Sie sind auf Seite 1von 29

Islamic Banks and Investment Financing Author(s): Rajesh K.

Aggarwal and Tarik Yousef Source: Journal of Money, Credit and Banking, Vol. 32, No. 1 (Feb., 2000), pp. 93-120 Published by: Ohio State University Press Stable URL: . Accessed: 19/08/2013 05:19
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact

Ohio State University Press is collaborating with JSTOR to digitize, preserve and extend access to Journal of Money, Credit and Banking.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


IslamicBanksandInvestment Financing
Islamic Law prohibitscharginginterest.We study financialinstruments used by Islamic banks and find that most are not based on profit-and-losssharing (equity) but, instead, are very debtlike in nature.We see some bias against providing financing for agricultureandindustry.Long-termfinancingis rarelyofferedto entrepreneurs. Our model shows that debtlike instrumentsare a rational response by Islamic banks to their contractingenvironments.As agency problemsbecome more severe, debt becomes the dominant instrumentof H1nance. We give conditions under which banning debt increases social welfare as well as conditions under which banningdebt decreases social welfare. IN THE LAST TWODECADES,Islamic banks have grown in size and numberaroundthe world. In this paper,we examine the types of financial contractsoffered by Islamic banks.We try to understandany departures from traditional Islamic principlesin the types of contractsoffered. We suggest an economic rationalefor the constraintsimposed on Islamic banks and try to determineif these constraintsare likely to be social welfare improving.We also examine the types of projectsin which Islamicbanksinvest. In this process we hope to shed some light on the efficiency of Islamic banks and Islamic economies. Islamicbanksoperatein over sixty countries,most of them in the Middle East and Asia. In three countries, Iran, Pakistan,and Sudan, the entire banking system has been convertedto Islamic banking. In the other countries, the banking systems are still dominated by conventional banking institutions operating alongside Islamic banks. Even so, Islamic bankingis the fastest growing segment of the credit market in Muslim countriesthathave Islamic banks:theirmarketsharehas risen from 2 percent in the late 1970s to about 15 percenttoday,as measuredby assets in the banking system (Babai 1995). Even conventionalcommercialbanks have startedto offer IsThe authors thank Sheri Aggarwal, Hesna Genay, Michael Knetter,Raghu Rajan, Steve Sung, Jeff Williamson, and seminarparticipants at DartmouthCollege, HarvardUniversity,and the Middle Eastern Economic Association Meetings for their helpful comments. Special thanks to two anonymousreferees whose comments greatlyimprovedthe exposition of the paper.The authorsalso thankNazim Ali and the HarvardIslamic Finance InformationProject for providing some of the data and MarciaDiefendorf for help in preparingthe figures.

RAJESH K. AGGARWAL is associate professor of economics at DartmouthCollege. TARIK YOUSEF ISassistantprofessor of economics at GeorgetownUniversity. Journalof Money,Credit,and Banking,Vol. 32, No. 1 (February 2000) Copyright2000 by The Ohio State University

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions



lamic financial contracts. Unlike other commercial banks in the Middle East that were eithercreatedby or received extensive supportfrom governmentsduringthe oil booms of the 1970s and 1980s, Islamic banks are generally the productof private initiativeand have appearedpredominantly in non-oil-exportingcountries. Islamic banks are supposed to offer instrumentsconsistent with the religious beliefs and culturalcharacteristicsof Muslim societies. According to prevailinginterpretationsof Islamic Law, financial instrumentsshould emphasize profit-and-loss sharing(equity). Interestis prohibited,which seems to exclude debt contracts.This is in markedcontrastwith Westernfinancialinstitutions.Banks in the United States, for example, primarilyoffer debt contractsto firms seeki-ng capital (regulationsprohibit banks from taking more than a S percent equity position in firms). Banks in Germanyand Japanmake equity investmentsin firms, but also use debt contracts. Non-Islamic banks operatingin Muslim countriesare able to offer debt contractsto firms.Even governmentsof Muslim countries,such as Saudi Arabia,borrowon the international capitalmarkets.That SaudiArabiadoes so is fascinatinggiven thatthe Saudi monarchyderives its legitimacy from upholding Islamic Law and is a major promoterof Islamic economics. Nonetheless, Islamic Law prohibitsIslamic banks from eitherreceiving or paying interest. We considertwo issues. First, do Islamic banks operateaccordingto the principle of profit-and-losssharing?If not, why not? Second, is it social welfare improvingto have a strictban on debt?Who benefits and who loses from such a ban? On the first issue, our evidence indicates that most of the financing provided by Islamic banks does not conformto the principleof profit-and-losssharing.Instead,much of the financingprovidedby Islamic bankstakes the form of debt-likeinstruments. On the second issue, proponentsof Islamic bankingarguethatprofit-sharing contracts (equity) are superiorfinancialinstrumentsto debt for a variety of reasons including the risk-sharing propertiesof equity (Ebrahimand Safadi 1995). We discuss some of these arguments in section 2. In addition,advocatesof Islamic bankingsuch as Chapra(1992) and Siddiqi (1983) have arguedthat Islamic banks will promote growth in Islamic countries by providing long-term financing to growth-oriented sectors of the economy. Contraryto the expectationsof Islamic banking'sadvocates, we find that Islamic banks rarelyoffer long-termfinancingto entrepreneurs seeking capital. In addition, the majorityof Islamic banks' financial transactionsat least initially were directed away from agricultureand industryand towardretail or trade financing.Further,it appearsthat much of the "lending"done by Islamic banks is secured, violating a legal prohibitionon collateral.While this evidence does not fully answer the question about the social welfare propertiesof a ban on debt, it does suggest that the claimed benefits of Islamic bankingmay be somewhatoverstated. In orderto understand why Islamic banks do not operateaccordingto the priniciple of profit-and-losssharingandto understand the social welfarepropertiesof a ban on debt, we develop a model of Islamic economies and Islamic banks. Islamic banks operate mostly in developing economies where financialmarketsare characterized by high degrees of imperfectinformationand rent-seekingbehavior.Data on corrup-

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


inefficiency for Iran, Pakistan, and Egypt suggest that these tion and bureaucratic countrieshave somewhatinefficienteconomies in which there are fairly high levels of rent seeking and corruption(Mauro1995). Of course, corruptionand inefficiency are widespreadin most developing economies, not just Islamic ones. In these counmay be willing to use funds from their firms for their own tries, many entrepreneurs investments.In such cases, consumptionof perquisitesor wasteful, negative-return banks providing financingface an agency problem.In our model, the agency problem arises from contractualincompleteness,as in the models of Bolton and Scharfstein (1990) and Hart and Moore (1998). The crucial assumptionis that the cash projectare not verifiableto a court of law. In flows (profits)from the entrepreneur's the Bolton and Scharfstein(1990) and Hartand Moore (1998) models, this assumpor equity contractsas financialinstrution rules out the use of outside profit-sharing ments as these are based on sharingthe cash flows. In our model, by weakening an utility function, we are able to characterizewhen assumptionon the entrepreneur's outside equity contractswill be optimal financialinstrumentsfor the bank. by agency problems will be biased toward We find that economies characterized debt financing.As these problemsbecome more severe, we show that debt will become the dominantinstrumentof finance.We also show thatequity financingcan be optimaland,in certaincircumstances,a ban on debt can be social welfareimproving. We draw connections to the literatureon optimal capital structureand highlight the importanceof agency problems and contractualincompleteness.We argue that the use of debtlikeinstrumentsis a rationalendogenous response on the partof Islamic banksto the contractingenvironmentsin which they operate. Several papers in the literaturehave examined restrictions on the use of debt. Glaeser and Scheinkman(1998) examine restrictionson usury and show that limiting the rateof interestcan be social welfareimprovingwhen marketsareincomplete. In theirmodel, banninginterestaltogetherwould not be social welfare improving,as it can be in our model. In work on Islamic banking,Khan and Mirakhor(1987) and Kuran(1993) have noted that adverse selection can result in the use of debtlike instrumentsby Islamic banks as well. We begin by describingthe financialcontractsoffered by Islamic banksin section 1. In section 2, we discuss the patternsand composition of credit that have emerged in Islamic banks.We presentour model in section 3. In section 4 we derive optimal contracts for the bank from the set of profit-sharingor pure equity contracts and combinationsof debt and equity contracts,which nest the special case of pure debt contracts.In section 5, we comparethe contractsfrom section 4 and examine social welfare implications.Section 6 concludes.

The Islamic legal principlesthat regulatethe conduct and content of commercial transactionsin Islamic banking date back to the early days of Islam in Arabia.The Muslim scholars of the Middle Ages made elaborateefforts to establish the funda-

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions



mentalprinciplesof finance and commerce.These principlesare supposedto govern economic activity for Muslims today.The most importantof these principlesis the prohibitionof riba, any predetermined or fixed returnin financial transactions.As stated in the Quran: 'cAllah forbids ribaand permitstrade." While there is much debate aboutthe exact natureof this prohibitionon riba,thereexists a widespreadperception that the ban on ribaimplies a ban on interest.Because this is the prevailing interpretation, we follow it in this paper. Alternative"interest-free" financing techniqueshave been developed by Islamic banksand the monetaryauthoritiesof severalcountries.The instruments of commercial financinghave been based on two principles:the profit-and-losssharing(PLS) principle and the markupprinciple.The PLS principle is unanimouslyaccepted in the Islamic legal and economic literaturesas the cornerstoneof financial transactions. According to the PLS principle,the bank may earn a returnon invested funds providedthatthe bank sharesin the risk of the investmentand bearsa loss if the project fails. Islamic banks utilize two instruments based on this principle:

* Mudarabah financing, where the bank provides capital and the entrepreneur
contributeseffort and exercises complete control over the business venture. In case of a loss, the bankearnsno returnor a negativereturnon its investmentand the entrepreneur receives no compensationfor her effort. In case of a gain, returnsare split accordingto a negotiatedequity percentage. * Musharaka financing, where the entrepreneur and the bank jointly supply the capital and manage the project.Losses are borne in proportionto the contribution of capitalwhile profitproportionsare negotiatedfreely. Both of these instrumentscan be thoughtof as equity investments,althoughmu-

darabah financingmay be more akin to a limited partnership and musharaka financing is closer to a traditional equity stake with rights of control. The markupprinciplehas its historical roots in commercial trade activities. The bank finances the purchase of assets in exchange for a negotiated profit margin. There are two widely used instrumentsin this category:

* Murabaha financing,where the bank purchasesan asset on behalf of an entrepreneur.The bank resells the asset to the entrepreneur at a predetermined price that covers the originalcost and an added,negotiatedprofitmargin.Paymentis made in the futurein lump sum or in installments.Ownershipresides with the bank until all paymentsare made. Murabaha financingis the classic instrument for tradefinancing,datingto ninth-century Arabia. * Ijarafinancing,where the bankpurchasesthe asset and allows the entrepreneur to use it for a fixed charge. The ownershipof the asset either remains with the bank or is graduallytransferred to the entrepreneur in a rent-to-own contract. Ijarafinancingis the traditional contractfor what is known as leasing today. Although markupinstrumentsare widely used, their acceptabilityunder Islamic Law is disputedbecause they can imply a fixed returnon investmentfor the bank. Many Islamic scholars have taken the position that markuptechniques, while per-

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


missible, should still be avoidedor restricted(Siddiqi 1983 and Khan 1987). Legally, the fear is that markupfinancingmay open a "backdoor"to interest.Economically, observersworry that markupfinancingmay stunteconomic growthby constraining from investing in new projects.In sections 4 and 5, we show that this entrepreneurs economic argumentagainstmarkupfinancingis flawed:markupinstrumentsgenerally expand the set of projects that can be undertaken.Nevertheless, an economic case can be made againstmarkupinstruments,which we discuss in section 5. We wish to stress thatthereis a formalequivalencebetween markupfinancingand debt, but the equivalence is not based on the paymentof interest.Following the incomplete contracts and control rights literature(Hart and Moore 1998), we argue thatthe salient featureof debt is thatit transferscontrolof an asset to the debtholder to disgorge cash in cases of default. This has the effect of forcing the entrepreneur flows by making regularly scheduled payments to prevent default. What prevents banksfrom contractingdirectlyover cash flows, ratherthanusing the indirectinstrument of debt?The assumptionis thatcash flows may be observableby the partiesbut not verifiableto a court.We will discuss these assumptionsin greaterdetail in section 3. The critical featureof markupcontractsis that the bank retainsownershipof the asset and can seize it in cases of default. Under PLS contracts,the bank has no Thus it is with the entrepreneur. such direct claim on the asset as it is in partnership it that differentiate markup financing conferred by the control rights over the asset from PLS financing. The control rights of markupcontracts are equivalent to the controlrights of debt contracts. Westerndebt contrasts One of the benefitsof markupcontractsrelativeto standard is that,in cases of default,thereis no ambiguityaboutcontrolof the assets. The bank retainstitle to the asset untilall paymentsaremade.In mostWesterncountries,default continuesto proceedingsduringwhich the entrepreneur/manager triggersbankruptcy law). Becauseof the decontrolthe assets (forexample,Chapter11 of U.S. bankruptcy proceedingsin the shiftin controlof the assets,barlays inducedby formalbankruptcy of gaining problems are introduced that can significantly decrease the efE1ciency contracts. investment.In principle,these problemsare avoidedunderIslamicmarkup While the rationalefor banninginterestis unambiguouslyrootedin theology, proponentsof Islamic bankinghave also found economic argumentsto supporta ban on interest.Many of these argumentsare similar to those put forth in medieval Christianity to restrict usury. Some of the economic rationales for the superiority of Asprofit-and-losssharingover the use of interestare describedby the International pp. 3-4): Islamic Banks (1995, sociation of
If interest is replaced by profit sharing, some imbalances are expected to be reduced. Allocation of investablefunds First, the returnon capital will depend on productivity. will be guidedby the soundnessof the project.This will in effect improvethe efficiency of capital allocation. Second, the creationof money by expandingcredit will be createdonly when there is a strong likelihood of a correspondingincrease in the supply of goods and services. In case the enterprise loses, repayment of capital to the bank is diminished by the system, the supplyof money is not allowed to amountof loss. Thus in the profit-sharing overstepthe supply of goods and services. This will eventuallycurb inflationarypressures in the economy.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions



Third,the shift to profit sharingmay increase the volume of investmentsthat translates into job creation. This is because the interest mechanism makes feasible only those projectswhose expected profitsare sufficientlyhigh to cover the interestrateplus added income. This filters out projects which otherwise would be accepted in the profit-sharing system. Fourth, the new system will also ensure more equitable distribution of wealth. Wealthwould bringmore wealth to its owners only when its use has actuallyresultedin the creationof additionalwealth. This would in time reduce the unjustdistributionof wealth which continuedfor decades duringthe interestregime. Fifth, the abolitionof interest,togetherwith the restrictionof forwardtransaction, as prescribedby Sharia, will curtailspeculationsmeasurably. But still, therewill be a secondarymarkettradingcommon stocks and investmentcertificatesbased on profit-sharing principles.This will bring sanity back to the marketand allow raising of funds for enterprisesand liquidity to equity holders.

Point 1 seems to suggest thatequity investorscare aboutthe qualityof the projects in which they invest while lenders are indifferentto the quality of the projects to which they lend. On the other hand, point 3 seems to suggest that lenders constrain the set of projectsto which they lend to the profitableones. While these points are contradictory, point 3 does espouse the commonly held view that lendersrestrictaccess to capital.Point 1 can also be interpreted as saying thatbanningdebt will eliminate conflicts between equityholdersand debtholderssuch as the asset substitution problem.We discuss this in section 5. Point 2 suggests thatthereis a macroeconomic benefit to a ban on interest associated with limiting the amount of liquidity in an economy and hence limiting inflationarypressures.This implies that the most importanteconomic issue facing developing economies is inflation, ratherthan low growthrates. It is not obvious why restrictingthe supply of money will not also restrictthe supply of goods and services. Further, it is likely thateliminatingthe credit channelof monetarypolicy will prove detrimental in cases of recession, so it is hard to see a clear-cut benefit associated with this point. Point 4 does not obviously hold it is not clear why the chargingof interestnecessarily leads to a skewed distributionof wealth, althoughprohibitionson usuryclearly adoptthis viewpoint. Because this argument is frequentlymade, we discuss it in greaterdetail in section S in the context of our model. Point S arguesthatbanninginterestmay decreasespeculation, presumablybecause investors borrow on margin and speculate. However, the ability to borrowon margingenerally leads to an increase in stock marketliquidity, not a decrease as seems to be argued.Anotherinterpretation of point S is that abolishing interest will decrease volatility in the market.However, the relationshipbetween liquidity (as createdby borrowingon margin)and volatility is unsettled and liquidityoften decreasesvolatility in markets. From this critique of interest-basedbanking, the proponentsof Islamic banking seem to be arguingthat the primarysocial costs of interest are that it constrainsthe access of entrepreneurs to capital(point 3) and thatit leads to an unequaldistribution of wealth (point 4). Point 3 implies that debt financing funds fewer projects than does equity financing, and there is a social welfare gain when debt is banned.We argue in section S that a ban on debt may, in certain circumstances,increase social welfare. However,this gain in social welfare does not come because debt constrains

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


the number of projects undertaken.Debt generally expands the set of projects financed, but there is a deadweightloss on each projectdue to the possibility of default and liquidation.In addition,there may be other costs of debt associated with adverse selection, risk shifting, or direct costs of bankruptcy(lawyers, court costs). The social welfare gain from a ban on debt comes when these costs arc sufficiently more to appropriate high. Point 4 implies thatbanningdebt will allow entrepreneurs of the returnsfrom their investmentprojects. In section 5 we show that this can be true,but we also show thatbanningdebt is not the most efficient way to achieve this goal. We can thereforeprovidesome economic rationalesfor a ban on debt, although not quite on the same groundsas proposedby the proponentsof Islamic banking.We are the key reiterate,however,that if economic efficiency and income redistribution objectivesof Islamic banking,then therearebettermeans to these ends thana ban on debt. In additionto markupand PLS financing, as partof their mission, Islamic banks are encouraged to make charitableloans to individuals or organizationsthat need them in the form of funds or real assets (materials,supplies, etc.). These are termed QardHassan loans, or social or benevolentloans. These loans aremade at no charge, with no interest due, and with no mark-up.They are clearly negative NPV investments for Islamic banks.

The academic and policy interestin Islamic bankinghas been sparkedin partby the seeming rapidgrowthof their assets and marketshare in the financialsectors in Muslim countries,primarilythose in the Middle East. Over the last decade the assets of Islamic banks experiencedan annualgrowthrate of 19 percent.Our calculations show that Islamic banks had approximately$100 billion in assets in 1995. Today many Islamic banks are among the five largest banks in their respective countries. We caution, however, that there is much speculation about the size of the Islamic bankingmarketand the variancein estimatesis large (TheEconomist 1996). The advocates of Islamic banking have presentedthe rise of Islamic banking as the primaryalternativeto interest-basedbanking.Two problemsfacing the majority of Middle Easterneconomies are low investmentrates and weak financialintermedi(WorldBank 1995). Islamicbanksare supposedto serve the function ationstructures of allocatinginvestmentfunds to long-termproductiveprojects.They areexpectedto who do not have access to creditin the conventionalbankfavor small entrepreneurs ing system and to extend theirlinks to ruralregions thatare often cut off from formal access to urbanfinancialmarkets.Islamicbanksmay be an engine of growthin Muslim countries.Advocates argue that PLS financingprovides the vehicle for accomplishing these goals in a fair and efficient manner(Chapra1992 and Siddiqi 1983). We find little evidence to support these claims. First, Islamic banks rely much more heavily on markupfinancing than on PLS financing. Second, most financing does not appearto be long term in nature.Third,the evidence on whetheror not Is-

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


lamic banksprovidefinancingto capital-intensive sectorsof the economy such as industryis mixed at best. The data available on Islamic banks are somewhatlimited. In this study, we use the most comprehensivedata availablefrom the International Association of Islamic Banks, as well as data from annualreports and other sources. We reportwhat the data suggest and we caution that our conclusions are not definitive.There seems to be little reason to believe that Islamic banks operatemuch differentlythan conventional banks. Egypt's Faisal Islamic Bank (FIBE) provides an interestingcase study.FIBE was foundedin 1979 andis today the fourthlargestIslamicbankin the world andEgypt's sixth largest commercial bank (Kazarian 1993 and Ray 1995). ConvertingKazarian's (1993) data to constant 1985 Egyptian pounds shows that the total assets of FIBE grew quite rapidly from 208 million Egyptianpounds in 1979 to 2.75 billion Egyptian pounds in 1985. Thereafter,assets declined and then leveled off in real termsto 2.05 billion Egyptianpoundsin 1990. Ray (1995) shows thatFIBE's market share of total deposits increasedfrom 0.32 percentin 1979 to 7.62 percentin 1984. Marketsharethen remainedroughly constantthrough1990. Because Islamic banks areexcludedfromcompetingwith otherbanksfor public-sectoroperations,theirimpact is betterassessed by examiningtheircontribution to private-sector finance.Ray (1995) finds thatFIBE's marketshareof deposits and financingsin the privatesector follows the same pattern rapidinitial growthin marketshareto 1983 followed by a leveling off. We believe thatEgypt's high rates of inflationhave led to large nominal growthbut little real growthfor FIBE. Several other points emerge about FIBE from Kazarian's(1993) analysis. First, FIBE has kept on average35 percentof its assets overseas.Second, of the assets kept domestically, 46 percent on average have been kept at the CentralBank of Egypt. Takentogether,65 percent of FIBE's total assets are not used for financing of domestic projects.On both the deposits at the centralbank andthe assets held overseas, FIBE is earning a marketrate of returnon relatively safe investments.Third, when looking at the stock of assets deployed for domestic financingof investmentprojects, over 90 percentof the financingis markupfinancing.FIBE seems to maintaina policy of restrictingPLS financingto 3 percentof total domestically held assets. FIBE does not appearto be growingrapidly,and it is not investingmost of its assets in entrepreneurial projects.Even when FIBE does provide financingto such projects,the financingis primarilymarkupfinancingand not PLS financing. The conclusion thatthe use of markupfinancingdominatesthe use of PLS financing is supported by datafrom otherIslamicbanks.In Figures 1 and2 we plot the percentage of new financing that is markupfinancing and PLS financing for several large Islamic banks. For expositionalease, we have omittedthe category of flows to other financing.In additionto the flow of new financingfor FIBE, we also plot data for the JordanIslamic Bank (JIB) and Bank Islam Malaysia(BIM). All threeof these bankscompete with conventionalcommercialbanksin theirrespectivecountries.We also plot data for the last decade for Iranwhose banking system is entirely Islamic. Except for a few years of net repaymentsfor FIBE (which coincided with severe re-

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions

Flows of MarkupFinancing

25%--/ \


1g1 -2s% -50% -75%

1982 1983 r984 198; 1986 1t87 1988\1989| 1990 1991 1992 1993 t934 | |



| \ |

=BIM , X Iran




-t 25% -150% -175% FIG.l. Flows of MarkupFinancing.FIBE=Faisal Islamic Bank of Egypt, JIB=JordanIslamic Bank, BIM=Bank Islam Malaysia, Iran=operationsof the entire Iranianbankingsystem. Flows of financing are calculated by computing the percentage of total new financing allocated to markupfinancing:(Markup(t)-Markup(t-1 ))I(Financing(t)-Financing(t-1)). The variousmarkupinstrumentsused were aggregatedto get the percentageof markupfinancing. Negative percentagesindicatenet repaymentto FIBE: (Financing(t)-Financing(t-1 ) < O. Percentagesgreater than lO0 percent (or less than-lO0 percent) indicate net repayment(or net financing in another category of financing, such as PLS. For example, PLS(t)-PLS(t-1) < 0 and Markup(t)-Markup(t-1) > Financing(t)-Financing(t-1). Sources:Authors'calculationsfromAhmad(1987), Kazarian(1993), Pourian(1995), Ray (1995), Saffari (1995), Shallah (l990), Bank NegaraMalaysia,Annual Reports, 1983-1994. l


This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions

40% 30%-

Flows of PLS Financing






\ \

/ /



$, X

BIM -Iran

1g1 -5%

\ iL*ff4/b\s
1982 1983 1984 1985 1987 1988 1990 1991 1992 1993 194


FIG.2. Flows of PLS Financing. FIBE=Faisal Islamic Bank of Egypt, JIB=Jordan Islamic Bank. BIM=Bank Islam Malaysia,Iran=operationsof the entire Iranianbankingsystem. Flows of financingare calculatedby computingthe percentageof total new financingallocatedto PLS financing:(PLS(t)-PLS(t-1 ))I(Financing(t)-Financing(t-1 )) . used were aggregatedto get the percentageof PLS financing. The variousPLS instruments Negative percentagesindicatenet inflows to FIBE and BIM: (Financing(t)-Financing(t-1) < O. Data are missing for 1986 and 1987 for JIB. Sources: Authors' calculations from Ahmad (1987), Kazarian(1993), Pourian (1995), Ray (1995), Saffari(1995), Shallah (1990), Bank NegaraMalaysia,Annual Reports, 1983-1994.


This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions



cessions in Egypt), markupfinancinghas consistently comprisedmore than 50 percent of flows of new financingfor all of these banks.For BIM, markupfinancinghas averaged95.3 percentof new financingover the period 1983-1994. Except for Iran, flows of PLS financinghave consistentlybeen below 10 percent(we are missing observationsfor JIB in 1986 and 1987). Only in Iranis there a significantPLS component to new flows of financing.But even in Iran,the majorityof financingis based on the markupprinciple.Kuran(1995a) and Nienhaus (1994) also reportthat the composition of markuptechniques for other Islamic banks in mixed financialsectors is consistently above 90 percent. PanelA of Table 1 presentsmore recent evidence from a large cross-sectionof Islamic banks in 1994 and 1995. Roughly 50 percent of the stock of financings is based on the markupprinciple(eithermurabaha or ijara). The use of markupinstruments seems to be growing given the datain Figure 1. PanelA also shows that there is some skewness in the distributionof financing between large and small Islamic banks.The percentageof averagedollars in murabaha financingsis higher than the averagepercentage,suggesting that largerbanks use murabaha financingmore frequently than do smallerbanks. This is consistent with the evidence in Figures 1 and 2. The banks in Figures 1 and 2 are some of the largest Islamic banks in the world and these banks use more markupfinancingthan is suggested by Panel A. Table 1 presents data on a representativecross-section of Islamic banks, including many small ones. In addition,Figures 1 and 2 representflows of new financing.Table 1 representsstocks of existing financing.We conclude that Islamic banks use PLS instrumentsmuch less frequentlythanmarkupinstruments. There is no evidence that Islamic banks are providing significant amounts of long-term capital to entrepreneurs.Metwally (1992) presents cross-sectional evidence from a surveyin 1990 of twenty-two Islamic banks and investmentbanks operating in thirteen countries. Using his data, we calculate that, on average, 56.7 percent of financingsby nominal value were for maturitieslasting less than a year. Medium-term(one to two years) and long-term (two to five years) financings averaged 0.7 percent and 1.9 percent, respectively. Note that financings do not extend past five years. Western banks would consider financings of this maturity to be medium-term.Islamic banks kept, on average, 20.6 percent of their assets as deposits with otherbanks and/orcentralbanks. Real estate investmentswere 0.66 percent of assets and Qard Hassanor social lending was 0.63 percent.This last point is noteworthy. Social lending does not seem to happen.Even in the Islamic Republicof Iran, where the state requiresbanks to allocate specific amounts of QardHassan loans, Pourian(1995, p. 92) notes a steady decline in their share of total financing from 10.5 percentin 1984 to 4.6 percentin 1993. Kazarian(1993) presentsevidence for the two Egyptian Islamic banking institutions contrastedwith other Egyptian commercial banks and investment banks for 1979-1990. He finds that for FIBE, which is comparableto a commercialbank, the ratio of long-term financing to all financings is a third of that for other Egyptian commercialbanks. For the Islamic Bank for International Development (IBID), an Islamic investmentbank, the ratio of longtermfinancingto all financingsis half that

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions

FINANCINGS BY ISLAMIC BANKS, PanelA Financingsby Instrument 1994 Percentages Mean Std Dev Nominal dollars (lOOOs) Mean Std Dev % of mean Mean Percentages Std Dev


TotalAssets Murabaha Ijara Musharaka Mudarabah Other

42.70% 9.12% 11.39% 8.68% 28.10%

36.04% 20.84% 19.37% 19.64% 36.77%

304,809 132,735 16,614 36,939 3S,339 86,182


609,013 273,094 51,281 170,790 159,285 245,443

43.55% 5.45% 12.12% 10.61% 28.27%

45.10% 10.66Go 16.33% 6.50% 21.40%

35.56% 22.98% 26.31% 14.03% 34.04%

Panel B Financingsby Sector Percentages Mean Std Dev Nominal dollars (lOOOs) Mean Std Dev Percentages Std Dev

% of mean


TotalAssets Trading Agriculture Industry Services Real Estate Other


29.29% 11.11% 27.85% 13.94% 5.47% 12.33%

28.70% 23.36% 29.85% 21.24% 13.11% 16.66%

304,809 70,476 49,159 74,303 55,322 18,242 37,306

609,013 174,534 271,111 138,018 281,323 74,978 118,029

23.12% 16.13% 24.38% 18.15% 5.98% 12.24%

23.33% 12.24% 24.34% 11.77% 9.45% 18.83%

26.83% 23.06% 27.93% 17.81% 19.03% 21.86%

NorEs: Numberof banks = 82 for 1994. (2) Numberof banks = 86 for 1995. (3) Excludes banks for which thereis missing data. (4) Excludes Bank Te Authors'calculationsfrom International Association of Islamic Banks (1994, 1995).

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions



of other Egyptian investment banks. This fact is connected to the dominance of markupfinancing as markupinstrumentstend to be short term in natureeither by regulation(as in the case of Iran and Pakistan)or by common practice in other Islamic banks. PLS financings are typically longer in duration;their low composition is underlinedby the small shareof medium-term to long-termfinancings.In the case of Egypt, conventional banks have outperformed Islamic banks in providing long-termfinancing. The evidence is mixed on whetherIslamic banksinvest in entrepreneurial projects in sectors that are typically viewed as growthoriented.Agricultureand industryare presumablysectors in which entrepreneurial projectswould have the greatestimplications for growth. Panel B of Table 1 shows that financingis balanced evenly between agriculture/industry and trade/services. Skewness of the data suggest that large banks provided more financing to agricultureand industrythan to trade and services in 1995. Financingto real estate also seems to have grown in 1995, although real estate financingis still relatively small. However, other evidence suggests that Islamic banks are not extending much financing to agricultureand industry.Kazarian(1993) finds that commercialbanks in Egypt extended37 percentof theirfinancingto industryand agriculture in 1979-90, while FIBE allocated only 10 percent.IBID allocated 11 percentto agricultureand industrycomparedto 17 percentfor Egypt's other investmentbanks. Saffari(1995) notes that the CentralBank of Iran imposed targets for financing by Iran's highly regulatedIslamic bankingsystem to variouseconomic sectors. From 1991 to 1993, realized financingsto tradeand services were more than double the targetedlevels while those to industry,construction,and agriculture were significantlyless thanthe targets. Table 2 provides summarystatistics for Bank Islam Malaysia (BIM) and Jordan Islamic Bank (JIB). BIM extended on average 9.7 percent of its new financing to manufacturing for the period 1983-1994 comparedto 34.1 percent for other commercial banks. However, the large standarddeviations do not make this difference statisticallysignificant.In Jordan,commercialbanks allocated significantlymore financing to agriculture than did JIB but JIB outperformed commercialbanks in allocating funds to industry.However,Shallah (1990) claims that most of this financing to industryis importfinancefor equipmentand supplies ratherthanprojectfinance. There is some evidence that Islamic banks requirecollateralfor their financings, apparentlyviolating a widely accepted legal position forbiddingany form of collateral in lending transactions.Kazarian(1993) reportsin his study of Egypt thatFIBE and IBID secured on average96 percent and 90 percentof their financings,respectively, comparedto 60 percentfor commercialbanks and 78 percentfor investment banks. Thus the great majority of the financing transactionsby Egyptian Islamic banks are secured by some sort of financialguarantee.However, such a conclusion must be treatedwith caution, as ownershipundermarkupfinancingresides with the bank until all payments are made. Ownershipof the assets by the bank may be reportedas collateralizedlending, in which case the high degree of collateralization is merely an artifactof markupfinancing.There is other evidence that Islamic banks

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions




Mean Manufacturing Commerce Housing Loans Other


9.69% 6.31% 24.08% 59.92%


54.17% 15.17% 23.88% 38.11%

34.12% 8.29% 24.58% 33.01%


28.09% 4.91% 44.04% 67.57%

JordanIslamic Bank StdDev

CommercialBanks StdDev

Agriculture Industry Real Estate Transport Trade Other

0.47% 30.36% 16.99% 7.95% 24.85% 19.36%

0.07% 2.90% 0.77% 2.73% 2.33% 2.38%

2.54% 14.53% 25.05% 3.41% 24.93% 29.55%

0.36% 0.85% 1.48% 0.99% 0.91% 2.69%

NarEs: (1) Data for Malaysia are in flows. (2) Data for Jordanare stocks of financing. (3) Otherfor Malaysia includes nonhousingreal estate, agriculture,transport and storage, insuranceand business services, and mining and quarrying. (4) Otherfor Jordanincludes financingto the service sector andprivateindividuals. (5) ForJIB, a significantproportionof credit to industryconsists of tradefinancefor the importof equipmentand stlpplies. Sozerces: Shallah (1990) andAnnual Reportsfrom Bank NegaraMalaysia, Jordan Islalnic Bank, and JordanCentralBank.

are violating the prohibitionon collateral. In addition to the purchasedor leased goods serving as collateral, Islamic banks have been reportedto requlreadditional collateraldependingon the size of the transaction.Kazarian(1993) claims that the size of this additionalcollateral for Egypt's Islamic banks has ranged between 40 and 85 percent of the total funds provided.This seems to suggest that much of the lending done by Islamic banks is overcollateralized. Let us recapitulatethe stylized facts that we want to explain; first, the preference by Islamic banks for debtlike instrumentsover equity-like instruments;second, the preferencefor short-termfinancing when using murabahah contracts;third, some preferencefor investmentsin the retail and tradesectors at the expense of agriculture and industry;fourth,the use of collateralwhen providingfunds. The model we present and analyzein the next threesections explains the stylized facts but does not address why Islamic banks seem to be more averse to equity and require more collateral than conventionalbanks. One possible explanationis that Islamic banks suffer an adverse selection problem they get the entrepreneursthat have been turneddown by conventionalbanks. These entrepreneurs are worse "types" they are more likely to divertfunds and more likely to have high-cost investmentprojects. Conventionalbanks may be more attractiveto all types of entrepreneurs simply because they impose fewer nonpecuniarycosts such as religious restrictionson entrepreneurs.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions



In this section, we presenta model of investmentand capitalstructure based on incomplete contracts.This model is in some respectssimilarto the models of Hartand Moore (1998) and Bolton and Scharfstein(1990). One theoreticalinnovationof this paperis that we allow for outside equity contractsin additionto debt contracts. At time t = 0, an entrepreneur has an investmentproject with randomuncorrelated cash flows xl andx2 generatedin periods 1 and 2. Let xt [0,xH] with cumulative distributionfunction F(xt) and continuousprobabilitydensity functiont(xt). We make the standard monotonehazardrate assumption:the hazardrate (1-F(x))lf(x) is non-increasingin x. The risk-neutralentrepreneur has no wealth so she needs to raise I dollarsto financethe project.In this simple model we assume thereis no discounting. In principle, the project could be financed throughany type of financial contract. For example, the project could be financed via an equity investment (profit-sharing) such as a musharakaor mudarabahcontract.It could be financed througha debtlike contract(markup)such as a murabahaor ijara contract.In this context, it may not be clear what debt means given that there is no discounting.We define debt as fixed paymentsdue in either period. These fixed paymentscan be interpretedas coupon payments and principalrepayments.However, the crucial element of debt is not the payment of interest, but the fixed natureof the payments. Thus, we do not dwell on the explicit representation of interestin this model. We assume the entrepreneur can divertcash flows from the projecteitherthrough the consumptionof perquisitesor throughwasteful spending.As cash flows accrue initiallyto the entrepreneur, she can always divertthe proceeds andreportlower cash flows to her investor.The investoris not fooled, but because cash flows are not verifiable the investorhas no recourse.The entrepreneur generatesutility from diverting cash flows but not dollar-for-dollar. The consumption of perquisites is constrained the entrepreneur cannotconsume exactly the bundle she wants. Diversion is not outright theft, but instead involves spending on negative NPV investments which benefit the entrepreneur alone (see Jensen and Meckling 1976). The entrepreneur's utility from divertingan amounty is u(y) = cy where 0 ' c ' 1. The entrepreneurcan divertcash flows in both periods. Because cash flows are assumedto be unverifiableto a court, there is no way to force the entrepreneur to disgorge these proceeds.The entrepreneur's utility from a directcash paymentof y from the firmis v(y) = y. Such paymentslead to unconstrained spendingby the entrepreneur. Therefore the entrepreneur would prefer a direct cash payment of y from the firm to divertingan amounty from the firm'scash flows. The cost of the investmentproject is I, where I is distributedcontinuouslyon [0, (Exl) + E(x2)]with probability densityfunctiong(I). The cash flows from the project are independentof the cost of the project,so the cost gives no additionalinformation aboutthe expected value of the project.The realizedcost of the project,I, is fully observableto all at date 0. Note that the entrepreneur's project is always positive expected NPV. From a social welfare perspective, society would be better off if the

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


banks offered contractsto every entrepreneur as they all have positive NPV projects, even if the banks lose money. In principle, the entrepreneur may defaulton a debt payment.If the entrepreneur defaults, the bank may choose to liquidatethe firm's assets. If the entrepreneur defaults in the second period, the second periodliquidationvalue of the firmis L2.L2is normalizedto zero because we assume thatthe firmends aftertwo periods.The liquidationvalue representsthe futurevalue of the investmentproject,which after the second period is zero. If the entrepreneur defaultsin the firstperiod, we assume that the bankcan seize the projectand liquidateit for L1.No second periodcash flows are then realized. We assume that E(x2)-L1, the liquidation value of the firm is boundedabove by the expected second-periodcash flow of the firm.Thereforeliquidation is socially less efficient than continuation.This capturesthe idea that even thoughthe entrepreneur is the most efficient user of the firm's assets, the bargaining problembetween the bankand the entrepreneur may lead to liquidationin the case of default.This assumptionimplies that bankruptcyis costly and introducesa cost of debt financing. Our last assumption is that the banks have all of the bargaining power, or, equivalently,the banks are monopoly banks. We discuss this assumption in greaterdetail in section 5.

In this section we considergeneralcontractsthatbankscan offer to entrepreneurs. Such contractscan be PLS contracts(musharaka and mudarabah) or markupcontracts(naurabaha and ijara)in which thereare specified fixed repaymentsor combinations of both. We characterize when PLS contracts will dominate the use of markupcontractsand when they will be dominated.Markupcontractshave the critical featurewe ascribe to debt: default on a paymenttriggersa shift in control over the asset from the entrepreneur to the bank. In a markup contract this is transparent the bank retains ownership of the asset until all of the payments are made. If a paymentis not made, then the bank liquidatesthe asset. Fromnow on, we referto PLS contractsas equity and markupcontractsas debt. A generalcontractcan consist of both debt and equity. It will be given by {oc,D1, D2, I), where ocis the amountof equity (the profit share) retainedby the entrepreneur, 1-oc is the amount of equity given to the bank, D1 is the face value of firstperiod debt, D2 is the face value of second-perioddebt, and I is the amountof funds providedby the bank to the entrepreneur. In principal,the marketvalue of debt may be lower than the face value of debt because the entrepreneur may default. Nothing in our analysis prevents one investor from holding debt and anotherinvestor from holding equity.Thus the contractneed not be concentrated in a single bank, although that is the way we will refer to the contract.Islamic banks can offer these general contacts. These general contractsalso nest the special cases of all-equity contracts, {oc,I), and pure debt contracts,{D1, D2, I). The bank will maximize its expected return,E(KB),over the set of possible con-

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


tractssubject to inducing the entrepreneur to fulfill her contractualobligations (that is, not to default or divertcash flows). The bank'sproblemis: max JwoLlf(xl)dxl + Jwx,(ocD1+ (1 oc)xl + .(x2 (oCD2 (

) 2)

The first component f E(KB)is ,(0 Llf(x1)dxl. The bank liquidatesthe project for L1if the entrepreneur defaultson first-perioddebt, D1. The entrepreneur will default if the realizationof the first-periodcash flow x1 is less thanx1', wherex1' will be determinedby the incentive compatibilityand limited liability conditions given below. If x1 -x1', the entrepreneur does not default and the bank receives D1 + (1-ot)(x1-D1) = oeDl + (1-ot)x1, the first-period debt payment and the equity shareof the excess firstperiod cash flow over the face value of debt. In addition,the bank receives its expected second-periodreturn.If the entrepreneur defaults in the second period, the bank receives L2.Because we have normalizedLo = O,this component drops out of the bank's expected return.The entrepreneur does not default if x2-x2', where x2' will be determinedby the conditions given below. If the entrepreneurdoes not default,the bankreceives D2 + (1-oc)(x2-D2) = ocD2+ (1-oc)x2, the second-perioddebt payment and the equity share of the excess second-period cash flow over the face value of debt. The bank'sproblemis subjectto incentive compatibilityand limited liability conditions in both periods: v(ocx)-u(x)
x2 xH

(la) ( lb)

oc(x1-D1) + JsO cx2f(x2)dx2 + ,(x2, oc(x2-D2)f(x2)dx2-cxl

X1-D1 '

oc(x2-D2) ' CX2 x2-D2 '

(ld) (le)

Condition(la) says thatthe entrepreneur's utility from paying out the cash proceeds to the equityholdersmust be greaterthan the entrepreneur's utility from diversion. Another way to write this is that oc ' c. The entrepreneur must be given at least a fractionc of the equity, otherwise she will divertcash flows and get a higher utility. Divertingcash flows from equityholders does not constitutedefaultandthereforecannot be prevented.Condition(lb) is the first-period incentivecompatibilitycondition. If the entrepreneur diverts the first period cash-flow, she gets cx1. If she makes the first-period debt paymentD1, the entrepreneur gets both her equiHty share,oc(xl-D1), and her expected second-periodreturn:,(0 CX2X(X2)dX2+ ix2 (x2-D2)f(X2)dX2The face value of first-period debt and the bank'sequity allocationcannotbe so high

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


that the entrepreneur divertsall of the funds, taking into account the entrepreneur's second-period expected return. Condition (lc) is the first-period limited liability condition. This condition can also be interpreted as a zero-wealthassumption.If the first-periodcash flow is not sufficientlyhigh to meet the debt repayment,the entrepreneurcannotbe held liable for the shortfall.She simply loses controlof the firmto the bank. Further,the entrepreneur cannot make up the shortfallbecause she has no wealth with which to do so. In addition,the entrepreneur cannot borrowagainst the second-periodreturnin orderto make up the shortfall,that is, the entrepreneur cannot renegotiatethe contract.In the analysis thatfollows, we show that second-period debt is not optimal for the bank, so that a bank would never be willing to allow the entrepreneur to borrow against the second-periodreturn.Condition (ld) is the second-periodincentive compatibilitycondition. The equity allocation to the bank and the face value of debt cannot be so high that it is optimal for the entrepreneur to divert funds conditional on the realizationof the second-periodcash flow. Condition (le) is the second-periodlimited liability condition and its interpretation parallels that of the first-periodlimited liability condition. In orderto simplify the problemand derive the optimalcontracts,we work backwardfrom the second period.We firstnote thatwe can restrictattentionin the second period to equity contractsbecause, in the second period, an equity contractdominates any type of debt contract.To see this, let ot = c and D2 = O.Then the incentive compatibilitycondition( 1d) is satisfiedwith equalityfor any realizationof x2 and the limited liability condition (le) is clearly met for any realizationof x2. This contract generatessecond-periodexpected returnsof (1-c)E(x2) for the bank. Now consider any alternative contractwith oc'E (c, 1] andD2' > O.Then thereexists a cutoffx2' = ot'D2'/(oc'-c) from condition (ld) where for x2 ' x2', the entrepreneur does not default and divertthe cash flow. For x2 > x2', the incentive compatibilitycondition is met but the entrepreneur earns excess returns(thatis, this contractis ex post suboptimal for the bank).At x2 = x2', the bank is indifferentbetween the (x = c and D2 Oand the ot' E (c, 1] andD2' > Ocontracts.Forx2 < x2', the incentive compatibility condition is not met so the entrepreneur defaults and divertsthe second-periodcash flow and the bank receives nothing.As the incentive compatibilitycondition is met with slack for x2 > x2', and is not met for x2 < x2', second-perioddebt contractsare dominated.Further, any contractwith t2' < c and D2' ' Owill violate (la) and (ld) for any realization of x2 and so is clearly suboptimal.Therefore, optimal secondperiod contractswill be ot* = c, D2 = O.This implies thatx2' = O. Now considerthe firstperiod.We firstverify thatot* = c is the optimalequity portion of the contractin the first-periodas well and solve for the optimal first-period debt level. Note thatD2 - Oimplies thatthe first-period incentivecompatibilitycondition (lb) reduces to ot(xl-D1 +E(x2)) ' cxl. For ot = c and D1 ' E(x2),condition ( lb) is satisfiedfor all realizationsof xl . Furthermore, for ot = c andD1 = E(x2),condition (lb) will be met with equalityfor all realizationsof x1. Note, however,thatthe limited liability condition need not be met for all realizationsof x1. Now consider any alternativecontractwith ot' E (c, 1] and D1' > E(x2).Then there exists a cutoff x1' = ot'(Dl'-E(x2))/(ot'-c) from condition(lb). Forxl > xl', condition(lb) is met

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


but the entrepreneur earnsexcess returns. At x1 = x1', the bankis indifferentbetween the ot = c and D1 ' E(x2)and the ot' E (c, 1] andD1' > E(x2)contracts.For E(x2) ' x1 < x1', condition (lb) is not met so the entrepreneur defaultsand the bankreceives L1.In this region, underthe oe= c and D1 = E(x2)contract,the bank earns D1 + ( 1-C)(X1-D1+E(X2)) ' E(X2)+ ( 1-c)xl > L1 (2)

Forx1 < E(x2),the entrepreneur defaultsunderboth the D1 = E(x2)andD1' > E(x2) contractsand the bank receives L1.Therefore,any contractwith ot' E (c, 1] and D1' > E(x2)is dominatedby the ot = c and D1 = E(x2)contract.This implies that in the firstperiod,optimalcontractswill be characterized by oe*= c andD1 ' E(x2).Conditions (la) and (lb) are satisfiedeverywhere.Therefore,the only conditionthatmay not be satisfiedis the limited liability condition (lc) which in turnimplies thatx1' D1. Forx1 < x1', the entrepreneur defaults,while for x1 ' x1' the entrepreneur repays D1 to the bankbecause the limited liability conditionis satisfied. Given the preceding analysis and ot* = c and D2 = O, the bank's optimization problemreduces to max JO Llf (xl)dx+ |D ( Dl + (1-c)(x1-Dl + E(x2 )))t(xl )dxl (3)

The monotone hazardrate ensures quasiconcavityof the bank's objective function. The first-order conditionis
AE(X ) B = (F-D1-(1-C)E(x2

))f (Dl) + C(1-F(D1 )),


and the second-orderconditionis satisfiedas well. We first characterizewhen all-equity contractswill be optimal (that is D1 = O). Evaluating AE(XB)I8D1 at D1 = O yields a necessary and sufficient condition for all-equitycontractsto be optimal: Ll-(1-c)E(x2) + f(O) ' O. (S)

There exists a cutoff level of moral hazardc* E (O, 1) such that for c ' c*, pure equity contracts will be optimal and the bank will only offer pure equity contracts. When c ' c*, the rate of diversionis low. The cutoff c* is given by
* _
c -

E(x2 )-A

E(x2)+ f(lO)


This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


Conversely,if c > c*, then D1 = D1 is optimal. From equation (4), D1 > O is given by Ll-Dl-(1-c)E(x2 )+ ( ( ) l )) = 0 . (7)

For c > c* (the rate of diversionis high), debt and equity contractswill be optimal relativeto pureequity contractsand the bank will offer only the combinationof debt and equity contracts. Finally,in orderto see when D1 = E(x2),we evaluateaE(xB)laDl at D1-E(x2) to get Lq-E(X2 )-(1-c)E(x2

+ C(1 F(E(X2))),


This conditionyields a cutoff for D1 = E(x2)to be optimal:

2E(X2 )-L1



)) f (E(X2

debt conFor c ' c**,the level of moralhazardis so high thata maximalfirst-period tact, D1 = E(x2),is optimal.There are two points to note aboutthis cutoff. First, the cutoff for maximaldebt is higher thanthe cutoff for any debt, c**> c*. This follows from the monotonehazardrate assumption.Second, it may be the case that c**> 1. In other words, it is possible that maximal debt may not be optimal for any level of
. . .


We can write optimalfirstperiod debt contractsas pO D1 =tD1 for c* ' c forc** >c>c* (10)

tE(x2 ) for c ' c** assuming c** < 1. Figure 3 depicts the optimal first-perioddebt contractas a function of the rate of diversionc. Using the implicit functiontheoremand the monotone hazardratepropertyyields the following comparativestatics:
aD* aD* aD* (11)

The face value of debt, D1, is increasingin the rate of diversionc while the amount

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions





D1 (c)



Debt Contractsas a Functionof the Level of Diversion FIG.3. OptimalFirst-Period

of outside equity finance 1-oe* = 1-c is clearly decreasingin the rate of diversion. This is the sense in which the optimalityof debt as a financialinstrumentdepends on the level of moral hazard in an economy. The face value of debt is increasingin the liquidationvalue of the firm as a higher liquidationvalue decreases the riskiness of debt due to default. The face value of debt is decreasingin the expected value of the second period cash flow: as the second-periodcash flow becomes more valuable, the cost of default increases because the second-periodcash flow is not realized. We have identifiedthree regions of the parameterspace for first-perioddebt contracts.However,we are primarilyinterestedin two regions: c ' c* so that D1 = O, and c > c* so thatD1 > O.The firstcorrespondsto the use of all-equitycontracts.We define IE = (1-c)(E(xl) + E(x2)) (12)

as the highest-cost project that can be financed under all-equity finance or profit sharing.The bank's profit share must yield (in expectation) at least the amount of funds providedby the bank.The maximumamountof equity thatcan be given to the bank is 1-c, for then (la) will bind. Because the bank has the bargainingpower, with costs of investmentI E [O,IE), the optimal when c ' c* and for entrepreneurs expected returnis contract is ot* = c, D1 = O, and. D2 = O. The entrepreneur's cE(xl) + cE(x2).If I > IE, then the bank's returnis insufficientrelativeto the investment to induce the bank to provide financing using equity. Equity contractswork well when the cost of the project is low or when the rate of diversionis low. When the rate of diversion is high, the agency problem implies that the project cannot be fundedwith pure equity. In the second region, the optimal debt and equity contractsare ot*= c, D1 > O, have an expected returnof and D2 = O-Entrepreneurs

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions







= cE(xl ) + c(l-F(D1 ))(E(x2 )-D1 ) . We define

IDE = F(D1' )L1 + (1-F(D1' ))(cDl + (1-c)E(x2 ))

( 13)






as the highest-cost project that can be financedunder the combinationof debt and equity.An entrepreneur with a projectthatcosts I ' IDE will have her projectfunded with debt and equity.For I > IDE,the projectcannotbe funded.The projectis more likely to be funded the higher the liquidationvalue of the assets, the higher the expected value of the second-periodcash flow, and the lower the rate of diversion.One additionalimplicationfollows from the optimalcontracts.As the rateof diversionincreases, a largershareof the projectwill be fundedwith debt and less will be funded with equity.As moralhazardincreases,debt becomes the predominant instrumentof finance.Note, however,thatdebt has costs as well as benefits. Here the cost of debt comes from the possibility of default.

In this section, we comparedebt and equity contractsto pure equity contractsand we draw social welfare implications. For c > c*, the level of moral hazard in the economy is high enough so thatdebt andequity contractsexpandthe region in which entrepreneurs can get financingrelativeto pureequity.The bankprefersto offer debt and equity contractsas these generatea higherreturnto the bank.The entrepreneurs, however, prefer pure equity because it allows them to retain more of the proceeds conditionalon the bankfinancingthe projectundereitherpureequity or debt and equity. Because the bank has the bargainingpower, it will not offer pure equity contracts.If the level of moral hazardis low, c ' c*, then pure equity contractswill dominate the use of debt and equity. This point is of independentinterest because agency models of capital structuresuch as Bolton and Scharfstein(1990) and Hart and Moore (1998) do not have the use of outside equity as an optimal financial instrument. We next examine the social welfare implications of the different types of contracts.Gross social welfare per projectfunded underpure equity contractsis GSWE= (1-c) (E(xl ) + E(X2)) + CE(xl) + CE(x2 ) = E(Xl) + E(X2) (15)

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


Gross social welfare per projectfundedunderdebt and equity contractsis GSWDE= F(D1 )L1+ (1-F(D1 ))(cDl + (1-c)E(x2 ))
+ (1-c)|D* Xlf(Xl )dXl + CE(X1 )

+ c(l-F(D1 ))(E(x2)-D1 ) < E(xl ) + E(x2) .


However,the social welfare implicationsare more ambiguousthan this comparison suggests because more projects may be funded under the combinationof debt and equity thanunderpureequity.To see this, note thatexpectednet social welfare (prior to time O)underpure equity is JO (E(x1) + E(x2 )-I)g(I)dI . (17)

Expectednet social welfare (priorto time O)underdebt and equity is

lo (F(D1 )A + (1-F(D1 ))E(X2 ) + (1-C)|De Xlf(Xl )dXl

+ cE(xl )-I )g(I )dI

( 18)

In orderto determinewhetherpure equity or risky debt and equity deliver higher social welfare [whether(17) is greaterthan (18)] we considerthreecases. In the first case, c ' c* (the rateof diversionis low) which means thatthe bank funds all investandbanksprementprojectswith costs up to IE with pureequity.Both entrepreneurs fer this outcome andequity contractsareoptimal.In this region, a ban on debt has no force as debt contractsare not used. Figure 4 depicts the second and thirdcases. In Figure 4, c > c* (the rate of diversion is high) which implies that IDE > IE. In this figure, debt and equity contracts fund more projects than do pure equity contracts.Banks will choose to offer only Net social welfare is given by regions debt and equity contractsto entrepreneurs. A+C. The deadweight loss of debt is given by region B, which measures the loss due to the possibility of default and liquidationfrom risky debt. The second case is thatregion C is smallerthan region B. Social welfare is improvedby a ban on debt. A ban on debt resultsin projectsin the region C not being funded.However,projects with costs of investmentup to IE can still be fundedwith equity.Because C < B, the value of the projectsforegone is -lessthanthe value lost because debt is costly, which leads to the social welfare gain by a ban on debt. The thirdcase is thatregionC is largerthanregionB. Not only does the combination of debt and equity fund more projectsthandoes pureequity,but the additionalincrement to social welfarefrom more projects(C) outweighsthe deadweightloss of risky debt(B). Net social welfareis A + C. In this case, a ban on debtreducessocial welfare.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


Gross Sociai Welfare





Cost of Investment Project (I)

FIG. 4. A Social WelfareComparisonof Equity versusDebt and Equity

There is an economic rationale for a ban on debt. Default and liquidation are costly, and if these costs are sufficientlyhigh, then a ban on debt can be social welfare improving.Thoughwe have not modeled otherpotentialcosts of debt, this conclusion generalizes to other costs as well. Costs of debt associated with adverse selection, or asset substitution,or direct costs due to financial distress and bankruptcywill affect the social welfare calculationof a ban on debt in the same way that costs of liquidationdid above. While therecan be an economic rationalefor the ban on debt, we should also note that banning any form of economic activity generally requiressome degree of enforcement. It is straightforward to see that enforcement costs will impose some deadweightlosses on our social welfare calculationsfor a ban on debt. Clearly,Islamic banks currentlyevade a "ban"on debt, and historicallythis has been the case as well (Kuran1993). Enforcinga ban on debt may simply be too costly for Islamic governments,even in those cases where a ban would be social welfare improving. The results in the second and third cases depend on the assumptionthat banks have the bargainingpower in the relationshipbetween the bankand the entrepreneur. This assumptionseems reasonablegiven that there are relativelyfew banks in most developing countries (including Muslim ones) and that most transactionsare initiated by entrepreneurs competing for funds for projects.Many proponentsof Islamic banking cite the "unjustdistributionof wealth which continued for decades during the interestregime"(International Association of Islamic Banks (1995, pp. 3-4)) as an economic reason to ban debt. One way to interpretthis critique is that banning debt is a way to shift bargainingpower from banks to entrepreneurs. Conditionalon being funded, entrepreneurs preferequity contractsbecause they generatehigher returnsfor the entrepreneurs. Therefore,a ban on debt can lead to a wealth redistribution. As noted above, this wealth redistribution comes at the cost of some projects not being funded.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


If Banning debt is not the most efficientway to achieve this wealth redistribution. we allow for competition between banks, then banks would lose bargainingpower In this case, competitivebanksmight offer pureequity conrelativeto entrepreneurs. Figure 4 suggests that competitive banks would offer pure tracts to entrepreneurs. equity contractsfor projectswith costs of investmentup to IE andthe combinationof risky debt and equity contractsto projectswith investmentcosts between IE andIDE. Low-cost projectswould be financedwith pureequity and high-cost projectswould be financedwith risky debt and equity.Anotherway to think aboutthis is as a form with higher-costprojectsget worse fiin which entrepreneurs of price discrimination with with lower-cost projects. Entrepreneurs nancing terms than do entrepreneurs low-cost projects would get to keep more of the cash flows from their projects, of wealth. Such an outcome would be unambigutherebyimprovingthe distribution ously social welfare enhancingbecause it would reduce the numberof defaults.Net social welfare would be A + B + C. Several points emerge from this analysis. First, for c > c*, debt and equity contractsexpand the set of projectsthat can be fundedrelativeto pure equity contracts. There is an efficiency loss because for low realizationsof cash flows, entrepreneurs with debt contracts will default and firms will be liquidated, resulting in a deadweight loss to society. Nevertheless, debt and equity contractscan improve social welfare by allowing some projects to be undertakenthat would otherwise not be. This resultis drivenby the assumptionthatthe level of moralhazardin the economy is high. We believe this assumption is realistic when thinking about developing countries. Second, as a corollaryof the previouspoint, for c > c*, debt and equity contracts are more profitable for banks than are pure equity contracts. Conversely, if both would prefer then the entrepreneur types of contractwere offeredto an entrepreneur, the pure equity corltractto the debt and equity contract.Monopoly and oligopoly banks prefer to offer debt and equity contracts.In the presence of competition, we should see pure equity contractsbeing offered. As more banks enter Islamic countries, theremay be a shift in the compositionof financetowardequity. Third, as the agency problem becomes more severe (increasing rates of diversion), the fractionof debt will rise relative to the fractionof equity in the composition of finance.This suggests thatthe high quantityof mark-upcontractsoffered by Islamic banks is a rational choice given the environment they face, one of high moralhazard.A shift towardequity can occur if the level of moralhazardin Islamic countries decreases. of debt and equity financing,most financing Fourth,when looking at the structure will be skewed towardlow-cost projects.High-cost projects(those thatrequiremore capital)are unlikelyto get funded.This is consistentwith some bias on the partof Islamic banks to lend to companies engaged in trade and commerce as opposed to agriculture,industry,or real estate. The latter are potentiallymore capital intensive than the former.Even though the assets are more intangiblein the case of tradeand commerce,projectsare more likely to be funded given the higher potentialreturns. Fifth, debt contractswill be short term in natureand equity contractswill be long

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


term. We view this as consistent with the fact that Islamic banks seem to lend short term primarily.Islamic banks also requirea high degree of collateralfor their loans. This may be because Islamic banksface a very high level of moralhazard.This may, however, simply be an artifactof the fact that the bank owns the asset in a markup contractuntil all of the paymentsare made, and so the loan is reportedon the bank's balance sheet as collateralized.

We study the set of optimal financial instrumentsfor banks (specifically Islamic banks)operatingin environmentscharacterized by agency problemsand incomplete contracts.We show that outside profit sharingor equity contractscan be optimal financial instrumentseven when contractsare incomplete with respect to cash flows. However, the optimality and use of equity contractswill decrease as the level of agency problems increases within an economy, and debt contractswill become the dominantform of finance. We drawseveralimplicationsfrom our analysis for Islamic banking.First, the observationthat,for reasonablelevels of moralhazard,debt and equity contractsdominate pure equity contractsfrom a bank's perspectiveaccordswell with the fact that Islamicbankshave chosen markupcontractsas theirpreferred mechanismof financing investment.Second, when debt and equity contractscan be used, debt will be shorttermin nature.Therefore,we can rationalizethe preferenceof Islamic banksto lend short term. Third, banks will generally prefer lower-cost investmentprojects. This may explain the prevalenceof markupcontractsused to financetradeand commerce:the cost of investmentprojectsmay be lower in these sectors than in agriculture and industryor real estate. Fourth,in orderto obtain financingfor more costly projects,entrepreneurs will need to show that their propensityfor diversion is low. The use of collateralsuggests thatIslamic banks seek out entrepreneurs who are unlikely to be serious moral hazardrisks. Seeking additionalcollateral,althoughcompletely rational,contradictsthe spiritof Islamic banking. Given that Islamic banks structure their lending to be mostly shortterm and quite heavily secured, we conclude that Islamic banks face severe agency problems in their attempts to provide funds to entrepreneurs.Heightened competition among banks might generatemore profit-sharing contractsas borrowersare able to choose among the varioustypes of contractson offer from the differenttypes of banks, but we have yet to see this in the data.Furthermore, it is not clear thatIslamic bankswill provide more competition to conventionalcommercial banks in Muslim countries. Our evidence suggests that Islamic banks are niche providersof capital, and within that niche do not operatemuch differentlyfrom conventionalbanks. These results have implicationsfor the literatureon religious norms and economics. Kuran(1983, l995a, l995b) argues that the principal shortcomingof Islamic economics is thatit does not properlyaccountfor the impactof economic incentives. In short, religious norms are unlikely to change humanbehaviorwhen fundamental

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


economic considerationssuch as wealth maximizationare present.Our results suggest that economic incentives are shaping the structureof Islamic bankingmore so than are religious norms.The key economic restrictionof Islamic Law is the ban on interest. While the religious basis for this restrictionis a concern with income inequality and unequal access to capital, there is an economic case for the ban on the use of interest.The inefficiencies necessary to justify the ban on interestprimarily have to do with incentivesandbargaining power.These inefficienciescan also lead tQ social welfare reductionsfrom a ban on interest.A betterappreciation of these inefficiencies would lead to betterpolicy prescriptionsfor Muslim countriesand perhaps reconcile the expectationsof Islamic banking'sadvocateswith the reality of Islamic bankingtoday, a point also made by Kuran(199Sa). Muslim countriescurrentlydo not seem to be in a situationin which a ban on interestincreasessocial welfare. We conclude from our analysis thatalthoughIslamic banks are or should be based on the profit-and-losssharingprinciple,given the economic environmentsin which they operate, using only this type of financing may not be possible. Moral hazard problemssuggest the need for some sort of debtlike instrument.The use of markup contractsis a rationalresponseto informational problems.Thus we feel the informational environmentwill be a more important determinant of the evolution of banking and growthin Muslim countriesthanwill attemptsto impose financialsystems based Onspec1ncre ilglOUS pr1nc1p ies.
., . . . .


Ahmad,Ausaf. Developmentand Problems of Islamic Banks. Jeddah:Islamic Development Bank, 1987. Babai, Don. Arab Banker'sAssociation of NorthAmericaNewsletter. 12 ( 1995), 1. Bolton, Patrick,and David Scharfstein."ATheoryof PredationBased on Agency Problemsin FinancialContracting." AmericanEconomicReview 80 (March 1990), 93-106. Chapra,M. Umer. Islam and the Economic Challenge. Leicester: The Islamic Foundation, 1992. Ebrahim,MuhammedS., and Akram Safadi. "BehavioralNorms in the Islamic Doctrine of Economics: A Comment." Journal of Economic Behavior and Organization 27 (June 1995), 151-57. Economist, The. "IslamicFinance:Turningthe Prophet'sProfits."The EconomistNewspaper (August 24, 1996), 58-9. Glaeser, EdwardL., and Jose Scheinkman."Neithera Borrowernor a Lender Be: An Economic Analysis of InterestRestrictionsand Usury Laws."Journal of Law and Economics 41 (April 1998),1-36. Hart,Oliver,andJohnMoore. "DefaultandRenegotiation: A Dynamic Model of Debt."QuarterlyJournal of Economics 113 (February1998), 1-41. International Association of Islamic Banks. Directoryof Islamic Banks and Financial Institutions: 1994.Jeddah:IAIB, 1994. . Directoryof Islamic Banksand FiszancialInstitutions:1995. Jeddah:IAIB, 1995. Jensen, Michael, and William Meckling. "Theoryof the Firm:ManagerialBehavior,Agency Costs, and Ownership Structure."Journal of Financial Economics 3 (October 1976), 305-60.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions


Kazarian,Elias G. Islamic versus Traditional Banking:Financial Innovationin Egypt. Boulder:Westview Press, 1993. Khan, Mohsin S., and Abbas Mirakhor.TheoreticalStudies in Islamic Bankingand Finance. Houston:The Institutefor Researchin Islamic Studies, 1987. . "IslamicBanking: Experiences in the Islamic Republic of Iran and Pakistan." Economic Developmentand CulturalChange 38 (January1990), 353-75. Khan, Shahrukh R. Profitand Loss Sharing:An Islamic Experimentin Finance and Banking. Oxford:Oxford UniversityPress, 1987. Kuran,Timur."BehavioralNorms in the Islamic Doctrineof Economics:A Critique." Journal of Economic Behavior and Organization4 (December 1983), 353-79. . "The Economic Impact of Islamic Fundamentalist." In Fundamentalismsand the State: RemakingPolities, Economies, and Militance, edited by MartinMarty and R. Scott Appleby,pp. 302-41. Chicago: Universityof Chicago Press, 1993. . "Islamic Economics and the Islamic Subeconomy."Journal of Economic Perspectives 9 (Fall 1995), 155-73 (a). . "Further Reflections on the Behavioral Norms of Islamic Economics."Journal of EconomicBehavior and Organization27 (June 1995), 159-63 (b). Mauro,Paolo. "Corruption and Growth." QuarterlyJournalof Economics 110 (August 1995), 681-712. Metwally,Samir."TheAggregateBalance Sheet andResults of Transactions andFinancialIndicatorsfor Islamic Banks and FinancialInstitutions." Journalof Islamic Bankingand Finance (1992), 7-61. Nienhaus, Volker."Conceptualand Economic Foundationsof Islamic Banking."In Islamic Banks in a Non-IslamicFramework edited by Gian MariaPiccinelli. Rome: IPO, 1994. Pourian,Heydar."TheExperienceof Iran'sIslamic FinancialSystem and Its Prospectfor Development."In Developmentof Financial Marketsin the Arab Countries,Iran and Turkey. Economic ResearchForumfor the ArabCountries,Iran,and Turkey,1995. Ray, Nicholas. Arab Islamic Bankingand the Renewal of Islamic Law. London:Grahamand Trotman,1995. Saffari, Said. "IslamicBanking in Theory and Practice:The Experience of Iran."Center for Middle EasternStudiesWorkingpaper,Harvard University,1995. Shallah,Ramadan."Jordan: The Experienceof the JordanIslamic Bank."In Islamic Financial Markets,edited by RodneyWilson, pp. 100-28. London:Routledge, 1990. Siddiqi, MuhammadN. Issues in Islamic Banking.Leicester:The Islamic Foundation,1983. WorldBank. Claimingthe Future: Choosing Prosperityin the Middle East and NorthAfrica. Washington: The WorldBank, 1995.

This content downloaded from on Mon, 19 Aug 2013 05:19:25 AM All use subject to JSTOR Terms and Conditions