The Role of Bank Reputation in Certifying Future Performance Implications of Borrowers Accounting Numbers*

Transcription

1 The Role of Bank Reputation in Certifying Future Performance Implications of Borrowers Accounting Numbers* Robert M. Bushman The University of North Carolina at Chapel Hill, Kenan-Flagler Business School Regina Wittenberg-Moerman The University of Chicago Booth School of Business First Draft: July 20, 2010 This Draft: December 30, 2010 * We thank Dan Amiram, Ray Ball, Ryan Ball, Phil Berger, Douglas Diamond, Merle Erickson, Rich Frankel, Christian Leuz, Michael Minnis, David Ross, Florin Vasvari, Jieying Zhang, seminar participants at the University of Texas at Dallas, and participants at the 2010 Dopuch Conference at Washington University, the 2010 Duke/UNC Fall Camp, 2010 FEA Conference at the University of Maryland, and the 2010 Stanford Summer Camp for helpful comments. We thank the Thomson Reuters Loan Pricing Corporation for providing loan data. We gratefully acknowledge the financial support of the Kenan-Flagler Business School, The University of North Carolina at Chapel Hill, and the University of Chicago Booth School of Business.

2 The Role of Bank Reputation in Certifying Future Performance Implications of Borrowers Accounting Numbers Abstract We investigate the role played by the reputation of lead arrangers of syndicated loans in mitigating information asymmetries between borrowers and outside investors. We hypothesize that reputation serves to certify the quality of borrowers by enhancing arrangers ability to credibly reveal favorable information about unobservable aspects of the borrowers quality and to credibly commit to rigorous ex post monitoring, which in turn manifests in the superior future performance of borrowers. In particular, we hypothesize that arranger reputation facilitates the certification of the quality of borrowers accounting numbers. Using attributes of borrowers location to instrument for lender-borrower matching, we find that high bank reputation is associated with higher profitability and credit quality in the three years subsequent to loan initiation. We also show that reputation certification is associated with enhanced long run sustainability of earnings via higher earnings persistence, and enhanced debt contracting value of accounting via a stronger connection between pre-loan profitability and future credit ratings. We further document that the enhanced earnings sustainability reflects both superior fundamentals (captured by higher cash flow persistence) and accruals more closely linked with future cash flows.

3 1. Introduction Information asymmetries create frictions that can impact costs of raising external funds. In this paper we investigate the role played by the reputation of the lead arrangers of syndicated loans in mitigating information asymmetries between borrowers and outside investors. The idea is that public revelation of new syndicated loan deals can convey new information about borrowers to the market, where the credibility of the information revealed increases in the lead bank s reputation. We hypothesize that reputation serves to certify the quality of borrowers by enhancing arrangers ability to credibly reveal favorable information about unobservable aspects of borrowers quality and to credibly commit to rigorous ex post monitoring, which in turn manifests in the superior future performance of borrowers. In particular, we hypothesize that arranger reputation facilitates the certification of the quality of borrowers accounting numbers. We examine relations between the reputation of lead arrangers and key dimensions of borrowers performance subsequent to loan origination and provide evidence consistent with our hypothesis. The syndicated loan market is a powerful setting in which to examine the implications of reputation. Syndicated loan transactions are arranged by a lead bank responsible for performing pre-loan due diligence, allocating loan principal to syndicate participants, monitoring borrowers after the loan is made and renegotiating terms over the life of the loan. Lead banks do not retain the entirety of syndicated loans and so do not capture the full benefits from costly ex ante evaluation or ex post monitoring of borrowers. This fact, together with unobservability of lead banks diligence efforts, creates potentially severe agency problems (e.g. Lee and Mullineaux, 2004, and Sufi, 2007). However, lead banks transact repeatedly in the marketplace allowing market participants to assess their reputation for veracity and diligence based on observable outcomes of previous loans arranged by a bank. Banks incentives to protect the stream of future 1

4 rents associated with costly investments in reputation supports a certification role whereby transactions associated with higher reputation banks provide more credible signals of borrowers prospects than do loans arranged by less reputable banks (e.g., Booth and Smith 1986). We take as our point of departure established evidence documenting that loan announcements generate positive abnormal returns to a borrower s equity. 1 Most pertinent to our analysis is research showing that borrowers stock price responses to loan announcements are relatively more favorable for loans arranged by higher reputation banks (Billett at al., 1995, and Ross, 2010). This evidence suggests that loan announcements reveal new information to the market about borrowers quality and that the credibility of this information is higher for loans arranged by reputable lead banks. Ross (2010) provides corroborating evidence that this response pattern is consistent with investors pricing beliefs that high reputation lead arrangers provide a higher level of certification of borrower quality than do other arrangers. 2 We extend this literature by examining whether the realized future performance of borrowers associated with high reputation lead banks substantiates investors beliefs via superior outcomes. We consider two key aspects of borrowers performance following loan origination: future profitability and future credit quality, as measured by credit ratings and CDS spreads. Further, we seek to isolate specific channels through which reputation certification operates. In this regard, we investigate the extent to which the superior future performance of borrowers associated with high reputation arrangers is reflected in the higher quality of borrowers 1 See James and Smith (2000) for a review of the empirical literature. See also Boot (2000) and Saunders and Cornett (2004) for thorough reviews of the special role of banks and relationship banking. 2 Ross (2010) shows that high reputation lead arrangers charge lower interest rates than lower reputation banks, and are less likely to be protected by a borrowing base which limits the outstanding principal on a loan to a fraction of the borrower s readily saleable assets. Together, these results support the contention that such banks have a particularly good reputation for evaluating a borrower s underlying business and true risk for default, and thus can offer borrowers more attractive terms while still inducing syndicate banks to participate in the transaction. 2

5 accounting numbers as captured by the relation between accounting numbers just prior to loan origination and future performance outcomes. It has long been argued that banks play a special certification role through private lending (Fama, 1985) and delegated monitoring activities (Diamond, 1984 and 1991). Chemmanur and Fulghieri (1994a) demonstrate analytically that higher reputation banks have incentives to perform more rigorous pre-loan evaluations of borrowers than lower reputation banks do, where more rigorous evaluation increases the likelihood that borrowers with poor prospects are discovered and denied loans. 3 Thus in equilibrium, high reputation banks are more likely to transact with firms that have superior future prospects. However, because borrowers true quality is not directly observable to the market, the fact that high reputation banks are more likely to transact with high performing borrowers implies that the observable act of a high reputation bank accepting a borrower reveals new information that reflects favorably on the borrowers future prospects. Building on this theory, we seek to empirically establish direct links between the enhanced credibility of the new information revelation associated with higher reputation lead bank and the future performance of borrowers. To isolate future performance implications of the credibility enhancing role of reputation, we must control for variables, unobservable and observable, that simultaneously impact the matching of borrowers with lead arrangers and the future performance of borrowers. Our setting involves private lending arrangements where lead banks are typically granted significant access to borrowers private information. This introduces the distinct possibility that there exists unobservable private information (to the researcher) that both underpins the decisions that 3 In a related paper, Chemmanur and Fulghieri (1994b) also show that high reputation banks credibly commit to more rigorous ex post monitoring of borrowers than do lower reputation banks. In Section 2 below, we thoroughly discuss the relation of extant theory to our empirical analysis. 3

6 determine bank-borrower matching and impacts borrowers future performance. Thus, our main analysis employs a self-selection framework designed to explicitly control for unobservable factors associated both with the likelihood of a borrower dealing with a high reputation bank and with future performance (see also Fang 2005, and Ross, 2010). 4 To implement this self-selection specification, we employ instruments based on the location of borrowers relative to potential lenders (Ross, 2010), embedding these instruments in a maximum likelihood specification that simultaneously determines the probability that a high reputation bank arranges a given loan and the impact of reputation on future performance and accounting quality. We measure bank reputation using a binary classification that classifies banks with a market share above a certain cutoff as high reputation banks. This captures the empirically observed structure of the syndicated loan market in which significant market share is concentrated among a few dominant banks, with the rest of the market split among many small players. Also, this characterization is consistent with arguments in Ross (2010) suggesting that reputation has a threshold rather than a continuous effect. 5 Turning to results, we first establish that higher lead arranger reputation is both statistically and economically associated with higher profitability and credit quality in the three years subsequent to loan initiation. For example, after controlling for all typical determinants of profitability and credit quality, in the third year following loan origination, borrowers of 4 In dealing with endogeneity, it is important to specify whether the endogenous choice of a treatment group depends on observable variables only, or whether it also depends on unobservables that are correlated with the outcome variable of interest (e.g., Heckman and Navarro-Lozano, 2004). Our empirical findings suggest that endogeneity with respect to important unobservable choice variables is a material concern in the context of lender reputation. However for robustness, we re-estimate the analysis using OLS, Propensity Matching, and firm fixed effect specifications, finding that the main results continue to hold. See Section 5 below for further discussion. 5 We classify J.P. Morgan Chase, Bank of America, Citibank, Bank One, Fleet Boston, Wachovia, Credit Suisse, First Boston and Deutsche Bank as reputable arrangers. We replicate all the tests using Ross s (2010) dominant bank measure, which classifies J.P. Morgan Chase, Bank of America and Citibank as reputable arrangers; the results are robust. In Section 2, we extensively discuss the competitive structure of the syndicated loan market. 4

7 reputable arrangers report profitability that is 2.0 percentage points higher and have a credit rating that is 2.8 notches lower (lower numerical credit ratings indicate higher credit quality), relative to the borrowers of less reputable arrangers. Next, we investigate whether the performance implications of reputation certification manifest in dynamic properties of borrowers accounting numbers. Specifically, we estimate the mapping between borrowers pre-loan profitability and future performance separately for high and low lead arranger reputation partitions using an endogenous switching regression framework. We find that earnings persistence is significantly higher for the high reputation bank partition relative to the lower reputation partition, implying that borrower profitability reported at the time of loan origination is more sustainable for high reputation banks. To exemplify, a one standard deviation increase in pre-loan profitability increases profitability in the third year after loan origination by 2.0 percentage points higher for borrowers of reputable arrangers than for borrowers of less reputable arrangers, representing 18.0 percent of the profitability of the latter. We also document that borrowers with high reputation lead arrangers exhibit enhanced debt contracting value of accounting via a stronger connection between pre-loan profitability and future credit quality. That is, pre-loan profitability is positively related to future credit quality, measured by both credit ratings and CDS spread, but this relation is significantly stronger for the high reputation partition. Finally, we examine whether the higher accounting quality documented for borrowers of high reputation lead banks is a consequence of better fundamentals, higher accrual quality, or both. We follow the approach in Minnis (2010) and regress one-year-ahead cash flows from operations on contemporaneous cash flow and accrual components of net income separately for the high and low reputation partitions. We find that borrowers using high reputation banks 5

8 exhibit relatively stronger fundamentals as captured by higher cash flow persistence, as well as significantly greater accrual quality. Accruals of borrowers in the high reputation partition are more strongly related to future cash flows than accruals of other borrowers. This paper makes several substantive contributions. First, we directly extend Billett et al. (1995) and Ross (2010) by establishing that the higher stock returns associated with the loan announcements of higher reputation banks are indeed consistent with rational investor beliefs that high reputation lead arrangers provide a higher level of certification of borrower quality than do other arrangers. We show that such beliefs are later substantiated via the superior economic performance of borrowers associated with higher reputation lead banks. Second, while a large literature examines the role of auditor reputation in influencing reported accounting numbers, much less is known about the role of financial intermediaries in establishing the credibility of accounting numbers for investors. 6 By documenting that high lead bank reputation is associated with higher earnings and cash flow persistence and earnings that are more diagnostic of future credit ratings, we show that bank reputation certifies aspects of accounting earnings that are distinct from those typically attributed to audit certification. Except in extreme cases where the status of a firm as a going concern is questionable, auditors do not generally speak to the long run performance implications of reported accounting numbers (see Weber and Willenborg, 2003). Our results on lead bank reputation illustrate that the process of establishing the credibility of accounting reports operates though multiple channels, where 6 The audit literature documents that auditor reputation is associated with less IPO underpricing (Beatty, 1989), fewer accounting errors and irregularities (Defond and Jiambalvo, 1991), higher earnings response coefficients (Teoh and Wong, 1993) and lower abnormal accruals (Becker et al., 1998). See Francis (2004) for a review of the literature. With respect to the reputations of underwriters and venture capital investors, Lee and Masulis (2009) show that high reputation is associated with significantly less earnings management, while Agrawal and Cooper (2009) find that venture capital reputation is associated with lower levels of post-ipo restatements. In contrast, in a related study, Cohen and Langberg (2009) find that venture backed firms have earnings that are less informative compared to non-venture-backed firms. 6

9 intermediaries other than auditors can use their reputation to certify deep properties of firms accounting numbers. Of notable interest to the debt contracting literature is our result concerning the credibility enhancing role of lead arranger reputation with respect to the debt contracting value of accounting numbers, whereby pre-loan profitability is more diagnostic of future credit ratings and CDS spreads for borrowers of higher reputation lead banks. 7 Where a growing literature examines the effects of accounting quality on the credit risk of borrowers (e.g., Francis et al., 2005, Ashbaugh-Skaife et al., 2006, Zhang, 2007, and Bharath et al., 2008), our analysis suggests that important determinants of the relation between accounting numbers and the long run credit quality of borrowers include unobservable private information and arrangers commitment to ex post monitoring, and that lead arranger reputation plays an important role in credibly revealing information about these determinants to the market. The rest of the paper is organized as follows. Section 2 presents the conceptual basis of reputation certification as developed in the extant literature. Section 3 discusses our empirical design and econometric approach to self selection issues. Section 4 describes the sample and data, and presents descriptive statistics, while Section 5 presents our main results. Section 6 discusses robustness issues and Section 7 concludes. 2. The conceptual basis of reputation certification and related literature Financial intermediaries play an important role as information producers (e.g., Leland and Pyle, 1977, Campbell and Keracaw, 1980 and Diamond, 1984). However, an intermediary s effectiveness in transmitting information hinges critically on market perceptions of its credibility. The literature posits reputation as a mechanism for projecting credibility to the market. Klein and 7 As discussed below, we also establish that pre-loan profitability is more diagnostic of future credit ratings downgrades for borrowers with high reputation lead banks. 7

10 Leffler (1981), DeAngelo (1981) and Shapiro (1983) demonstrate that reputable firms have greater incentives to supply high quality goods to the market. In equilibrium, rents flow to firms as compensation for sunk investments in reputational capital, and such rents serve to bond the firm since it is vulnerable to loss if shirking is detected. Booth and Smith (1986) apply this reputational capital paradigm to explain the role of investment bankers in certifying the pricing of equity and risky debt issues. 8 Chemmanur and Fulghieri (1994a, b) formally model the role of intermediary reputation in facilitating the credible transmission of information to the market. Chemmanur and Fulghieri (1994a) focus on the ex ante evaluation of borrowers by banks. They model a multi-period setting where firms seek to raise capital by employing the services of an investment bank. True firm quality is unobservable to those outside the firm. Banks evaluate firms future prospects by incurring costs that are increasing in the rigor of evaluation, where more rigorous evaluation makes it more likely that a borrower with poor prospects will be discovered and denied a loan. Borrowers who pass a more rigorous evaluation will perform better on average in the future than those firms granted financing on the basis of a less rigorous evaluation. However, a bank s choice of rigor is not observable to outsiders, and so, to alleviate moral hazard, investors assess a bank s reputation for supplying rigorous evaluations by conditioning on the observable outcomes of past transactions sponsored by the bank. In equilibrium, banks with higher reputations adopt more rigorous evaluation standards than lower reputation banks do, and this is reflected in more favorable financing terms and in firms raising capital from the most reputable intermediary that 8 Accounting theory shows that high reputation auditors can provide firms with a mechanism to reveal private information. See for example, Titman and Trueman (1986), Bachar (1989) and Datar et al. (1991). Similarly, Ball et al. (2009) show that commitment to a higher quality audit certifies the quality of management earnings forecasts. 8

11 will agree to provide certification. That is, the fact that a high reputation bank agrees to engage with a firm reveals favorable information about that firm. Focusing on ex post monitoring, Chemmanur and Fulghieri (1994b) develop a multi-period model where the ability to acquire a reputation provides banks with endogenous incentives to devote more resources to information production about firms who subsequently become financially distressed. Investors believe that more reputable banks supply a greater ex post monitoring effort, and, in equilibrium, they do. Taken together, Chemmanur and Fulghieri (1994a, b) show that the high rigor of banks ex ante evaluation and their commitment to extensive ex post monitoring is inextricably embedded in reputation via equilibrium incentives. These models provide the theoretical rationale for the empirical results in Ross (2010) and for our analysis. The public announcement that a high reputation bank has agreed to arrange a borrower s loan reveals new information to the market both through the fact that the borrower has passed a very rigorous evaluation and that the bank is bonded to supply high ex post monitoring efforts. The nature of reputation certification as characterized by Chemmanur and Fulghieri (1994a) implies that the economic consequences of bank reputation cannot be strictly interpreted in terms of a classical treatment effects model, where subjects are randomly assigned to two groups with one group receiving the treatment while the other does not. The reason for this is that even if borrowers were randomly assigned to high and low reputation banks, borrowers who failed to pass the rigorous evaluation process of high reputation banks would be rejected for a loan. The act of a high reputation bank accepting a borrower is itself an observable information event that credibly reveals private information to the market. To empirically isolate the future performance implications of reputation s credibility enhancing role, we must control for variables, both unobservable and observable, that simultaneously impact the endogenous 9

12 matching of borrowers with lead arrangers and the future performance of borrowers. We discuss this point further in Section 3 below in the context of our empirical design. In addition, there exists considerable empirical evidence that maintaining a high reputation is a primary concern for financial intermediaries. With respect to syndicated lending by banks, Gopalan et al. (2009) show that a default by a lead arranger's borrower adversely affects the bank s subsequent lending activity, consistent with a loss of reputation. More generally for banks, the recent financial crisis provides powerful evidence of the importance of reputation. Prior to the crisis, many banks sponsored structured investment vehicles (SIVs) to fund the purchase of long-term assets with short-term asset-backed commercial paper (ABCP). Many of these SIVs were structured so that the sponsoring bank had no contractual obligation to provide liquidity facilities or guarantees to the SIVs. However, when banks were unable to roll over the ABCP funding for these SIVs, to preserve their reputations, they took the SIVs assets on their balance sheets (Crouhy et al., 2008). Further, analogous to the result in Chemmanur and Fulghieri (1994a), where banks reject firms who fail the pre-loan evaluation to protect the bank s reputation, there is evidence that auditors reject risky firms as clients to protect the audit firm s reputation and litigation risk (Johnstone and Bedard, 2003). Finally, the syndicated loan market is dominated by a small number of banks that arrange a significant share of all loans in the U.S. We define a bank as a reputable lead arranger if its average market share over our sample period ( ) is above 2 percent (see Appendix A for a detailed discussion). We classify J.P. Morgan Chase, Bank of America, Citigroup, Wachovia, Credit Suisse First Boston and Deutsche Bank as reputable arrangers. We also classify J.P. Morgan, Bank One and Fleet Boston as reputable arrangers over the period prior to their merger/acquisition. Together, these banks syndicated over 65% of the loan issuance (by 10

13 volume) over the sample period. In contrast, the remaining syndicated loans were arranged by more than 1,000 banks, the vast majority of which had a market share of less than 0.02%. 9 While we interpret market share as reputation, such dominant market share could be the consequence of something other than reputation. First, as noted by Ross (2010), dominant banks syndicate most of the loan principal to other syndicate participants, and so their dominant market shares could not plausibly result from offering terms unattractive to other lenders or from passing on an advantage in raising money at favorable rates. Perhaps instead, the dominant banks have achieved oligopolistic market power through some structural barrier that suppresses competition in this market. If so, these banks should then be able to extract rents from borrowers, in the form of higher interest rates. However, Ross (2010) finds that these banks actually charge lower interest rates than less reputable banks, consistent with predictions of reputation certification demonstrated in Chemmanur and Fulghieri (1994a). Ross (2010) also finds that the dominant banks are more likely to lend without the protection of a borrowing base, which limits the outstanding principal on a loan over the course of its term to a fraction of the borrower s readily saleable assets. Together, these results are consistent with such banks having a strong reputation for evaluating borrowers underlying business and true default risk, allowing them to offer borrowers more attractive terms while still inducing syndicate banks to participate in the transaction. In a related paper, Fang (2005) shows that the dominant investment banks in the bond underwriting market obtain lower yields and charge higher fees, but that issuers net proceeds are higher; these results 9 Given that Bank of America and Citigroup had to be bailed out in the recent financial crisis, one may question our conjecture that these banks are reputable. However, Bank of America and Citigroup remained dominant arrangers over the period (with an average market share of 18.3 and 14.4 percent, respectively), suggesting that their reputation continues to be strong in the syndicated loan market. J.P. Morgan Chase, Credit Suisse First Boston, Deutsche Bank and Wachovia (until its merger with Wells Fargo) also continued to lead syndicated loan issuances over the crisis period (Bank One and Fleet Boston merged with J.P. Morgan Chase and Bank of America, respectively, in 2004). 11

14 are again consistent with predictions in Chemmanur and Fulghieri (1994a). Overall, this evidence provides support for the presumption that these banks have achieved a high market share by building strong reputations for reducing information asymmetry between the corporations that raise capital and the investors who ultimately provide it. 3. Empirical design In Section 3.1, we succinctly describe the econometric framework employed in our analysis and the instruments used in the estimation to facilitate identification of the impact of reputation certification on borrowers post-loan performance. In Section 3.2, we discuss our strategy for isolating the implications of reputation certification for borrowers future performance and accounting quality Econometric Framework and Instruments Consistent with recent research on the certification role of bank reputation (Fang, 2005, and Ross, 2010), we employ a self-selection framework designed to explicitly control for unobservable factors associated both with the choice of lead arranger and the borrower s future performance. The appropriateness of this framework rests on the observation that lead banks have significant access to private information (e.g., quality of a borrower s management, additional relationship business from the borrower, etc.) and as such, it is plausible that such private information underpins the decisions of both banks and borrowers that determine the observed matching between them. This private information may be also associated with a borrower s performance subsequent to loan origination. The basic framework consists of a binary outcome equation that models matching between borrowers and lead banks, and two regression equations, one for each reputation partition (see Maddala, 1983, Chapters 8 and 9). Formally, we have: 12

15 Reputation * i =Z i γ+ε i, (1) high Future Performance rep i =x i β high rep high +μ rep i, and (2) low Future Performance rep i =x i β low rep low +μ rep i. (3) Equation (1) is the latent borrower lead arranger matching equation. Given our binary approach to classifying lead banks as either high or lower reputation, we implement (1) with a dummy variable set equal to one if and only if a loan is arranged by a high reputation bank, zero otherwise. Conceptually, Reputation i 1 if Reputation * i 0, and Reputation i 0 Reputation 0. Equation (2) is the performance equation for reputable banks, and (3) is that for less reputable banks, where future performance represents either profitability or credit quality. The vector x i includes observable borrower and loan characteristics posited to directly influence borrowers future performance. Endogeneity is modeled by allowing the residuals in the performance equations (2) and (3) to be correlated with the residual in equation (1). The model high assumes that the three error terms ε i, μ rep low i, μ rep i have a trivariate normal distribution, and simultaneously estimates equations (1), (2), and (3) by maximum likelihood. 10 By incorporating this correlation structure into the estimation, we seek to control for the possibility that unobservable variables embedded in the error term of (1) also affect future performance via unobservables embedded in the residuals of equations (2) and (3). The vector Z i in (1) represents observable variables important to determining matching between borrowers and lead banks. Our estimation procedure dictates that Z i include variables 10 We verify that all results are robust to using a two-step procedure which: 1) estimates equation (1) as a probit model and derives the inverse Mills ratio from the fitted values of Reputation, and then 2) includes the calculated inverse Mills ratios in equations (2) and (3) to account for unobservable factors affecting both the decision to choose a high reputation bank and future performance. 13

16 that predict the likelihood that Reputation equals 1 in a probit regression, where some of these variables are valid instruments that predict this likelihood but are also independent of the borrowers future performance and thus are properly excluded from the second stage regression. 11 We use geography as the basis for constructing such instruments. As argued by Ross (2010), it is not plausible that a lender or borrower would change the location of its headquarters solely for the purpose of consummating a loan. Moreover, it is not obvious why two otherwise observationally equivalent borrowers should differ in their susceptibility to the loan certification effect and in post loan performance, merely because the borrowers headquarters are located in different regions of the United States. By contrast, location likely plays a large role in determining the propensity of lenders and borrowers to do business together. Prior research demonstrates that proximity to the borrower enhances lenders information gathering and monitoring abilities (Peterson and Rajan, 2002, Berger et al., 2005, Sufi, 2007, and Dass and Massa, 2009). The importance of proximity is supported by evidence that a shorter distance between the borrower and the lender facilitates lenders collection of soft information (Peterson and Rajan, 2002, and Berger et al., 2005) and access to private information (Dass and Massa, 2009). Further, the home bias literature also suggests that investors have better access to information about local firms (e.g., Coval and Moskowitz, 1999 and 2001). We construct two instruments. The first derives from the idea that a borrower is more likely to borrow from a high reputation bank if its own headquarters is in close proximity to a high reputation arranger. Lead arrangers screening and monitoring costs, as well as the transaction costs of the borrower (e.g., the travel time of its executives), may be lower in this 11 Strictly speaking, exclusion restrictions are not necessary in the self selection model because the model is identified by non-linearity. However, such identification by non-linearity cannot be guaranteed in a given setting. See Heckman and Navarro-Lozano (2004) for further discussion. 14

17 case, increasing the probability of an arranger-borrower matching. We define the variable Nearreputable-arranger as equal to 1 if the borrower s headquarters is within 60 km of any high reputation banks headquarters, zero otherwise. 12 We expect Near-reputable-arranger to be positively associated with the choice of a high reputation lead bank. The second instrument captures the idea that a borrower will be less likely to borrow from a high reputation bank if there are credible local alternative lenders nearby. We consider a local bank to be a credible alternative if it arranges on average 30 syndicated loan deals per year over the sample period, suggesting that it has a considerable expertise in syndicated lending. 13 This variable, Near-local-arranger, is set equal to 1 if a borrower s headquarters is within 60 km of a credible regional lead arranger s headquarters and zero otherwise. We expect Near-localarranger to be negatively associated with the choice of a high reputation lead bank. See Appendix A for a more detailed description of these instruments. 3.2 Estimating the future performance implications of bank reputation We first investigate the basic question of whether firms with high reputation lead banks exhibit superior future performance (Section 5.2 below). Here, we utilize a basic treatment effects model that includes a reputation dummy variable. 14 This represents a logical starting point as it directly extends the model used by Ross (2010) to estimate the impact of bank reputation on loan announcement equity returns. The specification simplifies equations (1), (2) and (3) to a two 12 Following Ross (2010), we conjecture that while reputable banks have satellite offices in many regions of the country, these offices rarely engage in the syndicated loan activities of large, public borrowers that are the focus of this study. 13 The results are not sensitive to the threshold number of syndicated loan deals. When we change the threshold to 40 or 60 deals per year, this instrument performs similarly in both the arranger-borrower matching equation and the future performance equation. 14 We use the treatment effects model primarily to facilitate simple comparisons of the results using the self selection model to alternatives such as OLS, propensity matching and firm fixed effects. We also compute treatment effects using counterfactuals constructed from the endogenous switching model estimated in Section

18 equation framework which retains equation (1), and replaces equations (2) and (3) with a single equation (4) that restricts β high rep β low rep. The single equation takes the form: Future Performance i = Reputation i x i β + μ i. (4) Reputation i is set equal to one if a loan is arranged by a reputable bank, zero otherwise. The error terms ε i, μ i are assumed to be jointly normally distributed, where this feature controls for unobservable variables impacting both reputation choice and future performance. The coefficient represents an unbiased estimate of the impact of reputation certification on future performance. That is, captures the future consequences that derive from firms having passed the more rigorous evaluations of high reputation banks and from any benefits of the superior ex post monitoring supplied by high reputation banks, after controlling for the observable and unobservable factors that influence both firm-bank matching and firms future performance. To investigate whether reputation certification operates through the quality of borrowers accounting numbers, we estimate the full switching regression framework represented by equations (1), (2) and (3) above (see Section 5.3 below). We estimate separate performance mappings for high and low reputation partitions. The implications of reputation certification for accounting quality are captured by significant differences in the mapping between borrowers pre-loan profitability and future performance across high and low reputation partitions. When future performance is profitability, significant differences in the mapping between borrowers pre-loan profitability and their future profitability capture differences in earnings persistence across reputation partitions. When future performance is credit quality, significant differences capture differences in the strength of the connection between pre-loan profitability and future credit ratings. We turn next to our empirical analysis. 16

19 4. Sample, data and descriptive statistics 4.1. Data sources and sample selection We employ the DealScan database provided by the Thomson Reuters Loan Pricing Corporation (TRLPC). DealScan covers a majority of syndicated loan issues in the U.S. and provides a wide range of loan characteristics, such as interest rate, amount, maturity, purpose, covenants and syndicate structure. We obtain firm characteristics from COMPUSTAT. Firms senior debt ratings, watchlist additions and outlook changes (at the firm level) are retrieved from the S&P historical database. If the S&P historical database does not cover a particular firm, we retrieve the Moody s, Fitch or DPR senior debt rating from Mergent Fixed Income Securities Database (FISD). CDS data is obtained from Markit. Table 1 summarizes the sample selection process. For the period 1998 to 2006, DealScan reports 68,368 facilities outstanding to U.S. firms and issued in U.S. dollars. Merging this data with COMPUSTAT allows us to identify 25,518 facilities issued to public firms. Next, we exclude facilities with insufficient loan and firm data, leaving 10,279 facilities. Most loans excluded due to insufficient data relate to firms without credit ratings on either the S&P or FISD databases. Lastly, we eliminate the facilities of firms without the 5 consecutive years of COMPUSTAT data, required for estimating a borrower s performance prior to a loan s issuance and following it. The remaining sample contains 6,002 facilities related to 1,352 firms Descriptive statistics Panel A of Table 2 reports that 67 percent of sample loans are issued by reputable arranger (detailed variable definitions are in Appendix A). Sample loans have, on average, a size of $460M, a maturity of 42 months and an interest-spread of 173 basis points. Institutional loans represent 14 percent of the sample loans, 17 percent of the loans are issued for restructuring 17

20 purposes, and 53 percent are syndicated by a relationship arranger. In terms of credit quality characteristics at loan origination, sample firms have a mean and median S&P senior debt rating of BB+ and BBB-, respectively. The mean CDS spread on the loans with traded CDS is 1.75 percent, and 8 (17) percent of the loans relate to firms on the S&P negative watch list (Outlook). Sample firms are relatively large, with a mean and median value of total assets of 7,525M and 2,300M, respectively, and they have average profitability, as measured by the ratio of EBITDA to total assets, of 13%. In Panels B and C of Table 2, we report summary statistics for the reputable and less reputable arranger samples, respectively. The two samples differ along a number of dimensions. Loans syndicated by reputable arrangers are larger, have a shorter maturity and are also more likely to be issued by arrangers that have a previous relationship with the borrower; loans with a prior lending relationship represent 56 percent of the reputable arranger sample relative to 45 percent of the less reputable arranger sample. This evidence is consistent with Fang (2005) who finds that reputable underwriters in the public bond market are more likely to have a prior relationship with the issuers, compared to less reputable underwriters. As suggested by the Interest-spread, Credit-rating, CDS-spread and Leverage variables, reputable arrangers loans are issued to less risky firms relative to the firms of less reputable arrangers. Reputable arrangers borrowers are significantly larger but do not differ in terms of profitability at the time of the loan s origination. Supporting the appropriateness of our instruments, we find that the firms of reputable arrangers are significantly more likely to be in close geographical proximity to the reputable lead banks and are significantly less likely to be located close to a credible local lead bank. 18

21 5. Empirical Results 5.1. The borrower-lead arranger matching model Table 3 presents the estimation of the following Probit model of borrower-arranger matching, which includes loan- and firm-specific characteristics and year and industry fixed effects: Reputation = α + Firm size + Credit ratings Leverage + Profitability + Interest coverage + Loan size + Maturity + Restructuring purpose+ (5) Prior relationship + Near-reputable-arranger + Near-local-arranger. Table 3 shows that reputable arrangers syndicate loans of larger firms and firms with higher credit quality, consistent with reputational concerns. Profitability does not affect arranger choice. Reputable arrangers are more likely to syndicate larger loans; these loans generally require larger syndicates which are likely more difficult to arrange. Reputable arrangers are less likely to issue restructuring purpose loans, consistent with reputational concerns deterring them from syndicating such loans. Restructuring purpose loans are often accompanied by significant changes in a firm s capital structure and thus are associated with high uncertainty about the firm s future prospects and creditworthiness. We also find that firms are more likely to hire the same lead arranger used for previous loan transactions if the lead arranger is reputable. Lastly, both of our geography instruments are significant with the predicted signs. First, Near-reputable-arranger, which equals 1 if the borrower s headquarters is within 60 km of a high reputation banks headquarters, and zero otherwise, is significantly positively associated with the choice of a high reputation lead. Economically, being in proximity to a reputable lead arranger increases the probability of a borrower-reputable arranger match by 7.4%. The second instrument, Near-local-arranger, which equals 1 if a borrower is within 60 km of a major local 19

22 bank, has a significantly negative coefficient, consistent with borrowers being less likely to borrow from a high reputation bank if there are credible local alternatives. Close proximity to a local credible lead arranger decreases the probability that a loan is syndicated by a reputable arranger by 6.2%. A partial-f statistic of (p-value of 0.00) indicates that the Nearreputable-arranger and Near-local-arranger variables are collectively strong instruments for the choice of bank reputation. Also, the partial R-Square of 2.9 percent reveals that these variables have significant explanatory power, increasing the R-Square by 19% relative to a specification excluding them Bank reputation and borrowers future profitability and credit ratings As discussed in Section 3.2 earlier, our first series of future performance analyses employ a treatment effects model of the form: Future Performance i = Reputation i Firm Controls +β 2 Loan Controls μ i. (6) In (6), Reputation i equals one if a loan is arranged by a reputable bank, while Future Performance i is either a borrower s profitability or credit rating in each of the three years subsequent to loan initiation. Firm controls include borrowers profitability, credit ratings, size, leverage, and interest coverage ratio, all measured prior to loan initiation. In terms of loan characteristics, all specifications control for initial interest spread, loan maturity, whether the loan was issued for restructuring purposes, and whether the loan is institutional. We also control for relationship lending. When Future Performance is credit ratings, we augment the controls with measures of whether the borrower is on the S&P watch list or has an S&P outlook (negative or positive) at the time of a loan s issuance. The coefficient then captures whether the reputation of the lead arranger has incremental explanatory power for future performance after 20

23 controlling for observable determinants of future performance and unobservable determinants via the self selection estimation. 15 Panel A of Table 4 provides a univariate comparison of the future profitability and credit ratings of borrowers of reputable and less reputable arrangers. In terms of profitability, while there is no significant difference in the mean profitability of the borrowers of reputable and less reputable arrangers prior to a loan s issuance, borrowers with reputable arrangers exhibit significantly higher mean profitability over the three year period following a loan s issuance. This result is driven by the drop in the mean profitability of borrowers of less reputable arrangers, from 0.13 prior to the loan issuance to 0.12 in the following years. In terms of credit ratings, Panel A of Table 4 reveals that borrowers of reputable arrangers have a lower credit rating, by approximately 2 notches, compared to borrowers of less reputable arrangers (remember that lower numerical credit ratings indicate higher credit quality). Table 4, Panel B reports the results of estimating equation (6) with Future Performance measured as a borrower s profitability in the first, second and third years following the year of a loan s issuance. We find that loans issued by reputable arrangers have a significantly higher profitability following loan issuance. The coefficient of on Reputable-arranger in column (1) indicates that in the year following the year of loan issuance, borrowers with reputable arrangers report an ROA that is 1.3 percentage points higher than it is for borrowers with less reputable arrangers. This difference is economically significant, representing 10.7% of the mean profitability of borrowers of non-reputable arrangers in that year. The difference in profitability across borrowers of reputable and non-reputable arrangers increases over time. In the second (third) year following the year of loan issuance, borrowers with reputable arrangers report an 15 All specifications also contain year and one digit industry fixed effects. The analyses cluster standard errors at the firm level. 21

24 ROA that is 1.8 (2.0) percentage points higher than that of borrowers with non-reputable arrangers, representing 14.9 (18.8) percent of the mean profitability of borrowers with less reputable arrangers in the second (third) year following the year of a loan s issuance. The coefficients on the control variables suggest that smaller and more leveraged firms have higher future profitability. As expected, a firm s past profitability is one of the most important determinants of its future performance. We also find that longer maturity loans are associated with higher future profitability. The interest spread, which captures firm/loan riskiness that is not captured by other control variables, is negatively related to future profitability. Note also that, the correlation between error terms in equations (5) and (6), is negative and significant in columns (1), (2) and (3); this evidence supports our proposition that unobservable factors affect both borrower-lead arranger matching and a borrower s future performance. Columns (4), (5), and (6) reflect results from running the OLS specifications for comparison. The OLS results are qualitatively similar to the results that control for self selection, with a positive, significant coefficient on the reputation dummy for a borrower s profitability in the first, second and third year following the year of a loan s issuance. However, the economic significance is reduced relative to the self selection specification, illustrating the importance of controlling for the endogeneity of arranger-borrower matching. Finally, Table 4, Panel C reports the results of estimating equation (6) with Future Performance measured as a borrower s credit ratings in the first, second and third year following the year of a loan s issuance (Columns 1-3). We find that borrowers with reputable arrangers have significantly better credit ratings (i.e., lower) following a loan s issuance. The difference in credit ratings across borrowers of reputable and less reputable arrangers is 1.6, 2.3 and 2.8 notches in the first, second and third year following the year of a loan s issuance, 22