (1981) and De Meza and Webb(1987) The economy consists of a continuum of risk-neutralentrepreneurs normalized to unity Each entrepreneur is the ownerof a blueprint for an investment project that requires a capital outlayfor the purchase of productive assets These assets can then be com-bined with entrepreneurial labor to produce a return on the invest-ment There is no moral hazard and entrepreneurs supply labor tothe project inelastically Once the capital is in place and the requiredunit of labor is put in projects yield either a high or a low return Werefer to these outcomes as success (S) and failure (F) respectively Thereare two exogenously given types of entrepreneurs characterized bythe probabilities of success of their projects pr and ps where

0 lt pr lt ps lt 1

Henceforth we will refer to them as risky and safe entrepreneurs Riskyand safe entrepreneurs exist in proportions h and 1 h in the popu-

4 Several recent studies (see Khanna 2000) cast doubt on economistsrsquo equation ofgroups and pyramids

5 We also focus solely on nancial interlinkage as a solution to asymmetric infor-mation problems in credit but there can be of course other economic functions thatbusiness groups perform In situations with imperfect markets for labor and capitalbusiness groups are able to act as surrogate labor markets and venture capitalists SeeKhanna and Palepu (1998a) for more detailed arguments Further theoretical work willtry to account for these functions explicitly

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 593

basis of this information Consequently banks can offer a menu ofcontracts that vary in the extent of nancial interlinkage to inducerms to self-select on the basis of the equilibrium composition ofthe business groups they can form By accessing information rmsmay possess about each other nancial interlinkage among businessgroups can improve efciency in the credit market3

The starting point for our theory is the well-known lemon prob-lem that arises in credit markets with asymmetric informationbetween borrowers and lenders (see Stiglitz and Weiss 1981 De Mezaand Webb 1987) In the presence of adverse selection in the creditmarket the equilibrium allocation with standard debt contracts maybe inefcient mdash deserving projects may not get funded [the underin-vestment problem of Stiglitz and Weiss (1981)] or undeserving projectsmay get funded [the overinvestment problem of De Meza and Webb(1987)] But what if rms which are better equipped than outsidelenders to judge project riskiness are allowed to have cross-holdingof equity and debt guarantees thus forming business groups char-acterized by nancial interlinkage In this event we show that byoffering contracts that involve nancial interlinkage along with stan-dard debt contracts lenders can induce borrowers to form groups thatdisplay assortative matching and self-select among these contracts Inparticular in the presence of mutual debt guarantees high-risk bor-rowers will not be able to induce low-risk borrowers to associate withthem even if side payments are allowed The intuition is simple whileany type of rm will prefer to have a low-risk rm in its group thevalue from having a low-risk member is strictly higher for low-riskrms since they are themselves less likely to default and hence morelikely to have to pay the debt guarantee amount for a defaulting mem-ber Business groups are therefore formed of rms with similar char-acteristics This idea is expressed more formally as Proposition 1 ofthe model

We then examine the properties of these business groupsCorollary 1 expresses the nding that the isocost curves of businessgroups formed as a result of the positive assortative matching displaythe single-crossing property In particular for a given reduction in theinterest rate a low-risk rm will be willing to offer a higher debt guar-antee on the loans of its member rms since it has low-risk rms inits group Consequently an outside lender can use this property to

3 It is important to emphasize that our focus is on reciprocal shareholding anddebt guarantees that are the pattern among horizontal groups This is distinct froma unidirectional chain of shareholding as in a pyramidal structure There is an exten-sive literature on pyramidal structures See for example Wolfenzon (1999) Bebchuket al (1998) and La Porta et al (1999)

594 Journal of Economics amp Management Strategy

offer a menu of contracts to sort the business groups by quality ofinvestment projects Low-risk business groups will pay lower interestrates but engage in a higher degree of cross-holding than businessgroups with higher-risk projects

Starting from a situation where under standard debt contractsprojects with negative social surplus are borrowing by offering appro-priate nancially interlinked contracts lenders can exclude theseprojects from the credit market Conversely if rms with low-riskprojects do not nd it worthwhile to nance their projects when facedwith a standard pooled debt contract they can be attracted back intothe market using nancially interlinked contracts thereby enhancingefciency Proposition 2 expresses this idea

Mutual debt guarantees are the driving force behind the resultsmentioned above We extend our basic model in two different ways toexplain mutual equity cross-holdings one based on mutual monitor-ing among group members and the other based on risk sharing Withthese extensions we provide explanations for equity cross-holdingsand show at the same time that the assortative matching and screen-ing results in our basic model due to mutual debt guarantees continueto hold

Our paper has links to several different literatures It contributesto the understanding of business groups a literature that has beenrecently reinvigorated by the interest in the industrial organizationof developing and transition countries (see Ghemawat and Khanna1998 Khanna and Palepu 1998a Kali 1999b) Within this literature ourpaper is closest in spirit to Berglof and Perottirsquos (1994) analysis of theJapanese nancial keiretsu Their model focuses on the role of cross-shareholding among rms within a group to provide the incentive tomonitor as well as the means through reciprocal voting rights andcoalition-enforced threats of removal To guard against the possibilityof collusive behavior by the coalition as whole (namely no managerexerts effort and all vote their rmsrsquo cross-holdings to protect currentmanagerial appointment) the role of external debt from the groupbank is emphasized Poor protability results in nancial distress andthe control of the rm is shifted to the main lender moving away frommutual governance by cross-shareholders Unlike that paper the mainfocus of our paper is cross-holding of debt and how it can resolveadverse-selection problems However the extension of our baselinemodel in Section 23 is similar to their explanation of cross-holding ofequity although the formal models are different Their paper analyzesthe problem of ex post moral hazard and how it can be alleviated in arepeated-game setup through cross-holdings of equity which throughreciprocal exchange of voting rights allows rms to punish a manager

Financially Interlinked Business Groups 595

who shirks We focus on ex ante moral hazard in a static model andshow how cross-holdings of equity are a way to directly induce themanager of a rm to monitor the manager of another rm

A recent paper by Clayton and Jorgensen (2000) provides a dif-ferent explanation for cross-holding of equity than the ones offeredby Berglof and Perotti and our paper They show that in the presenceof Cournot quantity competition in the product market cross-equityholdings can lead to higher joint prots by inducing each rm to inter-nalize the effect of its quantity decision on the prots of the other rmOur result on assortative matching can also be compared with a recentpaper by Matsusaka (2000) that models the diversication decision ofrms as a search problem by which rms seek businesses that aregood matches for their capabilities The main difference between thesetwo models is that sorting (between rms) in our model is inducednot by any technological features but by a contractual feature

Our paper is also related to the literature on the problem ofadverse selection in credit markets and the role of collateral andmutual loan guarantees in alleviating this problem In the economichistory literature analyzing the banking insurance experience in theUS during the antebellum period and the 1920s Calomiris (1990) ndsevidence of mutual-liability-based bank insurance schemes beingmore successful than others In the banking literature Bester (1987)shows that collateralization of loans can ameliorate adverse-selectionproblems by screening borrowers by riskiness of project In partic-ular safe borrowers will be willing to offer greater collateral thanrisky borrowers for a given reduction in the interest rate becausethey expect to repay (and hence not lose the collateral) more often Ifborrowers are too poor to offer collateral then the problem of adverseselection can lead to inefciencies in the allocation of credit Inspiredby the successful experience of the Grameen Bank of Bangladesh inlending to poor villagers without any collateral by asking borrowersto form self-selected groups and making the group members jointlyliable for each otherrsquos loans several recent papers in the developmenteconomics literature have examined how this mechanism can solvevarious informational problems [see Morduch (1999) for a recent sur-vey] In particular joint liability can be used as an instrument to sortborrowers according to the riskiness of their projects (see for exampleGhatak 2000)

While our paper is similar in spirit to this literature there areseveral differences First joint liability in the context of microlend-ing takes the form of denying future credit to the entire group ifany member defaults which is quite different from cross-holding ormutual guarantee of debt Second there is no cross-holding of equity

596 Journal of Economics amp Management Strategy

among borrowers within a group and we demonstrate here that assor-tative matching and sorting is robust to the introduction of equityinterlinkage

Our paper demonstrates that nancial interlinkage throughequity and debt cross-holding can be efciency-improving in somecircumstances This is in contrast to the minority-expropriation viewof nancial interlinkage put forward in the literature on pyramidalstructures (such as La Porta et al 1999) However it is important tonote that the environment we are concerned with involves reciprocalshareholding and debt guarantees while pyramids are unidirectionalchains of shareholding4 A recent paper by Wolfenzon (1999) developsa theory of pyramidal ownership and its implications for extractionand rm value5

We have organized the remainder of the paper in the followingway Section 2 presents the basic model and considers various exten-sions of it Section 3 explores empirical implications of the model andconcludes

2 The Model

We develop a simple static model of adverse selection in the creditmarket similar to Stiglitz and Weiss (1981) and De Meza and Webb(1987) The economy consists of a continuum of risk-neutralentrepreneurs normalized to unity Each entrepreneur is the ownerof a blueprint for an investment project that requires a capital outlayfor the purchase of productive assets These assets can then be com-bined with entrepreneurial labor to produce a return on the invest-ment There is no moral hazard and entrepreneurs supply labor tothe project inelastically Once the capital is in place and the requiredunit of labor is put in projects yield either a high or a low return Werefer to these outcomes as success (S) and failure (F) respectively Thereare two exogenously given types of entrepreneurs characterized bythe probabilities of success of their projects pr and ps where

0 lt pr lt ps lt 1

Henceforth we will refer to them as risky and safe entrepreneurs Riskyand safe entrepreneurs exist in proportions h and 1 h in the popu-

4 Several recent studies (see Khanna 2000) cast doubt on economistsrsquo equation ofgroups and pyramids

5 We also focus solely on nancial interlinkage as a solution to asymmetric infor-mation problems in credit but there can be of course other economic functions thatbusiness groups perform In situations with imperfect markets for labor and capitalbusiness groups are able to act as surrogate labor markets and venture capitalists SeeKhanna and Palepu (1998a) for more detailed arguments Further theoretical work willtry to account for these functions explicitly

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

594 Journal of Economics amp Management Strategy

offer a menu of contracts to sort the business groups by quality ofinvestment projects Low-risk business groups will pay lower interestrates but engage in a higher degree of cross-holding than businessgroups with higher-risk projects

Starting from a situation where under standard debt contractsprojects with negative social surplus are borrowing by offering appro-priate nancially interlinked contracts lenders can exclude theseprojects from the credit market Conversely if rms with low-riskprojects do not nd it worthwhile to nance their projects when facedwith a standard pooled debt contract they can be attracted back intothe market using nancially interlinked contracts thereby enhancingefciency Proposition 2 expresses this idea

Mutual debt guarantees are the driving force behind the resultsmentioned above We extend our basic model in two different ways toexplain mutual equity cross-holdings one based on mutual monitor-ing among group members and the other based on risk sharing Withthese extensions we provide explanations for equity cross-holdingsand show at the same time that the assortative matching and screen-ing results in our basic model due to mutual debt guarantees continueto hold

Our paper has links to several different literatures It contributesto the understanding of business groups a literature that has beenrecently reinvigorated by the interest in the industrial organizationof developing and transition countries (see Ghemawat and Khanna1998 Khanna and Palepu 1998a Kali 1999b) Within this literature ourpaper is closest in spirit to Berglof and Perottirsquos (1994) analysis of theJapanese nancial keiretsu Their model focuses on the role of cross-shareholding among rms within a group to provide the incentive tomonitor as well as the means through reciprocal voting rights andcoalition-enforced threats of removal To guard against the possibilityof collusive behavior by the coalition as whole (namely no managerexerts effort and all vote their rmsrsquo cross-holdings to protect currentmanagerial appointment) the role of external debt from the groupbank is emphasized Poor protability results in nancial distress andthe control of the rm is shifted to the main lender moving away frommutual governance by cross-shareholders Unlike that paper the mainfocus of our paper is cross-holding of debt and how it can resolveadverse-selection problems However the extension of our baselinemodel in Section 23 is similar to their explanation of cross-holding ofequity although the formal models are different Their paper analyzesthe problem of ex post moral hazard and how it can be alleviated in arepeated-game setup through cross-holdings of equity which throughreciprocal exchange of voting rights allows rms to punish a manager

Financially Interlinked Business Groups 595

who shirks We focus on ex ante moral hazard in a static model andshow how cross-holdings of equity are a way to directly induce themanager of a rm to monitor the manager of another rm

A recent paper by Clayton and Jorgensen (2000) provides a dif-ferent explanation for cross-holding of equity than the ones offeredby Berglof and Perotti and our paper They show that in the presenceof Cournot quantity competition in the product market cross-equityholdings can lead to higher joint prots by inducing each rm to inter-nalize the effect of its quantity decision on the prots of the other rmOur result on assortative matching can also be compared with a recentpaper by Matsusaka (2000) that models the diversication decision ofrms as a search problem by which rms seek businesses that aregood matches for their capabilities The main difference between thesetwo models is that sorting (between rms) in our model is inducednot by any technological features but by a contractual feature

Our paper is also related to the literature on the problem ofadverse selection in credit markets and the role of collateral andmutual loan guarantees in alleviating this problem In the economichistory literature analyzing the banking insurance experience in theUS during the antebellum period and the 1920s Calomiris (1990) ndsevidence of mutual-liability-based bank insurance schemes beingmore successful than others In the banking literature Bester (1987)shows that collateralization of loans can ameliorate adverse-selectionproblems by screening borrowers by riskiness of project In partic-ular safe borrowers will be willing to offer greater collateral thanrisky borrowers for a given reduction in the interest rate becausethey expect to repay (and hence not lose the collateral) more often Ifborrowers are too poor to offer collateral then the problem of adverseselection can lead to inefciencies in the allocation of credit Inspiredby the successful experience of the Grameen Bank of Bangladesh inlending to poor villagers without any collateral by asking borrowersto form self-selected groups and making the group members jointlyliable for each otherrsquos loans several recent papers in the developmenteconomics literature have examined how this mechanism can solvevarious informational problems [see Morduch (1999) for a recent sur-vey] In particular joint liability can be used as an instrument to sortborrowers according to the riskiness of their projects (see for exampleGhatak 2000)

While our paper is similar in spirit to this literature there areseveral differences First joint liability in the context of microlend-ing takes the form of denying future credit to the entire group ifany member defaults which is quite different from cross-holding ormutual guarantee of debt Second there is no cross-holding of equity

596 Journal of Economics amp Management Strategy

among borrowers within a group and we demonstrate here that assor-tative matching and sorting is robust to the introduction of equityinterlinkage

Our paper demonstrates that nancial interlinkage throughequity and debt cross-holding can be efciency-improving in somecircumstances This is in contrast to the minority-expropriation viewof nancial interlinkage put forward in the literature on pyramidalstructures (such as La Porta et al 1999) However it is important tonote that the environment we are concerned with involves reciprocalshareholding and debt guarantees while pyramids are unidirectionalchains of shareholding4 A recent paper by Wolfenzon (1999) developsa theory of pyramidal ownership and its implications for extractionand rm value5

We have organized the remainder of the paper in the followingway Section 2 presents the basic model and considers various exten-sions of it Section 3 explores empirical implications of the model andconcludes

2 The Model

We develop a simple static model of adverse selection in the creditmarket similar to Stiglitz and Weiss (1981) and De Meza and Webb(1987) The economy consists of a continuum of risk-neutralentrepreneurs normalized to unity Each entrepreneur is the ownerof a blueprint for an investment project that requires a capital outlayfor the purchase of productive assets These assets can then be com-bined with entrepreneurial labor to produce a return on the invest-ment There is no moral hazard and entrepreneurs supply labor tothe project inelastically Once the capital is in place and the requiredunit of labor is put in projects yield either a high or a low return Werefer to these outcomes as success (S) and failure (F) respectively Thereare two exogenously given types of entrepreneurs characterized bythe probabilities of success of their projects pr and ps where

0 lt pr lt ps lt 1

Henceforth we will refer to them as risky and safe entrepreneurs Riskyand safe entrepreneurs exist in proportions h and 1 h in the popu-

4 Several recent studies (see Khanna 2000) cast doubt on economistsrsquo equation ofgroups and pyramids

5 We also focus solely on nancial interlinkage as a solution to asymmetric infor-mation problems in credit but there can be of course other economic functions thatbusiness groups perform In situations with imperfect markets for labor and capitalbusiness groups are able to act as surrogate labor markets and venture capitalists SeeKhanna and Palepu (1998a) for more detailed arguments Further theoretical work willtry to account for these functions explicitly

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 595

who shirks We focus on ex ante moral hazard in a static model andshow how cross-holdings of equity are a way to directly induce themanager of a rm to monitor the manager of another rm

A recent paper by Clayton and Jorgensen (2000) provides a dif-ferent explanation for cross-holding of equity than the ones offeredby Berglof and Perotti and our paper They show that in the presenceof Cournot quantity competition in the product market cross-equityholdings can lead to higher joint prots by inducing each rm to inter-nalize the effect of its quantity decision on the prots of the other rmOur result on assortative matching can also be compared with a recentpaper by Matsusaka (2000) that models the diversication decision ofrms as a search problem by which rms seek businesses that aregood matches for their capabilities The main difference between thesetwo models is that sorting (between rms) in our model is inducednot by any technological features but by a contractual feature

Our paper is also related to the literature on the problem ofadverse selection in credit markets and the role of collateral andmutual loan guarantees in alleviating this problem In the economichistory literature analyzing the banking insurance experience in theUS during the antebellum period and the 1920s Calomiris (1990) ndsevidence of mutual-liability-based bank insurance schemes beingmore successful than others In the banking literature Bester (1987)shows that collateralization of loans can ameliorate adverse-selectionproblems by screening borrowers by riskiness of project In partic-ular safe borrowers will be willing to offer greater collateral thanrisky borrowers for a given reduction in the interest rate becausethey expect to repay (and hence not lose the collateral) more often Ifborrowers are too poor to offer collateral then the problem of adverseselection can lead to inefciencies in the allocation of credit Inspiredby the successful experience of the Grameen Bank of Bangladesh inlending to poor villagers without any collateral by asking borrowersto form self-selected groups and making the group members jointlyliable for each otherrsquos loans several recent papers in the developmenteconomics literature have examined how this mechanism can solvevarious informational problems [see Morduch (1999) for a recent sur-vey] In particular joint liability can be used as an instrument to sortborrowers according to the riskiness of their projects (see for exampleGhatak 2000)

While our paper is similar in spirit to this literature there areseveral differences First joint liability in the context of microlend-ing takes the form of denying future credit to the entire group ifany member defaults which is quite different from cross-holding ormutual guarantee of debt Second there is no cross-holding of equity

596 Journal of Economics amp Management Strategy

among borrowers within a group and we demonstrate here that assor-tative matching and sorting is robust to the introduction of equityinterlinkage

Our paper demonstrates that nancial interlinkage throughequity and debt cross-holding can be efciency-improving in somecircumstances This is in contrast to the minority-expropriation viewof nancial interlinkage put forward in the literature on pyramidalstructures (such as La Porta et al 1999) However it is important tonote that the environment we are concerned with involves reciprocalshareholding and debt guarantees while pyramids are unidirectionalchains of shareholding4 A recent paper by Wolfenzon (1999) developsa theory of pyramidal ownership and its implications for extractionand rm value5

We have organized the remainder of the paper in the followingway Section 2 presents the basic model and considers various exten-sions of it Section 3 explores empirical implications of the model andconcludes

2 The Model

We develop a simple static model of adverse selection in the creditmarket similar to Stiglitz and Weiss (1981) and De Meza and Webb(1987) The economy consists of a continuum of risk-neutralentrepreneurs normalized to unity Each entrepreneur is the ownerof a blueprint for an investment project that requires a capital outlayfor the purchase of productive assets These assets can then be com-bined with entrepreneurial labor to produce a return on the invest-ment There is no moral hazard and entrepreneurs supply labor tothe project inelastically Once the capital is in place and the requiredunit of labor is put in projects yield either a high or a low return Werefer to these outcomes as success (S) and failure (F) respectively Thereare two exogenously given types of entrepreneurs characterized bythe probabilities of success of their projects pr and ps where

0 lt pr lt ps lt 1

Henceforth we will refer to them as risky and safe entrepreneurs Riskyand safe entrepreneurs exist in proportions h and 1 h in the popu-

4 Several recent studies (see Khanna 2000) cast doubt on economistsrsquo equation ofgroups and pyramids

5 We also focus solely on nancial interlinkage as a solution to asymmetric infor-mation problems in credit but there can be of course other economic functions thatbusiness groups perform In situations with imperfect markets for labor and capitalbusiness groups are able to act as surrogate labor markets and venture capitalists SeeKhanna and Palepu (1998a) for more detailed arguments Further theoretical work willtry to account for these functions explicitly

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

596 Journal of Economics amp Management Strategy

among borrowers within a group and we demonstrate here that assor-tative matching and sorting is robust to the introduction of equityinterlinkage

Our paper demonstrates that nancial interlinkage throughequity and debt cross-holding can be efciency-improving in somecircumstances This is in contrast to the minority-expropriation viewof nancial interlinkage put forward in the literature on pyramidalstructures (such as La Porta et al 1999) However it is important tonote that the environment we are concerned with involves reciprocalshareholding and debt guarantees while pyramids are unidirectionalchains of shareholding4 A recent paper by Wolfenzon (1999) developsa theory of pyramidal ownership and its implications for extractionand rm value5

We have organized the remainder of the paper in the followingway Section 2 presents the basic model and considers various exten-sions of it Section 3 explores empirical implications of the model andconcludes

2 The Model

We develop a simple static model of adverse selection in the creditmarket similar to Stiglitz and Weiss (1981) and De Meza and Webb(1987) The economy consists of a continuum of risk-neutralentrepreneurs normalized to unity Each entrepreneur is the ownerof a blueprint for an investment project that requires a capital outlayfor the purchase of productive assets These assets can then be com-bined with entrepreneurial labor to produce a return on the invest-ment There is no moral hazard and entrepreneurs supply labor tothe project inelastically Once the capital is in place and the requiredunit of labor is put in projects yield either a high or a low return Werefer to these outcomes as success (S) and failure (F) respectively Thereare two exogenously given types of entrepreneurs characterized bythe probabilities of success of their projects pr and ps where

0 lt pr lt ps lt 1

Henceforth we will refer to them as risky and safe entrepreneurs Riskyand safe entrepreneurs exist in proportions h and 1 h in the popu-

4 Several recent studies (see Khanna 2000) cast doubt on economistsrsquo equation ofgroups and pyramids

5 We also focus solely on nancial interlinkage as a solution to asymmetric infor-mation problems in credit but there can be of course other economic functions thatbusiness groups perform In situations with imperfect markets for labor and capitalbusiness groups are able to act as surrogate labor markets and venture capitalists SeeKhanna and Palepu (1998a) for more detailed arguments Further theoretical work willtry to account for these functions explicitly

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 597

lation The outcomes of the projects are assumed to be independentlydistributed for the same types as well as across different types Thereturn of a project is R gt 0 if it is successful and 0 if it fails for bothtypes of projects Our formulation of how risky and safe projects differis similar to De Meza and Webb (1987) In Stiglitz and Weiss (1981) incontrast risky and safe projects are assumed to have the same meanreturn but risky projects have a greater spread around the mean Ourmain results extend easily to this case as we discuss at the end ofSection 3

Entrepreneurs have no wealth and hence have to rely on externalnance An entrepreneur who chooses not to use his blueprint obtainsthe reservation payoff u We use the terms entrepreneur and rm inter-changeably in the paper There is a risk-neutral external lender whichwe will refer to as a bank that can provide investment nancing toeach entrepreneur at a gross (inclusive of principal) interest rate rBorrowers have limited liability So if their projects fail entrepreneursare liable up to the amount of collateralizable wealth they possesswhich we take to be zero for simplicity The opportunity cost of cap-ital for the bank is q per loan We assume it is economically efcientto pursue only safe investment projects

Assumption 1

psR u q gt 0 gt prR u q (1)

Following existing models of adverse selection in the credit mar-ket we will focus only on debt contracts6

The type of each borrower is unknown to the bank Howevereach entrepreneur possesses some information about a group of otherentrepreneurs Specically the type of each entrepreneur belongingto the same information network is common knowledge within thenetwork We could think of each entrepreneur as belonging to suchan informational network because of being a member of a social orethnic group7

6 In this model there is no difference between debt and equity contracts Whena borrower fails she pays nothing whereas when she succeeds she pays a positiveamount to the bank Since both types of borrowers earn the same revenue when theirprojects succeed whether the bank is paid a fraction of the success revenue or anamount independent of it makes no difference In contrast in Stiglitz and Weissrsquos modelrisky and safe borrowers earn different amounts of revenue when their projects succeedAs a result debt and equity contracts have different implications

7 These informational networks may or may not be connected with one another

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

598 Journal of Economics amp Management Strategy

21 The Adverse-Selection Problem

In the environment that we have specied simple debt contracts mayrun into the following type of adverse-selection problem The reason-ing is straightforward Because the bank cannot identify ex ante whichblueprints are better and which are worse it will offer a pooling con-tract to all entrepreneurs based on the average repayment rate in thepopulation p h pr (1 h )ps Thus the bank will charge an interestrate r q p However it is possible that projects that are not worthyof being nanced from the point of view of economic efciency maybe nanced at this interest rate We assume this condition to hold

Assumption 2

pr Rq

pgt u (2)

Notice that this condition may hold even when risky projects areunproductive (ie pr R u q lt 0) because they are cross-subsidizedby safe borrowers Here the average repayment rate would be p andthe level of expected social surplus would be h pr (R q p) u (1h ) ps(R q p) u pR q u h (pr R q u) (1 h )(psRq u) If the bank had perfect information about the type of a rmthen it would lend to safe rms only leading to an average repay-ment rate of ps and the level of expected social surplus would be(1 h )(psR q u) Given (1) both these surpluses are strictly higherthan those achieved under adverse selection This is the overinvest-ment problem in credit markets with adverse selection (De Meza andWebb 1987) Notice that while social surplus is higher when riskyrms are excluded welfare comparisons are less clear-cut Safe bor-rowers are better off under the full-information allocation (since theyhave to cross-subsidize risky borrowers under the pooling debt con-tract) while risky borrowers are strictly worse off

22 Financial Interlinkage andAssortative Matching

In this section we explain how the cross-holding of equity and debtbetween rms can be a solution to the adverse-selection problem inthe market for credit A set of rms that are interlinked through debtand equity in this fashion is what we refer to as a business group

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 599

221 Assortative Matching and Single Crossing Firstwe show that for any given interest rate r rms with debt and equitycross-holding will always choose business-group partners of the sametype That is the equilibrium in the group-formation game satisesthe optimal-sorting property namely rms not in the same businessgroup cannot form a business group without making at least one ofthem worse off Our proof of this property explicitly allow rms to beable to make side payments to each other In principle then a riskyrm can pay a safe rm to join its business group For simplicity wewill consider groups of size two8

Let frac14ij denote the expected payoff of an entrepreneur of type iwho forms a business group with an entrepreneur of type j We focuson symmetric equity stakes here and consider the implications ofrelaxing this assumption in Section 25 Let a be the share of his ownproject returns that he retains 1 a is his stake in his partnerrsquos rmLet 0 c 1 be the extent of the liability that a rm in the businessgroup has on a member rmrsquos loans In the event that a project inwhich an entrepreneur is a shareholder fails and goes bankrupt theentrepreneur will pay a fraction c of the failed rmrsquos debt obligation

Then

frac14ij pi pj a (R r) (1 a )(R r) pi(1 p j ) a (R r c r)

p j (1 pi )(1 a )(R r c r)

R (1 c )r a p i (1 a )p j pi pj c r

Thus if a risky rm were to switch from forming a business groupwith a risky partner to one with a safe partner instead the expectedgain would be

frac14rs frac14rr (ps pr )[(1 a ) R (1 c )r pr c r]

Similarly if a safe rm were to switch from forming a business groupwith a safe partner to one with a risky partner instead the expectedloss would be

frac14ss frac14sr (ps pr )[(1 a ) R (1 c )r ps c r]

The question now is whether the gain for the risky entrepreneurfrom forming a group with the safe entrepreneur is greater than theloss that the safe entrepreneur incurs from agreeing to allow riskyentrepreneurs to form a group with him If this is so then the risky

8 The results generalize to business groups of size more than two

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

600 Journal of Economics amp Management Strategy

entrepreneurs can negotiate a bribe for the safe entrepreneurs thatwill induce them to team with the risky entrepreneurs and will makeboth parties better off However since ps gt pr comparing the twoexpressions we see that

frac14ss frac14rr gt frac14rs frac14sr

and therefore business-group formation will display positive assorta-tive matching

Consequently as in Beckerrsquos (1993) analysis of the marriage mar-ket entrepreneurs in our model will form business groups with nan-cial interlinkage only with those who have a similar risk prole Theintuition is the following From the point of view of both types of bor-rowers a safe partner is preferred both because of higher expectedreturns through cross-shareholding and lower expected liability ontheir loans through the cross-guarantee of loans However as far ascross-shareholding is concerned the gain of a risky type from havinga safe partner is exactly equal to the loss of a safe type from havinga risky partner As a result with side payments being possible rmswill be indifferent between choices of partners However as far ascross-guarantee of loans is concerned a safe rm will have a highervaluation of having a safe partner in the business group than a riskyrm This is because the benet of having a safe rather than a riskypartner is realized only when a rm does not itself default and theprobability of this is high for safe rms This implies that a risky rmwill never nd it protable to attract a safe rm to be a group mem-ber after compensating the latter for the loss of having a risky rmas a partner We can state the preceding analysis in the form of thefollowing proposition

Proposition 1 Financial interlinkage within business groups in the formof cross-shareholding and cross-guarantee of loans leads to positive assorta-tive matching in the formation of business groups

Given assortative matching in equilibrium the payoff of eachtype of entrepreneur under a contract (r c ) will be

frac14i i( c r) R (1 c )r pi p2i c r i s r

Hence the isocost curve of an entrepreneur of type i in ( c r) space isrepresented by pi(1 c )r p2

i c r k i s r where k is a constantThe slope of an isocost curve of an entrepreneur of type i is then

drd c

r1

11 p i

clt 0 i s r

Since ps gt pr the isocost curve for the safe entrepreneur is atterthan one for the risky entrepreneur in the ( c r) plane That is the

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 601

Indifference curve of safe firms

Indifference curve of risky firms

Preference direct ion

Interest rate

Cross liability

g

r

FIGURE 1

isocost curves satisfy the single-crossing property which implies thatby offering a menu of contracts that vary in r and c banks can inducevarious business groups to self-select

Corollary 1 Isocost curves of rms satisfy the single-crossing propertyin the ( c r) plane

Figure 1 represents these isocost curvesSince the isocost curves are negatively sloped the higher is the

interest rate r the lower will have to be the degree of cross-guaranteeof loans c It also follows that to receive a small reduction in theinterest rate safe rms will be willing to offer a higher degree ofcross-guarantee than risky rms because they have safe partners

Banks can use this property to screen rms by offering contractsto business groups that differ in the interest rate r and the degree ofcross-guarantee c We show that this can improve social surplus bydriving out risky rms that were not initially borrowing under thestandard debt contract

Two observations are worth making regarding the assortative-matching result and the single-crossing property before we move onto derive optimal screening contracts First our proof relies only onrms having different probabilities of success and on their types (ieprobabilities of success) being complementary in the payoff function

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

602 Journal of Economics amp Management Strategy

induced by a nancial interlinkage9 In particular it does not dependon whether safe and risky rms have the same or different expectedproject returns and hence this result extends to Stiglitz and Weissrsquosenvironment as well Second and related to the previous observationthe extent of equity-cross holding plays no role in these proofs either(ie they go through for every value of a ) Even if there were nocross-holding of equity these results would continue to hold

In Section 23 where we discuss extensions of the current modelwe outline two alternative theories of equity-cross holding one basedon mutual monitoring and the other on risk sharing In particularwe show that the assortative-matching result and the single-crossingproperty continue to hold in these extensions

222 Optimal Screening Contracts Next we derive opti-mal screening contracts To do so rst we need to prove the followinglemma

Lemma 1 If the contracts ( c r rr ) and ( c s rs) are incentive-compatiblethen assortative matching will still result in the formation of business groups

Proof Suppose not Then it must be that two heterogeneous groupsearn greater joint prots by borrowing under either of the two con-tracts that are offered [say ( c s rs)] than two homogeneous groups canearn under the contracts ( c r rr) and ( c s rs) that is

frac14rs( c s rs) frac14sr ( c s rs) gt frac14rr( c r rr ) frac14ss( c s rs)

By Proposition 1 if the contract ( c s rs) had been the only one offeredby the bank assortative matching would have resulted That is

frac14rr( c s rs) frac14ss( c s rs) gt frac14rs( c s rs) frac14sr ( c s rs)

Together these inequalities imply frac14rr( c s rs) gt frac14rr ( c r rr) But that vio-lates the incentive compatibility constraint for risky rms acontradiction u

9 That is 2frac14i j (r c ) pipj c r gt 0 Technically this is the reason why positiveassortative matching results with such a contractual form Beckerrsquos analysis showedthat if the cross partial derivatives of the types of agents are negative then negativeassortative matching results In the context of business groups if we allow for a moregeneral production technology such that member rms have comparative advantage indifferent tasks (or products) and these enter a joint production function of the group asa whole then we could have positive or negative assortative matching depending onwhether these tasks are strategic complements or substitutes

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 603

Our goal is to nd a pair of contracts ( c s rs) and ( c r rr) suchthat risky borrowers do not borrow Without loss of generality wecan take ( c r rr ) (0 q pr ) the same contract they would be offeredunder full information The bankrsquos zero-prot condition from lendingto safe borrowers is10

rsps 1 c s(1 ps) q

We can use this to solve for rs in terms of c s

rs

q

ps 1 c s(1 ps) (3)

The incentive compatibility constraint of risky borrowers require themto prefer not borrowing at all to borrowing under the contract ( c s rs)

pr[R rs 1 c s(1 pr) ] u (4)

Substituting (3) in (4) we get the following condition for the existenceof a separating equilibrium there exists c s [0 1] such that

prR upr 1 c s(1 pr)ps 1 c s(1 ps)

q (5)

Finally we need to ensure that another condition is satised forthe optimal cross-guarantee contract to be feasible namely the con-tract ( c s rs) must satisfy the following limited liability constraint

rs(1 c s) R

Using (3) this condition can be written as

1 c s

ps 1 c s(1 ps)q R (6)

This guarantees that a rm cannot make any transfers to the bankwhen its project fails and that the sum of its own liability and theliability for member rms of the business group it belongs to throughcross-guarantees r(1 c ) cannot exceed the realized revenue fromthe project when it succeeds We assume that the following pair ofconditions hold

10 If the bank is a monopolist maximizing its expected prots then the optimalcontracts will be similar to those derived in this section but they will lie on the respec-tive participation constraints of the borrowers as opposed to the zero-prot constraintsof the bank

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

604 Journal of Economics amp Management Strategy

Assumption 3

pr

pq u prR lt

pr (2 pr)ps(2 ps)

q u

q u lt psR qp2

s

p2r

u

Observe that Assumptions 1 and 2 are implied by this assump-tion11 Recall that these assumptions implied that safe projects gen-erate positive social surplus risky projects generate negative socialsurplus and under a standard debt contract risky borrowers borrowThen we are able to prove

Lemma 2 Suppose Assumption 3 holds Then there exists a pair of con-tracts ( c s rs) and ( c r rr ) that satisfy the zero-prot condition of the bankand the limited-liability constraint such that risky borrowers do not borrow

Proof See the Appendix u

The solution to the optimal-separating problem will not in gen-eral be unique We prove that there exists a critical value of the degreeof optimal cross-holding of debt for safe borrowers c (0 1] suchthat for any c s c there exists a corresponding interest rate for safeborrowers rs (from the bankrsquos zero-prot condition) such that theincentive compatibility constraint of risky borrowers and the limited-liability constraint are satised At the same time so long as Assump-tion 3 is satised it is possible to offer a pair of contracts such thatonly safe rms get to borrow by forming business groups with cross-shareholding and cross-guarantee of loans and risky rms do notreceive loans12

Since the contracts ( c s rs) and (0 q pr ) lie on the respectivezero-prot equations the expected payoff of a safe rm is equal topsR u q and the repayment rate is equal to ps Hence the averagerepayment rate and social surplus under this pair of contracts are attheir full-information levels and strictly higher than those under ordi-nary debt contracts The main result of this section readily follows

11 This is not obvious for the part of Assumption 1 that says pr R lt q u Inthe proof of Lemma 2 below we show that pr (2 pr) ps(2 ps) lt 1 and so prR ltpr(2 pr ) ps(2 ps) q u implies pr R lt q u

12 A counterfactual implication of this model is that no rm borrows under a stan-dard debt contract In order to highlight the potential efciency gains from nanciallyinterlinked business groups using the simplest possible model we assumed there areonly two types of borrowers with risky borrowers having inefcient projects Thisassumption can be readily relaxed to allow a third type of borrower risky borrowerswho have efcient projects Then under the optimal screening contracts risky borrow-ers who have efcient projects will borrow under the standard debt contract

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 605

Proposition 2 If parameters satisfy Assumption 3 nancial interlink-age within business groups in the form of cross-shareholding and cross-guarantee of loans will produce greater expected social surplus and repay-ment rates than standard debt contracts

It is straightforward to extend this result to an environmentwhere adverse selection leads to the exclusion of safe projects withpositive net surplus from the market as in Stiglitz and Weiss (1981)rather than the form of inefciency we focus on (ie risky projectswith negative net surplus receiving loans) As observed before Propo-sition 1 Corollary 1 and Lemma 1 do not depend on the distributionof revenues of the projects and hence these results continue to applyThe key difference between these two setups lies in which type ofborrowerrsquos participation constraint binds In the above model safetypes have a higher expected payoff for any given contract and soit is the participation constraint of risky borrowers that we focus onHowever if both types of projects have the same mean returns thensafe borrowers have a lower expected payoff under a standard pool-ing debt contract This is because expected interest payments to thebank are higher for safe borrowers (since they pay back the sameamount more often than risky borrowers) while expected revenuesare by assumption the same Hence it is possible that under standarddebt contracts the participation constraint of safe borrowers will notbe satised and only risky borrowers will borrow at an interest rateof q pr If nancially interlinked contracts are allowed the bank canoffer two contracts (rs c s) and ( q pr 0) such that safe borrowers willchoose the former and risky borrowers will choose the latter Noticethat the welfare implication of nancial interlinkage is quite differentin this case Financially interlinked contracts would attract safe rmsback into the market while risky borrowers would continue to borrowunder debt contracts As a result the welfare of safe borrowers socialsurplus and repayment rates will all be higher but risky borrowerswould be no worse off In contrast in the basic model social surplusand the welfare of safe borrowers are higher but risky borrowers arestrictly worse off with nancial interlinkage

23 Optimal Cross-Holding of Equity

The main results in the previous section were driven by the cross-holding of debt How much a rm valued having a safe rm inits group was positively correlated with its own type due to cross-holding of debt leading to assortative matching in the formation ofbusiness groups This in turn allowed banks to exclude risky rmswith inefcient projects from borrowing which might not have been

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

606 Journal of Economics amp Management Strategy

possible with standard debt contracts However equity cross-holdingis a prominent characteristic of business groups across the world andits extent played no role in the results of the previous section Inthis section we address the issue of determining the optimal degreeof cross-holding of equity using two alternative models The rst isbased on mutual monitoring and the second on risk sharing

231 Mutual Monitoring Consider a simple extension ofour basic model where the probability of success of each type ofproject depends (apart from an intrinsic quality component) on theeffort put in by managers of rms within a business group In par-ticular the probability of success of the ith type of project is nowp i pi a b where ps gt pr a [0 a] is the effort level chosenby the manager of a rm and b [0 b] is the effort level chosen bythe manager of its partner rm A sufcient condition for p i lt 1 fori r s is a b lt 1 ps The key assumption is that a and b areunobservable among the managers of the rm and to the bank Alsothese efforts are subject to some disutility costs which are taken to bequadratic for simplicity 1

2 c1a2 and 1

2 c2b2 The effort of the manager of

a rm devoted to the project of a rm that is a member of the samebusiness group can be interpreted as monitoring effort or as help13

We show that (a) equity-cross holding can work as an optimal incen-tive device to elicit effort (or other noncontractible resources) fromother group members and (b) debt cross-holding can still be used asa screening device when the effort levels are endogenous

Consider a given contract (r c a ) for cross-holding of debt andequity and a group consisting of two types of rms i and j Weshow that for any given contract (r c a ) the assortative-matchingresult still goes through when the effort levels are endogenous Nextwe show how a bank can screen borrowers by offering contracts thatdiffer in the extent of cross-holding of debt and equity For rm i thedecision problem is to

maxa i bi

R r(1 c ) a p i(a i bj ) (1 a )p j (a j bi)

pi (a i bj )pj (a j bi ) c r 12 c1a

2i

12 c2b

2i

where p i(a i bj ) p i a i bj and pj (a j bi ) p j a j bi We focus onthe choice of effort levels by group members that constitute a Nash

13 The latter interpretation is favored by Itoh (1991) The moral-hazard part of ourstory is similar to his model

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 607

equilibrium14 Solving for the optimal values of a i and bi we get

a i

a R r(1 c ) c rpj

c1

bi

(1 a ) R r(1 c ) c rpi

c2

Substituting the values of a i and bj in pi and those of a j and bi in pj

we get

p i pi A Bp j

p j pj A Bp i

where A R r(1 c ) a c1 (1 a ) c2 and B c r 1 c1 1 c2 Solving simultaneously we get

p ij

11 B2 A(1 B) pi Bpj

where p ij is the probability of success for type i when its business-group partner is of type j In the choice of the optimal contracts weneed to make sure that c and r satisfy the condition B lt 1 in order tohave pij gt 0

It is to be noted that the amount of effort supplied by the man-ager of rm i on her own project depends on the type of project itspartner rm j has because of the presence of a cross-guarantee Ifrm j is very likely to fail then rm i has lower incentives to supplyeffort to its own project because of the higher level of expected cross-guarantee payments In contrast the effort supplied by the managerof rm i on the project of its partner rm j is increasing in her owntype If rm i is more likely to succeed then rm j rsquos expected cross-guarantee payments are lower Since rm i gets a share of rm j rsquosprots her incentive to supply monitoring or helping effort is higher

It is straightforward to check that Proposition 1 goes throughie frac14ss frac14rr gt frac14rs frac14sr The proof is in the Appendix The onlydifference from the previous section is that now the probability ofsuccess of a rm depends not only on its own type (as before) butalso on the type of the rm that it is grouped with As before the valueof having a safe rm as partner is higher for a safe rm because itis more likely to be in a position to make cross-guarantee paymentsMoreover in this case its own probability of success is higher if it

14 Since these efforts are unobservable to group members we do not have to worryabout the possibility of collusion

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

608 Journal of Economics amp Management Strategy

has a safe partner through the choice of effort which reinforces theformer effect

For a given (r c ) we can derive the optimal value of a by max-imizing the payoff of a representative i-type borrower with respect toa The relevant condition is (see the Appendix for details)

R r(1 c ) a pi i c rR r(1 c ) (1 a ) p ii c r

c2

c1

1 c r1 B

1c1

1c2

1 c r1 B

1c1

1c2

Notice that the right-hand side is increasing in c2 and decreasing in c1while the left-hand side is increasing in a Also for c1 c2 a 1

2 Itimmediately follows that a gt 1

2 if c2 gt c1 and a lt 12 if c2 lt c1 Intu-

itively if eliciting own effort is less costly than eliciting monitoringeffort (ie c1 is lower than c2) the level of equity holding should behigher in onersquos own rm than in a partner rm and vice versa if c1 ishigher than c2 The next step of our analysis will be simplied if weassume that the difference between c1 and c2 is not very large so thata is in the neighborhood of 1

2 We can show that there exists a pair of nancially interlinked

contracts (rs c s a s) and (rr c r a r ) such that risky projects with neg-ative net surplus that were being funded under a standard poolingdebt contract can be excluded The proof is in the Appendix

232 Risk Sharing An alternative way to derive optimalequity cross-holdings is one using a risk-sharing model Let us retainour basic model of Section 2 and add the feature that the borrowersare risk-averse and there is no market insurance available Then ef-cient risk sharing dictates that rms within a business group smooththeir income streams by holding claims on each otherrsquos projectsrsquoreturns

The expected payoff of a borrower of type i that forms a businessgroup with a borrower of type j is now

frac14ij pi pj u(R r) pi (1 p j )u( a R r(1 c ) )

(1 pi )pj u((1 a ) R r(1 c ) )

Faced with any contract (r c ) any two rms would optimally sharerisk by choosing a 1

2 which follows from maximizing frac14ij frac14j i withrespect to a Intuitively risk sharing within the group implies havingthe same income in each state of the world Given this we show thatthe assortative-matching and single-crossing properties still apply In

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 609

particular the expected gain of a risky borrower from having a safepartner is

pr(ps pr) u(R r) uR r(1 c )

2

while the expected loss of a safe borrower from having a risky part-ner is

ps(ps pr) u(R r) uR r(1 c )

2

The former is less as ps gt pr Given assortative matching theexpected payoff of a type i borrower is

frac14i i p2i u(R r) 2pi (1 p i)u

R r(1 c )2

The slope of an indifference curve of a type-i borrower in the (r c )plane is

drd c

11 c

rp i

1 pi

u (R r )

u R r(1 c )2 r

It is clear upon inspection that indifference curves still satisfy thesingle-crossing property From this the proof of Proposition 2 can bestraightforwardly adapted to show that a pair of screening contractsexist such that only safe borrowers borrow in equilibrium

24 Other Extensions

In the basic model we assumed that shocks facing rms are perfectlyuncorrelated Our model can be extended to allow for partial correla-tion in these shocks However if the shocks are perfectly correlatedthen cross-holding of debt will not have any real effect and our resultswill no longer hold If all rms within the business group fail thenthere will be no debt repayment nor any cross-guarantee paymentsIf they all succeed there is no need for any cross-guarantee paymentsHence shocks facing rms within a business group must have anidiosyncratic component for business groups to solve the adverse-selection problem This assumption is justied by the fact that mostbusiness groups tend to be diversied or heterogeneous in their busi-nesses15

15 Fismanrsquos (2001) paper on political connections in Indonesia is an example of asituation when shocks could be correlated

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

610 Journal of Economics amp Management Strategy

The assumption of competitive credit markets is without loss ofgenerality If the bank is a monopolist maximizing its expected protsthen the optimal contracts will be similar to those derived in the basicmodel but they will lie on the respective participation constraints ofthe rms as opposed to the zero-prot constraints of the bank

Cross-holdings of debt and equity in our model are symmet-ric within a business group We allow borrowers to differ in a singlecharacteristic namely the riskiness of their projects and show thatborrowers are going to sort in terms of this single characteristic Thisleads to perfectly homogeneous groups and hence symmetric con-tracts In reality however these contracts are often asymmetric rmA may hold a share a A of its own returns while rm B may holda B a A of its own returns The same is possible regarding the extentof cross-holdings of debt c A separate source of heterogeneity otherthan riskiness can be added to the model to generate asymmetric con-tracts within groups A complete analysis of the issue of multidimen-sional heterogeneity is beyond the scope of this paper and so we limitourselves to the following example based on the mutual-monitoringmodel Suppose after groups are formed and contracts are signed (sothat regrouping is not feasible) each rm receives a mutually observ-able shock to the cost of monitoring with some probability In partic-ular the cost of monitoring its partner rm becomes very high (iec2 ) for a rm affected by this shock Then there will be somebusiness groups for which the members (say A and B) will have dif-ferent costs of monitoring (ie cA

2 lt cB2 ) and hence the optimal equity

cross-holdings will be asymmetric (ie a A 1 gt a B)16

In our model we assume that borrowers have no wealth at allwhich is clearly an unrealistic assumption Suppose borrowers havesome wealth w gt 0 In that case the bank will ask borrowers to pledgesome collateral which is taken away if the project fails If this wealthlevel is high enough the use of collateral will be sufcient to screenout risky borrowers For our results to go through all that is needed isa binding limited-liability constraint that implies that even if collateralis used risky borrowers still prefer to borrow under a standard debtcontract

3 Empirical Implications

The analyses of the previous sections suggest ways to interpret exist-ing empirical literature in the area and avenues for further empiricalresearch In this section we outline some of these ideas

16 Other potential sources of heterogeneity are the protability of the rm (R)how much wealth a rm can put up as collateral (or equivalently how much capital isneeded) and risk aversion

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 611

The model suggests that the cost of borrowing should be cor-related across group members and that the degree of cross-holdingshould be negatively correlated with the cost of borrowing We believethese implications are potentially testable using data on emerging-market business groups

For instance several of the prominent business groups in coun-tries such as South Korea and Mexico where the prevalence of cross-guarantees has been noted borrow funds on the international debtmarket and consequently have been assigned international risk assess-ments (ratings) by investment advisory rms (such as Moodyrsquos GlobalInvestor Service) The interest rate that these groups pay on such debtis also public information Information on the extent of intragroupcross-liability is not available from public sources but in principleit too should be obtainable This would then allow the testing of theimplications with regard to positive assortative matching in the forma-tion of business groups and the screening of such groups by externallenders

Another interesting implication of our theory for the study ofbusiness groups is their hierarchy which is a hitherto unexploreddimension to such organizations17 If as the theory suggests businessgroups are composed of rms with similar quality characteristics thenat rst blush we would expect this to mean that actual groups shouldbe composed of rms producing similar products And indeed wecan use this interpretation to understand actual business groups com-posed of rms operating in similar markets18 But how can we explainthe existence of business groups composed of rms engaged in verydiversied markets19

There are two answers to this First while these rms may beengaged in very different markets their activities may be similar inthe quality dimension20 Second diversication in activities make theshocks facing group members less likely to be correlated and the effect

17 It should be noted however that this implication is not unique to the specicmechanism for positive assortative matching suggested in this paper

18 Such as Grupo Cemex of Mexico which has rms engaged in the productionof cement contracting for bridge and building construction and producing ancilliaryconstruction materials

19 For example the House of Tata in India has interests in steel watches detergentstea automobiles and computer software Grupo Luksic of Chile has interests in bankshotels mining beer and pasta Grupo Carso of Mexico has rms in telecoms Internetservices television department stores and nance See ldquoWhen Eight Arms Are BetterThan Onerdquo The Economist Sept 12 1998 pp 67ndash68

20 Thus although the House of Tata in India has interests in very diverse markets(steel watches detergents tea automobiles and computer software) they are all per-ceived to be similar in qualitymdashin this case high quality since Tata is considered aldquoblue-chiprdquo brand See Khanna et al (1998) for more details on Tata

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

612 Journal of Economics amp Management Strategy

of debt cross-guarantees on efciency crucially depends on the shocksbeing not perfectly correlated If they were perfectly correlated debtcross-guarantees would never bite Either all rms would succeed inwhich case there would be no need to pay up loan guarantees for amember rm or all rms would default in which case there wouldbe nobody to pay loan guarantees This perspective leads to a relatedimplication We can rank business groups according to the quality oftheir projects suggesting a quality-based hierarchy In fact in manycountries we do observe a class hierarchy of business groups fromthe so-called ldquoblue-chiprdquo groups on down In combination with theprevious interpretation this implies that we should expect to observethis kind of hierarchy among groups of related rms as well as amonggroups of diversied rms Testing this would require obtaining riskratings both on group-member rms and on the overall group Forprominent business groups in several emerging economies one shouldbe able to obtain or construct such indices

The above observation is related to another robust recent empir-ical nding in the business-group literature Studies covering variouscountries [see Khanna (2000) for an excellent survey] nd that rmsassociated with business groups show better nancial performanceand productivity as well as better risk sharing than unafliated rmsWhile these may be explained by the presence of better mutual mon-itoring and risk sharing among business groups our paper suggeststhe possibility of reverse causality In particular low-risk and high-productivity rms are precisely those that are likely to form businessgroups

Recent empirical work on Chilean business groups by Khannaand Rivkin (1999) nds that equity interlocks explain a limitedamount of covariance between earnings of business-group membersThis suggests there are other mechanisms of nancial interlinkagethat contribute to the observed covariance such as debt guarantees orintragroup loans which are the main focus of this paper21 Our papersuggests that future empirical work should pay greater attention tothese alternative instruments of nancial interlinkage

An implication of the model is that if an economy is very net-worked and rms have access to good information about each othernancially interlinked debt contracts of the kind described in thepaper should improve repayment rates On the other hand ineconomies with low levels of networking such contracts should reducerepayment rates In a large economy such as India it is well known

21 The paper by Lincoln et al (1996) also nds evidence of business groups func-tioning as redistributive income-smoothing mechanisms

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 613

that some business communities (such as the Marwaris) have muchstronger close-knit networks than others In principle it should be pos-sible to obtain information on business groups composed of membersof these communities and their repayment rates on loans to test thisimplication

Our results imply that the cross-holding of debt can be inter-preted as a way to ameliorate adverse-selection problems in the creditmarket (in the absence of collateral) while cross-holding of equityprovides incentives to mitigate moral-hazard problems It is unlikelythat either type of cross-holding by itself will solve both types ofasymmetric information problems Therefore equity and debt cross-holding arrangements may or may not go together depending onwhether both kinds of asymmetric problems are prevalent or not in agiven economy Empirically this suggests we might be able to use thepresence or absence of each of these types of arrangements togetherwith the presence or absence of collateral as a marker for the kindof asymmetric-information problem that is more serious in a giveneconomy

The only role played by equity in our theory is as a device toelicit monitoring effort or to share risk Debt has been the sole instru-ment available to rms for obtaining needed investment nance Thisof course ignores the important role that the sale of shares plays inraising external funds But the raising of capital through the sale ofequity depends crucially on the existence of a well-functioning stockmarket and associated nancial intermediation A principal activity ofthese institutions is in fact the gathering and processing of informa-tion about rms thereby easing the asymmetric information that is atthe heart of our approach

In many emerging and transition economies the nancial sec-tor is still underdeveloped Specialized nancial intermediaries thatperform informational and monitoring services are absent or thereis a serious lack of skills and incentives in such intermediaries asdo exist22 Stock markets do not work well Indeed there is empir-ical evidence that in many developing countries stock markets areatrophied with limited otation and few listed rms [see Castaneda(1999) for Mexico and Pistor (1999) for the Czech Republic Hungaryand Poland] Our model ts into such environments Conversely asthe nancial sector develops and information problems in the econ-omy become less severe we should expect to observe an unravelingof the kinds of ties that bind group rms that we have focused on in

22 For a theory of this see Holmstrom and Tirole (1993) For empirical work seeKhanna and Palepu (1998b)

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

614 Journal of Economics amp Management Strategy

this paper This may be one way to understand why such businessgroups are rare in developed countries such as the US

In a related vein the past three years of nancial crisis in SouthKorea may provide a laboratory to test our theory As mentionedearlier the clearest examples of debt and equity interlinkage comefrom the Korean chaebols It is conceivable that the recent aggregateeconomic shock has also had idiosyncratic effects at the rm levelthat have altered the distribution of riskiness across Korean rms Interms of our model this means that the type of some rms may havechanged which would imply a change in the composition of groupsAs we might expect in consequence there is currently underway aprocess of unraveling of existing relationships and restructuring ofmany of the large Korean business groups23 Once the economy sta-bilizes it should be possible to examine whether the new groupingsreect a different distribution of riskiness across rms the way ourassortative matching result suggests

Recent cross-country empirical work across a spectrum of emerg-ing economies24 by Singh (1995) shows that in developing countriesexternal nance takes precedence as a source of funds for rms Thisis the reverse of the ldquopecking orderrdquo pattern of nance found inadvanced economies wherein rms mostly use retained prots tonance their investment needs followed by long-term debt withequity nance only as a last resort Since rms in emerging economiesare compelled to look toward external nance credit rationing islikely to be a serious problem on account of the absence of adequatenancial intermediation in these economies Our analysis suggeststhat nancially interlinked business groups can be interpreted as asolution to this problem

Appendix

A1 Proof of Lemma 2

It is easy to verify that the left-hand side of (6) is an increasing func-tion of c s and assumes the value 2 ps(2 ps) q for c s 1 Simi-larly the right-hand side of (5) is an increasing function of c s andassumes the value pr(2 pr) ps(2 ps) q for c s 1 Observe thatprR u lt R and for any c s [0 1] (1 c s) ps 1 c s(1 ps) gtpr 1 c s(1 pr) ps 1 c s(1 ps) As R gt q ps [by (1)] for c s 0

23 See for instance the article ldquoEntrepreneurial Fresh Airrdquo in The Economist Jan 112001

24 India Republic of Korea Jordan Pakistan Thailand Mexico Malaysia Turkeyand Zimbabwe

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 615

condition (6) is satised with strict inequality and so it also holds forc s small enough However for c s 0 condition (5) cannot hold as itis equivalent to the condition pr R u (pr ps) q which is ruled outby Assumption 2 [equation (2)] which says there is some inefciencyunder standard debt contracts A necessary condition for the existenceof a c s [0 1] that satises (5) is

pr(2 pr)ps(2 ps)

q prR u (A1)

Note that x(2 x) is increasing in x for x [0 1] Therefore pr (2 pr ) ltps(2 ps) and the assumption that pr R u lt q in Assumption 1[equation (1)] is not sufcient to ensure the condition (7) will holdHowever while necessary (7) is not sufcient as we have to checkwhether (6) is satised as well There are two cases to consider If(6) is satised for c s 1 which will be the case if R 2 ps(2ps) q then it is satised for all c s [0 1] and so we have provedthe existence of a critical value c (0 1) such that for all c s c ascreening contract exists Suppose R lt 2 ps(2 ps) q instead Thenthere exists a c s (0 1) such that (6) holds with equality This valueof c s is c (psR q ) q ps(1 ps)R Note that as by assumptionR lt 2 ps(2 ps) q in this case we have c (0 1) A necessary andsufcient condition for a screening contract to exist in this case is thatc must satisfy (5) Straightforward algebra shows that this condition ispsR q (p2

s p2r )u Observe that as ps gt pr this condition is consistent

with (1) which requires psR gt q u This completes the proof

A2 The Mutual-Monitoring Case

A21 Proof of Assortative Matching The expected pay-off of a rm of type i that has a type-j rm in its group when facingthe contract (r c a ) is

frac14ij R r(1 c ) a p ij (1 a )pj i r c p ij pj i12 c1a

2i

12 c2b

2i

where pij 1 (1 B2) A(1 B) p i Bpj a i [a R r(1 c )c rpj i] c1 and bi [(1 a ) R r(1 c ) c rpij ]c2 Then

2frac14ij

p ipj

r c

(1 B2)2 1r c

c2B B B

r c

c1gt 0

since B r c (1 c1 1 c1) lt 1 by assumption Since the types ofborrowers are complementary in the payoff functions the assortativematching result follows directly from Becker (1993 Ch 4)

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

616 Journal of Economics amp Management Strategy

A22 Derivation of the Optimal Value of a Sincegroups are homogeneous the direct effect of a change in a cancelsout By the envelope theorem we can ignore the effect of a change ina on frac14ii through variations in the level of the own effort level of agiven rm Rather the choice of a is based on its effect on the otherrmrsquos effort level Differentiating frac14ii with respect to a and using thisfact we get

[ R r(1 c ) a p ii c r]bi

a[ R r(1 c ) (1 a ) pi i c r]

a i

a0

Sincea i

a

R r(1 c )c1

1c r

1 B1c1

1c2

andbi

a

R r(1 c )c2

1c r

1 B1c1

1c2

the condition in the main text follows

A23 Proof of Existence of Separating ContractsNotice that Lemma 1 applies with a minor modication if incentive-compatible contracts (rs c s a s) and (rr c r a r ) exist then assortativematching still takes place The proof consists of four steps

Step 1 Take a given value of a (0 1) throughout this analysis Con-sider a contract (r c ) such that the bank makes zero expected protsif only safe borrowers borrow under that contract By construction

rpss 1 (1 pss) c q

We want to show that under some conditions if risky borrowers wereto borrow under this contract their expected payment to the bankwould be higher than that of safe borrowers Notice that both pss andprr depend on the specic contract (r c ) through endogenous effortchoice The condition that

rprr 1 (1 prr) c gt rpss 1 (1 pss) c

simplies to

pss prr gt1

1 c1

Clearly this condition cannot hold for c 0 However for a largeenough value of c it will be satised For example when c 1 thiscondition is equivalent to ps pr gt 3

2 2A (1 B) (1 B) Since B lt 1and A (1 B) lt 1 by assumption the condition will be satised fora range of values of ps and pr

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 617

Step 2 In the previous step we considered only a part of the expectedpayoff of a rm namely the expected payment to the bank Considerthe remaining component of the expected payoff of a borrower oftype i (say frac14ii ) namely p iiR

12 c1a

2i

12 c2b

2i As before consider a

contract (r c ) such that the bank makes zero expected prots if onlysafe borrowers borrow under that contract ie rpss 1 (1 pss) c q Consider a simple debt contract (r 0) such that pssr q We wish toshow that frac14rr(r c a ) lt frac14rr (r 0 a ) ie

(pr a r br)R12 c1a

2r

12 c2b

2r lt (pr ar br )R

12 c1a

2r

12 c2b

2r

where a i [ a R r(1 c ) c rp ii ] c1 and bi [(1 a ) R r(1 c )c rpii ] c2 are the effort choices of a rm of type i under the contract(r c ) while a i a (R r ) c1 and bi (1 a )(R r ) c2 are effortchoices of a rm of type i under the contract (r 0) Since effort levelsare lower than what would be achieved if these were contractible(namely a R c1 and b R c2) if we can show that ar gt a r andbr gt br the proof will be complete That is we require

a r(1 c ) r c rprr gt 0 (1 a ) r(1 c ) r c rprr gt 0

Recall that by construction rpss 1 (1 pss) c pssr Therefore r(1c ) r c rpss gt c rprr Hence so long as min a 1 a pss gt pr thiscondition will be satised So long as ps is large enough compared topr and a is neither too close to 1 nor too close to 0 that condition inturn will be satised Since we assume c1 and c2 are close enough weare guaranteed that a does not take extreme values

Step 3 We must make sure that safe borrowers are strictly better offunder the contract (r c ) than under the contract (r 0) Step 1 alreadyshows that their expected payment to the bank is the same Now welook at the remaining component of their payoff frac14ss The argument issimilar to the one used in step 2 but the aim is exactly the oppositeNow we want to show that as lt a s and bs lt bs whereas previously wewanted to show that ar gt a r and br gt br Since a r(1 c ) r c rpss

(1 a ) c rpss lt 0 and (1 a ) r(1 c ) r c rpss a c rpss lt 0our proof is complete

Step 4 Starting with a situation where risky borrowers borrowingunder a standard pooling debt contract (r 0) suppose the contract(r c ) is offered Safe borrowers will be better off and under someparameter conditions risky borrowers will be strictly worse off If theexpected payoff of risky borrowers frac14rr falls below u then they willwithdraw from the credit market thereby improving efciency

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

618 Journal of Economics amp Management Strategy

References

Aoki M 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 193ndash70

Bebchuk L R Kraakman and G Triantis 1998 ldquoStock Pyramids Cross-Ownershipand Dual Class Equity The Creation and Agency Costs of Separating Control fromCash Flow Rightsrdquo Mimeo Harvard Law School

Becker G 1993 A Treatise on the Family Cambridge MA Harvard University PressBerglof E and E Perotti 1994 ldquoThe Governance Structure of the Japanese Financial

Keiretsurdquo Journal of Financial Economics 36 259ndash284Bester H 1987 ldquoThe Role of Collateral in Credit Markets with Imperfect Informationrdquo

European Economic Review 31 887ndash899Brioschi F L Buzzacchi and M Colombo 1989 ldquoRisk Capital Financing and the Sepa-

ration of Ownership and Control in Business Groupsrdquo Journal of Banking and Finance13 747ndash772

Calomiris C 1990 ldquoIs Deposit Insurance Necessary A Historical Perspectiverdquo Journalof Economic History 50 283ndash295

Castaneda G 1998 La Empresa Mexicana y Su Gobierno Corporativo (The Mexican Firmand Its Corporate Governance) Alter EgoUDLAP Press

1999 ldquoLaggard Economic Growth and Ill-Functioning Stock Marketsrdquo Mimeo Uni-versidad de Las Americas Puebla Mexico

Clayton MJ and BN Jorgensen 2000 ldquoCross Holding and Imperfect Product Mar-ketsrdquo Mimeo Harvard Business School

De Meza D and D Webb 1987 ldquoToo Much Investment A Problem of AsymmetricInformationrdquo Quarterly Journal of Economics 102 (2) 281ndash292

Fisman R 2001 ldquoEstimating the Value of Political Connectionsrdquo American EconomicReview forthcoming

Ghatak M 2000 ldquoScreening by the Company You Keep Joint Liability Lending andthe Peer Selection Effectrdquo Economic Journal 110 (465) 601ndash631

Ghemawat P and T Khanna 1998 ldquoThe Nature of Diversied Business GroupsA Research Design and Two Case Studiesrdquo Journal of Industrial Economics 66 (1)35ndash61

Goto A 1982 ldquoBusiness Groups in a Market Economyrdquo European Economic Review 1953ndash70

Granovetter M 1994 ldquoBusiness Groupsrdquo in NJ Smelser and R Swedberg eds TheHandbook of Economic Sociology Princeton University Press

Holmstrom B and J Tirole 1993 ldquoMarket Liquidity and Performance MonitoringrdquoJournal of Political Economy 101 (4) 679ndash707

Itoh H 1991 ldquoIncentives to Help in Multi-agent Situationsrdquo Econometrica 59 (3)611ndash636

Kali R 1999b ldquoBusiness Groups the Financial Market and Modernizationrdquo MimeoUniversity of Arkansas

Khanna T 2000 ldquoBusiness Groups and Social Welfare in Emerging Markets ExistingEvidence and Unanswered Questionsrdquo European Economic Review 44 748ndash761

and K Palepu 1998a ldquoIs Group Afliation Protable in Emerging Markets AnAnalysis of Indian Diversied Business Groupsrdquo Working Paper 96-051 HarvardBusiness School Journal of Finance 55 (2) 867ndash891

and 1998b ldquoEmerging Market Business Groups Foreign Investors and Cor-porate Governancerdquo Working Paper No 6955 Cambridge MA National Bureau ofEconomic Research

Financially Interlinked Business Groups 619

and JW Rivkin 1999 ldquoTies That Bind Business Groups Evidence from ChilerdquoMimeo Harvard Business School

and Y Yafeh 1999 ldquoBusiness Groups and Risk Sharing around the Worldrdquo MimeoHarvard Business School

K Palepu and DM Wu 1998 The House of Tata The Next Generation (A) Case9-798-037 Harvard Business School Publishing

Keister L 2000 Chinese Business Groups The Structure and Impact of Interrm Relationsduring Economic Development Oxford University Press

La Porta R F Lopez-De-Silanes and A Shleifer 1999 ldquoCorporate Ownership aroundthe Worldrdquo Journal of Finance 54 (2) 471ndash517

Lincoln JR ML Gerlach and CL Ahmadjian 1996 ldquoKeiretsu Networks and Corpo-rate Performance in Japanrdquo American Sociological Review 61 67ndash88

Matsusaka JG 2000 ldquoCorporate Diversication Value Maximization and Organiza-tional Capabilitiesrdquo Journal of Business 74 (3) 409ndash432

Morduch J 1999 ldquoThe Micronance Promiserdquo Journal of Economic Literature 37 (4)1564ndash1614

Nakatani I 1984 ldquoThe Economic Role of Financial Corporate Groupingrdquo in M Aokied The Economic Analysis of the Japanese Firm New York North Holland

Pistor K 1999 ldquoLaw as a Determinant for Equity Market Development The Experienceof Transition Economiesrdquo in P Murrell ed Assessing the Value of Law in the EconomicTransition from Socialism Ann Arbor MI University of Michigan Press

Shin HH and YS Park 1999 ldquoFinancing Constraints and Internal Capital MarketsEvidence from Korean Chaebolsrdquo Journal of Corporate Finance 5 169ndash191

Singh A 1995 ldquoCorporate Financing Patterns in Industrializing Economies A Com-parative International Studyrdquo Technical Paper No 2 International Finance Corpo-ration

Stiglitz J and A Weiss 1981 ldquoCredit Rationing in Markets with Imperfect Informa-tionrdquo American Economic Review 71 (3) 393ndash410

Wolfenzon D 1999 ldquoA Theory of Pyramidal Ownershiprdquo Mimeo University ofMichigan