1、Bank lending behaviour under a liquidity constraintTakashi Hatakeda*Department of Finance, University of Marketing and Distribution Sciences, 3-1,Gakuen-Nishi-machi, Nishi-Ku Kobe, Hyogo, 651-2188, JapanReceived 2 June 1999; received in revised form 20 September 1999; accepted 13 December 1999Abstra
2、ctIn this paper, we propose a new model of bank lending behaviour, in which there are threepossible regimes under asymmetric information. We have found empirical evidence for theexistence of the third regime, in which banks lend under their liquidity constraint. In this regime,both the land price in
3、dex and bank capital have large and positive effects on bank loans. On theother hand, the call rate and economic activity (real GDP) have negative effects. Moreover, we havea new understanding of how financial liberalization and the regulation of bank capital also affectbank lending behaviour. # 200
4、0 Elsevier Science B.V. All rights reserved.JEL classification: E4; E5Keywords: Bank loans; Liquidity constraint; Collateral; Economic activity1. IntroductionThe ModiglianiMiller theorem argues that, in a setting of perfect capital markets,investment decisions do not depend on financial structure. I
5、n their theorem, banks are merefinancial veils. However, a number of papers (e.g. Brainard and Tobin, 1963; Blinder andStiglitz, 1983; Bernanke and Gertler, 1987; Williamson, 1987) have questioned therelevance of this proposition for macroeconomic analysis. The basic premise is that, inthe absence o
6、f intermediary institutions, problems caused by asymmetric informationmake financial markets incomplete. In this situation financial intermediaries, or banks, helpreduce market imperfections and improve the allocation of resources by specializing ingathering information about loan projects. Thus, ba
7、nks have important roles in aggregateeconomic activity.Japan and the World Economy12 (2000) 127141*Tel.: 81-78-796-4966; fax: 81-78-794-1084.E-mail address: takashi_hatakedared.umds.ac.jp (T. Hatakeda)0922-1425/00/$ see front matter # 2000 Elsevier Science B.V. All rights reserved.PII: S 0922-1425(9
8、9)00040-7Recently, a large number of empirical studies in Japan have discussed the above question(Iwabuchi, 1990; Hoshi et al., 1991; Kuroki, 1993; Ogawa et al., 1994; Hosono, 1995;Hatakeda, 1997; Ogawa and Suzuki, 1997; Ogawa and Kitasaka, 1998). Except for the firsttwo papers, these papers support
9、 the hypothesis that banks or bank loans have importantroles in aggregate economic activity. In particular, Ogawa et al. (1994), Ogawa and Suzuki(1997) and Ogawa and Kitasaka (1998) present empirical evidence that land plays an activerole as collateral in loan contracts and amplifies the magnitude o
10、f the business cycle.In this paper, we show how bank loans should be determined under asymmetricinformation and explore the empirical relationship between financial variables andmacroeconomic activity in Japan.Our analysis is based on Bernanke and Gertler (1987), who have shown that the landprice an
11、d bank capital the availability of banks internal funds should be importantdeterminants of bank loans when there are problems of asymmetric information betweenbanks and depositors. They approached the question from a quite different angle from otherrecent works. In particular, they have argued that
12、banks are subject to their liquidityconstraints, which are defined in Section 3, in deciding the quantity of bank loans.They implicitly assume that there are no alternative financing methods to bank loans.Therefore, an expansion of economic activity allows banks to increase bank loans, and theincrea
13、sed bank loans allow firms to invest more actively. In reality, however, firms do havesome alternative financing methods (bonds and equities). If bank loans are perfectsubstitutes to these direct financing methods, firms investment decisions will be inde-pendent of the fluctuations of bank loans. In
14、 this case, therefore, banks are mere financialveils.We extend their model to a new model where direct financing methods exist as well asbank loans. We assume that firms have a project with a risk but do not have funds to invest.On the other hand, banks do not have a project themselves but have fund
15、s to invest. What isimportant is that there exists private information about the realization of projects. In orderfor firms to carry out a project, firms choose to gather funds either from consumers directlyor from a bank indirectly. When firms borrow funds from a bank, firms receive some rents1to g
16、ive up any possible return from the project. Thus, firms have no risk by giving up theresidual return on the project.2On the other hand, when firms gather funds directly, firmsstill have a risk and the residual claim. In this case an extension of economic activity maynot necessarily further increase
17、 bank loans. When a banks liquidity constraint is binding,1Rents represent the banks commitment to the investment project. For example, if a firm faces thecontingency of bankruptcy, the bank will save the firm by extending rescue loans.2Readers may think that our framework should not be consistent w
18、ith the usual description of indirectfinancing whereby the bank gets the interest, the firm gets the profit and they share the risk. Readers, however,will accept validity to our framework by the following reasons. A large number of Japanese firms have the mainbank system, which is, for example, defi
19、ned as a long-term relationship between a firm and a particular bankfrom which the firm obtains its largest share of borrowings. Furthermore, the main bank takes large equitypositions in firms to which she lends and even places executives in top management. In this respect, bank loansare the so-call
20、ed debt with the control of the management. In addition, when these firms face financial distress,the main bank often provides them rescue loans. As far as firms raise funds from the (main) bank, these firmsfaces no risk in the bad situation and make decisions with taking consideration into the bank
21、s profitmaximization. In our paper, for simplicity, we presume that, in indirect finance, the decision of the investment isentirely in the hands of the (main) bank.128 T. Hatakeda / Japan and the World Economy 12 (2000) 127141banks cannot give firms enough rents to have attractions for them. As a re
22、sult, firms willshift to direct financing to get more profits. In our empirical analysis, we find evidence tosupport this view in Japan. In short, we present a model in which the choice between directfinancing and indirect financing is endogenously determined. Furthermore, we also presentempirical e
23、vidence, which Bernanke and Gertler (1987) do not have.The remainder of the paper is organized as follows. In Section 2 we outline some of thekey assumptions in our model. In Section 3 we develop a simple equilibrium model withfirms, banks, and consumers, and characterize the equilibria. In Section
24、4 we report ourestimation results by using aggregate data. We conclude in Section 5 with a brief discussionof the implications of our work.2. Basic assumptionsBernanke and Gertler (1987) focussed on the relation between banks and depositors,showed that banks should have an important role in real all
25、ocation and should not be merefinancial veils.3We will extend their model by introducing the relation between banks andfirms. For simplicity, we consider a static model. In this section we describe someassumptions before developing our model.Consider an economy that is composed of three types of age
26、nts: firms, banks, andconsumers. All agents live for two periods. In the first period, financial contracts among thethree types of agents are signed and investment decisions made. In the second periodinvestment returns are realized and claims settled. All agents are ex ante identical and riskneutral
27、.We will start with firms. We assume that firms do not have their own endowments.4However, firms have stochastic projects to produce outputs from endowments. The rate of(gross) return Ris a random variable that is distributed over the closed interval R1, Rh(Y)with R10. We call R1the collateral value
28、 of a stochastic project.5Weassume that an increase of economic activity, or a positive productivity shock, raises theaverage rate of return on the stochastic projects. Therefore, the expected return depends onthe collateral value and economic activity; ER R(R1,Y) with 0 qER/qR1Rl1 K=Y 1 drf(ii)In t
29、his case, though the banks liquidity constraint is binding, the minimum earning rateis larger than or equal to the market rate rf. The optimal investment Iis now given byIY K1 d Rl;Y;K ; wheredY0;dK0: (3.8)rfRl1 K=Y 1 d(iii)Now, the banks liquidity constraint is binding and the minimum earning rate
30、is less thanor equal to the market rate rf. In this case, the optimal investment Iis given byIrfK1 d Rl;Y;rfrfRl; wheredRl0;dY0;drf0: (3.9)We can differentiate Eqs. (3.7), (3.8) and (3.9) totally and get the following equationsrespectively.T. Hatakeda / Japan and the World Economy 12 (2000) 127141 1
31、31dIa1dY a2dK a3dRl; (3.10)wherea11 d R=Y1 d 20; a211 d0; a3R=Rl1 d 20dIb1dY b2dK b3dRl; where b20 (3.11)dIg1dY g2dK g3dRlg4drf; (3.12)where9:g1R=Y1 d rfRl20; g211 d rfRl0;g3R=Rlrf11 d rfRl20; g41 rfK Rl1 d rfRl20:The sign of the derivative of the banks capital is positive in all regimes. This impli
32、es thatan increase in bank capital increases the incentive for banks to increase bank loans. In ourmodel, bank loans are the most attractive assets for banks since it is assumed that theexpected return on bank loans exceeds the market rate (see footnote 20(a).The signs of the derivative of consumer
33、endowments and collateral value are different inthe third regime from those in the first regime. The reason for these differences betweenregimes is that firms have access to alternatives to bank loans such as bonds and equities.As we will illustrate, increases of consumer endowments or collateral va
34、lue have stronglyopposite effects on bank loans. In either case, the dominant effect determines the sign ofconsumer endowments or collateral value.We will start with an increase of consumer endowments. As a positive effect on bankloans, an increase of consumer endowments, or economic activity, leads
35、 to an increase ofdemand deposits. Banks, consequently, tend to increase bank loans in order to pursue theirprofits. On the other hand, the main cause of an increase of economic activity is to besought in positive productivity shocks. Positive productivity shocks raise the expectedreturn on stochast
36、ic projects. According to Eq. (3.3), firms that finance funds via bank loanstend to claim more payments d from banks. If banks do not accept firms claims forpayments, firms tend to finance directly to pursue more profits. Accordingly, banks mustmake enough payments to keep firms investing in stochas
37、tic projects via indirect finance,so that banks may decrease bank loans for increased payments. In the first regime, the signof the derivative of consumer endowments is dominated by a positive effect rather than anegative effect. In the third regime, however, banks cannot be prevented from decreasin
38、gloans because their liquidity constraint is binding and because the minimum earning rate ofloans is too low. Thus, the sign of consumer endowments is dominated by a negative effectalone.A similar discussion is true of collateral value. When the liquidity constraint is binding, arise in collateral v
39、alue relaxes the liquidity constraint and allows banks to increase their9For the sign of g3, we may safely assume that R=Rlrfis less than one. The reason is as follows. Supposethat R=Rlis large, for example, 0.4, 0.6, or 0.8. In order for R=Rlrfto be larger than one, the short-termrate rfis required
40、 to be at least 2.5, 16.7, and 1.25, respectively. However, these required rates are too high inreality.132 T. Hatakeda / Japan and the World Economy 12 (2000) 127141loans. On the other hand, a rise in collateral value raises the expected return on bank loans.Therefore, firms will claim more payment
41、s d from banks than before. Banks must acceptfirms claims in order to keep firms investing in stochastic projects via indirect finance,sothat they decrease bank loans for increased payments. In the first regime, a negative effectacts alone. In the third regime, on the other hand, a positive effect d
42、ominates a negativeeffect.Finally, the sign of the market rate is solely negative in the third regime. This is because,so far as the minimum earning rate exceeds the market rate, banks decisions areindependent of the market rate.Putting Eqs. (3.10), (3.11) and (3.12) together, we can obtain the foll
43、owing equation:dIy1dY y2dK y3dRly4drf; y20; y40: (3.13)Bank loans depend on economic activity, bank capital, collateral value, and the marketrate. In particular, when the liquidity constraint is binding, an increase of economic activityhas a negative effect on bank loans and a rise in collateral val
44、ue has a positive effect.As we have seen, we can obtain this proposition by introducing direct financing methodsinto Bernanke and Gertler (1987) model. If bank credit was the sole means of firmsfinancing, as assumed in Bernanke and Gertler (1987) model10, banks lending decisionswould be entirely con
45、sistent with firms investment decisions. In our model, however,banks must decide the amounts of bank loans, taking into account the possibility that firmsmay choose direct financing rather than bank borrowings. In other words, the negative signof economic activity and the positive one of the collate
46、ral value indicate not only that banksneed to decide their lending behaviour under their liquidity constraint but also that theexistence of alternatives to bank loans plays an important role in banks decisions.4. Empirical modelIn the previous section we have derived a new model of the bank lending
47、behaviour, inwhich there are three possible regimes under asymmetric information. We can nowestimate a bank lending function Eq. (3.13) and examine which regimes apply whenbanks are making decisions on their lending. We also examine whether or not financialliberalization11and the regulation of bank
48、capital (BIS)12exert larger effects on bank loans.Eq. (3.10) for estimation is given as follows.10Bernanke and Gertler (1987) assume that d is given exogenously, i.e. d/Y d/K d/Rl0 in Eqs.(3.10), (3.11) and (3.12). Then in any regime, I is a non-decreasing function of Y, Rl, and K, and a non-increas
49、ingfunction of rf. Thus, the signs of the coefficients in (3.13) are y10;y20;y30; and y40. Note that thecoefficient of economic activity dY is always non-negative in their model.11Japanese financial liberalization has been in progress since the middle of the 1980s, in particular, theopening of a commercial paper market. Japanese financial liberalization has allowed firms to finance at lowercost and in a variety of ways, such as corporate bonds, equities, convertible bonds and warrants.12In this paper, the Basle Accord is defined as
