1、THE JOURNAL OF FINANCE VOL. LXIV, NO. 3 JUNE 2009Bank Loan Supply, Lender Choice, and CorporateCapital StructureMARK T. LEARYABSTRACTThis paper explores the relevance of capital market supply frictions for corporatecapital structure decisions. To identify this relationship, I study the effect on fir
2、ms fi-nancial structures of two changes in bank funding constraints: the 1961 emergence ofthe market for certificates of deposit, and the 1966 Credit Crunch. Following an expan-sion (contraction) in the availability of bank loans, leverage ratios of bank-dependentfirms significantly increase (decrea
3、se) relative to firms with bond market access. Con-current changes in the composition of financing sources lend further support to therole of credit supply and debt market segmentation in capital structure choice.ARE CAPITAL MARKET SUPPLY FRICTIONS relevant for corporate capital structure deci-sions
4、? If the supply of different forms of capital is infinitely elastic, as assumedby Modigliani and Miller (1958), then debt levels are determined solely by afirms demand for debt. However, several recent pieces of evidence suggest thismay not be the case. First, anecdotal (Titman (2002) and survey (Gr
5、ahamand Harvey (2001) evidence suggest that practitioners view supply conditionsas important inputs to the capital structure decision. Second, Faulkender andPetersen (2006) present evidence that firms with a bond rating have higherleverage ratios than those without, even after controlling for debt d
6、emand.These authors interpret this finding as suggesting that debt market segmenta-tion may put constraints on some firms ability to borrow. As a result, observedleverage ratios may not reflect those demanded.Identification of a supply effect with cross-sectional evidence is complicatedby concerns o
7、ver direction of causality and potential endogeneity. As Faulkenderand Petersen (2006, p. 75) note, however, their “findings raise the possibilitythat shocks to certain parts of the capital markets affect firms finances dif-ferentially.” In particular, if some firms cannot easily move from private t
8、oLeary is from Johnson Graduate School of Management, Cornell University. I would like tothank the Acting Editor, Mitchell Petersen. I also thank Alon Brav, John Graham, Roni Michaely,Manju Puri, Michael Roberts, David Robinson, Vish Viswanathan, Jaime Zender, two anonymousreferees, and seminar part
9、icipants at Duke University, Boston College, Columbia University, Cor-nell University, Emory University, Harvard Business School, Northwestern University, Ohio StateUniversity, Stanford University, University of Chicago, University of Illinois, University of NorthCarolina, University of Notre Dame,
10、University of Rochester, University of Utah, University ofWashington, Vanderbilt University, the 2006 Federal Reserve Bank of Chicago Conference onBank Structure and Competition, and the 2006 Western Finance Association Meetings for helpfulcomments.11431144 The Journal of FinanceRpublic debt markets
11、, shocks to the banking system may have the largest im-pact, an issue they leave to future research. Put differently, the link betweendebt market segmentation and capital structure may be more sharply identi-fied by studying the differential impact of a shift in loan supply on the capitalstructures
12、of firms with varying degrees of bond market access.In this paper, I exploit a loosening and a tightening in bank funding con-straints to identify the effect of supply frictions on both leverage ratios and thechoice of capital provider. The loosening event of interest is the emergence ofthe market f
13、or negotiable certificates of deposit (CDs) in 1961. This financialinnovation allowed banks to reverse the outflow of deposits caused by a shiftin corporate cash holdings from noninterest bearing bank deposits to othermoney market instruments. The tightening event is the 1966 Credit Crunch,in which
14、government pressure for credit restraint and the imposition of Regu-lation Q interest rate ceilings limited banks ability to access time deposits andextend new loans. As I discuss below, these events provide a particularly well-suited quasi-experiment for identifying the effect of supply frictions o
15、n capitalstructure. First, there is strong evidence that they represent changes in creditsupply. Second, the supply shocks are bank-specific rather than shocks to totalcapital supply. Third, their effects on the cross-section of financial structureswere unlikely to be driven by any concurrent change
16、s in credit demand.The expected response to these loan supply shocks depends on a firms accessto different segments of the capital markets. For example, when faced with acontraction in loan supply, firms without access to public debt markets willneed to find alternate sources of capital to avoid cap
17、ital constraints.1Thesemay include internal funds, external equity, trade credit, or nonbank privatedebt. With the exception of the last possibility, all of these substitutions wouldresult in relatively lower (higher) leverage following a loan supply contraction(expansion). Larger firms, on the othe
18、r hand, will likely be less affected for tworeasons. First, banks lending to small, risky firms may be more sensitive tocredit supply than their lending to larger firms (e.g., Holmstrom and Tirole(1997). Second, larger firms can more easily substitute toward nonbank publicdebt sources in response to
19、 changes in the cost or availability of bank debt. Bytreating firms with debt market access as a control group, I can exploit thisdifferential sensitivity to the supply shocks in order to identify the effect ofsupply frictions on capital structure.I begin by showing that, consistent with these predi
20、ctions, the leverage ofsmall, bank-dependent firms rises (falls) relative to that of large, less bank-dependent firms following positive (negative) loan supply shocks. If these lever-age changes are caused by changes in bank loan availability, they should beaccompanied by relative shifts in the comp
21、osition of debt finance. Consistentwith this prediction, I show that the ratio of long-term bank debt to totallong-term debt increases (decreases) for small, bank-dependent firms, relative1Gertler and Gilchrist (1994), Kashyap, Lamont, and Stein (1994), and Hancock and Wilcox(1998) show evidence tha
22、t loan supply shocks lead to capital constraints for small private firms.As Faulkender and Petersen (2006) discuss, however, such constraints are likely to be less severefor publicly traded firms.Bank Loan Supply, Lender Choice, and Corporate Capital 1145to firms with public market access, following
23、 positive (negative) loan supplyshocks. Additionally, the use of equity capital by small firms, relative to largefirms, is negatively correlated with loan supply shifts. Finally, the use of publicdebt by firms with access to public markets increases, relative to that of smallfirms, following the 196
24、6 Credit Crunch.These results suggest that bank loan supply movements are also an impor-tant determinant of variation in firms debt placement structures. They alsohelp clarify the mechanisms behind the relative leverage changes. First, themix of debt sources appears more sensitive to loan supply shi
25、fts for small firmsthan for large firms. This suggests that in response to these supply shocks,loan markets do not clear simply through a single price mechanism. Rather,lending terms appear to change differentially across borrower types. Second,the shift by large firms into public debt in response t
26、o tight loan supply canexplain the persistence of the leverage effect, given the significant adjustmentcosts associated with public capital markets.I also show evidence that the effects of credit supply shocks on capital struc-ture are not limited to the particular events I study. Using an extension
27、 ofthe empirical model in Faulkender and Petersen (2006) and several proxiesfor the tightness of credit conditions, I show that the cross-sectional differencein leverage between firms with and without bond market access is negativelycorrelated with loan supply over a 30-year sample period.This study
28、s findings have several implications for the capital structure liter-ature. First, the informational asymmetries and fixed transaction costs that cre-ate segmentation in debt markets are shown to be relevant for capital structurechoice. Second, these results suggest that differential responses to cr
29、edit sup-ply shocks may account for some of the previously unexplained heterogeneityin capital structures (e.g., Welch (2004), Lemmon, Roberts, and Zender (2008).As discussed by Titman (2002) and Brealey and Myers (1996), one way thatfirms can use financing choices to enhance firm value in imperfect
30、 markets isby altering the security issued in response to a supply-demand imbalance inthe capital market. In the case of segmented lending markets, frictions lim-iting banks available loan supply can create such imbalances. Finally, whilemany previous studies have documented a positive relationship
31、between firmsize and leverage, uncertainty remains as to the economic interpretation ofthis relationship.2For example, while many authors have interpreted firm sizeas a proxy for expected bankruptcy costs, cross-country evidence documentedby Rajan and Zingales (1995) does not support this interpreta
32、tion. To the ex-tent that smaller firms are subject to relatively larger transaction costs andinformational frictions, however, my results suggest size may be an importantcapital structure determinant because it proxies for debt market access.The remainder of the paper is organized as follows. Secti
33、on I provides somebackground on the historical events I study and presents the empirical hypothe-ses. Section II discusses the experimental design and identification strategy.Section III describes the data sources. Results for the impact of these events on2See, for example, Rajan and Zingales (1995)
34、 and Kurshev and Strebulaev (2006).1146 The Journal of FinanceRleverage ratios, debt placement structure, and financing choice are presented inSection IV. Section V extends the leverage analysis to a longer time period andalternative credit condition proxies. Section VI discusses further implication
35、sof these findings and concludes.I. Credit Market Background and HypothesesA. 1961: The Emergence of the Bank CDAs discussed by Mishkin (2003), before the 1960s banks viewed their liabil-ities as essentially fixed by investor demand for deposits. Yet, throughout the1950s, as corporations became more
36、 adept at cash management, they increas-ingly moved idle cash from noninterest bearing bank deposits to other moneymarket instruments. According to Morris and Walter (1993, p. 37), “Demanddeposits and currency as a percentage of total financial assets of nonfinancialcorporate businesses declined fro
37、m 29% in 1946 to 16% in 1960.”The result for banks was the loss of an important source of loanable funds.As reported by the New York Times in early 1961 (Kraus (1961a, p. 58),Corporations have pared their demand deposits and increased their in-vestments in Treasury bills, acceptances and commercial
38、and finance pa-per. New York banks doing business chiefly with large corporations.have sustained deposit declines.Indeed, GNP growth in the 1950s (80%) far outstripped that of demand de-posits (21%) (Nadler (1964). From 1950 to 1960, checking account deposits andtime and savings accounts at commerci
39、al banks increased by $25 billion and$35 billion, respectively. By comparison, time and savings accounts outside ofcommercial banks rose by $106 billion (Kraus (1961b).At the time, banks had limited ability to stop this deposit outflow. Federalregulation prevented the payment of interest on demand d
40、eposits. While theywere permitted to pay interest on time deposits, this was capped by the Fed-eral Reserve at 3%, below rates offered by savings and loans and, importantly,these deposits lacked the liquidity offered by Treasury bills and commercialpaper. As described in the Wall Street Journal on J
41、anuary 4, 1961, “Commer-cial banks.have made repeated attempts in recent years to get the FederalReserve Board to boost the permitted ceiling on savings interest payments.”(p. 9) This suggests that the loss of corporate deposits was creating a bindingfunding constraint, an idea echoed in the 1959 an
42、nual report of First NationalCity Bank of New York (later Citibank):3The extent to which banks can liquidate investments for the purpose ofadding to loanable funds is limited.Except as deposits increase, regu-latory requirements impose an approximate ceiling on lending capacity.As Wojnilower (1980)
43、discusses, prior to the emergence of bank CDs, bank-dependent firms were often constrained in their access to capital because “their3See Cleveland and Huertas (1985, p. 243).Bank Loan Supply, Lender Choice, and Corporate Capital 11470510152025303540451955 1956 1957 1958 1959 1960 1961 1962 1963 1964
44、 1965 1966 1967 1968 $ Billions0%2%4%6%8%10%12%14%Net Incr in Liabilities Large Time Deposits %Figure 1. Commercial bank funding. Data are from the Flow of Funds Accounts of the UnitedStates. The bars (left-hand axis) represent the annual net change in total liabilities at U.S. com-mercial banks. Th
45、e line (right-hand axis) represents large time deposits as a percentage of totaloutstanding deposits (checkable, small time, and large time deposits).lead banks remained essentially dependent on narrow local deposit markets”(p. 284).In an effort to ease this constraint, the first large denomination
46、negotiablecertificate of deposit was issued by First National in February of 1961. Thenegotiability of this instrument, combined with the agreement of a governmentsecurities dealer, The Discount Corporation of New York, to make a secondarymarket for such instruments led to a surge in their use short
47、ly after (Roussakis(1997). Large denomination CDs at the Federal Reserve Systems weekly re-porting banks rose from less than $1 billion in 1961 to $26.1 billion in 1970,where this latter amount represented 32% of the outstanding commercial andindustrial loans (Friedman (1975).This new financing inst
48、rument allowed banks to bid for capital from a muchbroader base of investors and therefore enabled them to expand their loanportfolios. The impact of the CD on bank funding can be seen in Figure 1,which shows the annual net increase in liabilities at U.S. commercial banksalong with the percent of de
49、posits accounted for by large time deposits, basedon data from the U.S. Flow of Funds Accounts. As can be seen, the emergenceof this new market led to a rapid increase in commercial bank growth thatcorresponds precisely to a rapid increase in the use of large time deposits (i.e.,CDs) as a source of bank funds. As described by Cleveland and Huertas (1985,p. 256) in their history of Citibank:The CD.would solve the funding problem, thereby opening the way tofaster growth. Instead of matching loan commitments to the supply of1148 The Journal of Financ
