Are banks still special when there is a secondary market for loans?

Transcription

1 Are banks still special when there is a secondary market for loans? Amar Gande and Anthony Saunders October 2006 Abstract It is commonly argued that banks play a special role in the financial system because they resolve an important information asymmetry. The recent development of an active secondary market for loans could however potentially diminish this special role. This study utilizes a unique dataset of secondary market loan prices to examine this issue. We find that new loan announcements are associated with a positive stock price announcement effect even when a borrower s loans trade on the secondary market. This result also holds true for distressed borrowers who are ex ante expected to be most adversely affected by a potential reduction in bank incentives to monitor as a result of a secondary market for loans. Moreover, when a borrower s existing loans trade for the first time in the secondary loan market, it elicits a positive stock price response. Overall, our results suggest that banks continue to be special in the presence of a secondary market for bank loans, and that the bank monitoring function and the secondary market for bank loans are complementary sources of information about borrowers. This implies that banks and markets can co-exist as information producers. JEL Classification Codes: G14, G21, G22, G23, G24 Key Words: loans, monitoring, spillovers, stocks Amar Gande is from the Edwin L. Cox School of Business, Southern Methodist University (SMU), and Anthony Saunders is from the Stern School of Business, New York University. We thank Cliff Ball, Rob Hansen, Manju Puri, Anjolein Schmeits, Hans Stoll and the seminar participants at the Federal Reserve Bank of Chicago s Bank Structure Conference, the Financial Management Association Annual Meeting, the SMU Lone Star Finance Symposium and at Vanderbilt University for helpful comments. Please address all correspondence to Amar Gande, Finance Department, Edwin L. Cox School of Business, Southern Methodist University, 6212 Bishop Blvd, Dallas, TX Tel: (214) Fax: (214) agande@cox.smu.edu.

2 1. Introduction It is commonly argued that banks play a special role in the financial system because they resolve an important information asymmetry. The recent development of an active secondary market for loans could however potentially diminish this special role. 1 This study utilizes a unique dataset of secondary market loan prices to examine this issue. Several empirical studies provide evidence on the specialness of bank loans (see James and Smith (2000) and section 2 of our paper for a comprehensive review of the research on the special nature of bank loan financing). Overall, the research to date finds a robust, favorable, impact of bank loan announcements on borrowers stock returns. This result is in contrast to the insignificant or negative response by investors to announcements of most other forms of financing, including private placements of debt, straight public debt, preferred stock, convertible debt, convertible preferred, and common stock. The last decade or so has witnessed an increased commoditization of bank loans due to a well-developed secondary market for loans. Figure 1 shows that there has been a tenfold increase in secondary market loan transactions between , with the secondary market loan transactions currently exceeding $100 billion per annum. This ten-fold increase is significantly larger than the growth rate of the corporate bond market over the same period (See Section 3 for details). The rapid development of a secondary market for loans provides a unique opportunity to address whether such a market enhances or erodes the specialness of banks and bank lending. Prior studies in this area, such as James (1987), Lummer and McConnell (1989), Best and Zhang (1993), and Billett, Flannery, and Garfinkel (1995) use data from the 1970s and 1980s (a time period during which a well-developed secondary market for loans did not exist to show robust results supporting the specialness of banks). However, to the best of our knowledge, ours is the first study to examine the bank specialness issue in the presence of a well-developed secondary market for loans. 1 Banks are considered special for several reasons, including reducing the agency costs of monitoring borrowers. See, Saunders (2002) for a comprehensive review of why banks are considered special. Many theoretical models highlight the unique monitoring functions of banks (e.g., Diamond, 1984; Ramakrishnan and Thakor, 1984; Fama, 1985). These studies generally argue that banks have a comparative advantage as well as enhanced incentives (relative to public debt holders) in monitoring debt contracts. 1

3 We hypothesize that a secondary market for bank loans results in two potentially adverse influences on the specialness of banks. It can reduce the incentives of banks to monitor a borrower since a bank can use the secondary market to reduce its lending exposure to a borrower. If so, such a loan sale could significantly reduce or even eliminate a bank s incentives to monitor the borrower, and signal that the borrower is a lemon. 2 In addition, loan trading in the secondary loan market may serve as an alternative source of information to the information that is traditionally gathered by banks through loan monitoring activities. For example, given that an increase in trading activity is likely to result in a revision of the market price, part of such a change in the market price represents new information. See Stoll (2003) for some evidence from the market microstructure literature. 3 The existence of a secondary market for bank loans also allows us to draw insights into the question of whether banks and (secondary loan) markets can co-exist. Allen (1993) and Allen and Gale (1999) argue that banks and other types of secondary markets, such as stock markets fundamentally differ in the ways they process information. For example, although banks may be effective at eliminating duplication of information gathering and processing, which is likely to be helpful when people agree about what needs to be gathered and how it should be processed, banks may be ineffective in non-standard environments. For the purposes of our study, loan trading in a secondary loan market may provide some new additional information on borrower credit worthiness, at least for certain type of borrowers. This issue is directly related to the question of whether the information from loan trading in the secondary loan market complements or substitutes for the information traditionally 2 There is empirical evidence that provides support for this view. These studies examine the impact of the termination of a banking relationship on a borrower s stock returns. Specifically, Slovin, Shuska and Polonchek (1993) examine termination due to lender distress. In particular, they analyze stock price effects of firms with lending relationships with Continental Illinois Bank during its de facto failure and subsequent FDIC rescue. They document that firms where Continental Illinois Bank was the direct lender or lead manager incurred -4.16% (t-stat 3.79%) during the bank s impending insolvency ([-5,-2]). More recently, Dahiya, Puri and Saunders (2003) analyze termination through a loan sale. Based on a sample of 29 borrowers during , they find that while the stock returns of borrowers are significantly negatively impacted (e.g., a three-day ([-1,1]) CAR of -1.74%, z-stat -3.60), the stock returns of the selling banks are not significantly impacted surrounding the announcement of a loan sale. 3 Stoll (2003) provides a framework for decomposition of trading costs into information-related and noninformation related components. 2

4 gathered by banks through the loan monitoring process. Clearly, banks and secondary loan markets can co-exist more easily if the information sets are complements rather than substitutes. 4 We use a set of specific hypotheses and a unique dataset of secondary market loan prices to address the above questions, and other related issues. Our main results can be summarized as follows: First, we find that new loan announcements are associated with a positive announcement effect on the borrower s stock price even when a borrower s loans trade on the secondary market. Second, we find that distressed borrowers, for whom bank monitoring is likely to be most valuable, are not adversely affected by the presence of a secondary market for loans. Third, when a borrower s existing loans trade for the first time in the secondary loan market, it elicits a positive announcement effect on the borrower s stock price. Fourth, we find that borrower stock price effects are larger for loans that trade more. Finally, we document that borrower stock price effects are larger for loans where the lead lender bank has stronger incentives to monitor, as proxied by the lead lender bank s stake in a loan. Importantly, our results are qualitatively similar when we specifically control for the fact that a secondary loan market could improve the liquidity of a bank loan resulting in cost savings to the borrower s equity holders. Based on the above results, we conclude that banks continue to be special even in the presence of a secondary market for bank loans and that the bank monitoring function and the secondary market for bank loans are complementary rather than substitute sources of 4 There is a large literature on the banks versus capital markets debate as information producers. See Levine (2002) for the most recent survey of this debate. For example, Stiglitz (1985) argues that since well-developed markets quickly reveal information to investors at large, this dissuades individual investors from spending much time and money researching firms. This free-rider problem is less severe in bank-based systems since banks can make investments without revealing their decisions immediately in public markets. Shleifer and Vishny (1986) suggest that a liquid stock market encourages more diffused ownership, such that each investor has fewer incentives to oversee managers actively. However, Rajan (1992) argues that once banks acquire substantial, inside information about firms, banks can extract rents from firms. That is, in terms of new investments or debt renegotiations, banks with power can extract more of the expected future profits from the firm (than in a market-based system). Allen and Gale (1995) find that the Anglo-Saxon model of financial intermediation in which financial markets play a dominant role (as in the United States) does not necessarily dominate the German model in which banks dominate. Dow and Gorton (1997) argue that banks and stock markets are alternative institutions for the savings/investment process. Allen and Gale (1999), described earlier, suggest that banks may not be effective gatherers and processors of information in new, uncertain situations. 3

5 information about borrowers. Consequently, our results have important implications for the comparative advantages of banks versus markets as information producers, and for the substitutability or complementarity of markets and banks as monitors and information producers. The remainder of the paper is organized as follows. Section 2 presents a brief review of the prior studies on bank specialness. Section 3 describes the growth of the secondary market for bank loans. Section 4 describes our data and sample selection. Section 5 presents our test hypotheses and methodology. Section 6 summarizes our results and Section 7 concludes. 2. Studies of Bank Specialness As described in Section 1, many theoretical models highlight the unique monitoring functions of banks (e.g., Diamond, 1984; Ramakrishnan and Thakor, 1984; Fama, 1985). These studies generally argue that banks have a comparative advantage, as well as enhanced incentives (relative to public debt holders), in monitoring debt contracts. For example, Diamond (1984) contends that banks have scale economies and comparative cost advantages in information production that enable them to undertake superior debt-related monitoring. Ramakrishnan and Thakor (1984) show that banks as information brokers can improve welfare by minimizing the costs of information production and moral hazard. Fama (1985) argues that banks, as insiders, have superior information due to their access to inside information whereas outside (public) debt holders must rely mostly on publicly available information. The empirical research to date on bank specialness generally finds a robust, favorable, impact of bank loan announcements on borrowers stock returns. This result is in contrast to the insignificant or negative response of investors to the announcement of most other forms of financing, including private placements of debt, straight public debt, preferred stock, convertible debt, convertible preferred, and common stock. Table 1 presents a summary of estimates of loan announcement effects from prior studies. Note that these studies used data from time periods when a well-developed secondary loan market did not exist. 4

6 James (1987) finds a 1.93% two-day ([-1,0]) cumulative abnormal return (z-stat 3.96) around the announcement of 80 commercial bank loans to non-financial firms during Extending James work, Lummer and McConnell (1989) examine 728 loan announcements from 1976 to 1986 and document a 0.61% two-day CAR (z-stat 2.69). They distinguish between new bank loans and loan renewals, and show that only firms renewing a bank loan have significantly positive announcement period excess returns. However, later studies by Best and Zhang (1993), and Billet, Flannery, and Garfinkel (1995) do not find a statistically significant difference between loan initiations and renewals once they control for differences in borrower and lender characteristics, such as precision of analysts forecasts and the credit quality of lenders, suggesting that valuable information is revealed through initial screening by banks as well as in the renewal process. 6 Specifically, Best and Zhang (1993) examine 491 loan agreement announcements from and document a two-day CAR of 0.32% (z-stat 2.31). They argue that if banks produce valuable private information about borrowers, then loan announcements should convey good news to stock market investors only when public information about firm value is noisy or difficult to interpret by outside investors. Consistent with this view, they find that firms experience positive abnormal returns to loan announcements when analysts forecast errors are high, and insignificant abnormal returns when forecast errors are low. Billet, Flannery, and Garfinkel (1995) analyze 626 loan announcements from 1980 to 1989 and document one-day [day 0] abnormal returns of 0.68% (z-stat 4.33). They further show that lenders with a higher credit rating are associated with larger abnormal borrower returns, suggesting that a lender s perceived quality signals valuable information about the firm value of the borrower to uninformed market investors. 5 This compares favorably with similarly measured cumulative abnormal return of -0.11% for public debt, and -0.91% for private placement agreements (primarily with insurance company lenders), both are statistically not different from zero. See James (1987) for more details. 6 James and Smith (2000) suggest that one reason that duplicating the results of Lummer and McConnell has been difficult is that properly classifying loan announcements as initiations or renewals is tricky because many announcements do not make this distinction. Even when new loans are classified correctly as initiations, most firms in this category will have already established a lending relationship with the bank through a previous lending arrangement or unused loan commitment. They further state that it is difficult to imagine that there are many publicly traded firms that do not have some sort of ongoing banking relationship. 5

7 More recently, Billett, Flannery and Garfinkel (2005) re-examine the uniqueness of bank loans from a long-run perspective. They find that in the long-run bank loans are no different (and hence not special) from seasoned equity offerings or public debt issuance in terms of long-run stock returns, operating performance, and earnings announcement returns over the three years following a loan announcement. 3. The growth of the secondary market for loans The secondary market for loans has grown rapidly during the past decade. The market for loans typically includes two broad categories, the first is the primary or syndicated loan market, in which portions of a loan are placed with a number of banks, often in conjunction with, and as part of, the loan origination process (usually referred to as the sale of participations). The second category is the seasoned or secondary loan sales market in which a bank subsequently sells an existing loan (or part of a loan). Banks and other financial institutions have sold loans among themselves for over 100 years. However, this market grew slowly until the early 1980s when it entered a period of spectacular growth, largely due to expansion in highly leveraged transaction (HLT) loans to finance leveraged buyouts (LBOs) and mergers and acquisitions (M&As). With the decline in LBOs and M&As in the late 1980s after the stock market crash of 1987, the volume of loan sales fell to approximately $10 billion in However, since then the volume of loan sales has expanded rapidly, especially as M&A activity picked up again. 7 Figure 1 shows a tenfold growth in the secondary market loan transactions from , with the secondary market loan transactions exceeding $100 billion a year since The secondary loan sales market is sometimes segmented based on the type of investors 7 Specifically M&A activity increased from$190 billion in 1990 to $500 billion in 1995, and to over $1,800 billion in 2000 (Source: Thomson Financial Securities Data Corporation). 8 This ten-fold increase is significantly larger than the growth rates in the corporate bond market over this period. According to Thomson Financial Securities Data, corporate bond issuance grew three-fold from $224.2 billion in 1991 to about $650.4 billion in A comparison with the growth rate in corporate bond trading volumes is not feasible since the data for this period is not publicly available. 6

8 involved on the buy-side, e.g., institutional loan market versus retail loan market. An alternative way of stratifying loan trades in the secondary market is to distinguish between par loans (loans selling at 90% or more of face value) versus distressed loans (loans selling at below 90% of face value). In our empirical analysis, we will distinguish between par and distressed loans. Figure 1 also shows an increasing proportion of loan sales are distressed loans, reaching 42% in However, the level of distressed loan sales has come down subsequently to more moderate levels in 2004 and Data and sample selection The sample period for our study is January 1, 1987 through October 31, Our choice of sample period was primarily driven by data considerations. LPC Dealscan data on loan originations goes back to However, the daily secondary market loan price data, which is used in some of our tests (which we refer to simply as the loan price dataset ) is available only from November 1, For example, we obtain the first date of trading of a loan in the secondary market from the loan price dataset. The loan price dataset is a unique dataset of daily secondary market loan prices from the Loan Syndications and Trading Association (LSTA) and Loan Pricing Corporation (LPC) mark-to-market pricing service, supplied to over 100 institutions managing over $200 billion in bank loan assets. This dataset consists of daily bid and ask price quotes aggregated across dealers. Each loan has a minimum of at least two dealer quotes and a maximum of over 30 dealers, including all top loan broker-dealers. 9 These price quotes are obtained on a daily basis by LSTA in the late afternoon from the dealers and the price quotes reflect the market events for the day. The items in this database include a unique loan identification number (LIN), name of the issuer (Company), type of loan, e.g., term loan (Facility), date of pricing (Pricing Date), average of bid quotes (Avg Bid), number of bid quotes (Bid Quotes), 9 Since LSTA and LPC do not make a market in bank loans and are not directly or indirectly involved the buying or selling of bank loans, the LSTA/LPC mark-to-market pricing service is believed to be independent and objective. 7

9 average of second and third highest bid quote (High Bid Avg), average of ask quotes (Avg Ask), number of ask quotes (Ask Quotes), average of second and third lowest ask quotes (Low Ask Avg), and a type of classification based on the number of quotes received, e.g., Class II if 3 or more bid quotes. We obtained borrower stock returns, and stock index (i.e., NYSE/AMEX/NASDAQ Value-weighted index) returns for computing abnormal returns from the daily stock and indices files of the Center for Research in Securities Prices (CRSP). We take the earliest of the dates for a loan from LPC Dealscan as a proxy for the loan announcement date. That is, Dealscan mentions several different dates, such as active date, signing date, closed date, etc. We verified these loan announcement dates in Factiva for a random sample of 30 loans. Finally, security-specific characteristics and borrower characteristics, such as seniority, and collateral were obtained from the Dealscan database of the Loan Pricing Corporation (LPC), and Compustat database of Standard and Poor s (S&P). 5. Test hypotheses and methodology We start this section with outlining our test hypotheses. We next describe the methodology to test these hypotheses. The empirical results are summarized in Section Test hypotheses We seek to test the following hypotheses regarding bank specialness and substitutability or complementarity of banks relative to markets as monitors and information producers: H1: A secondary market for loans reduces bank specialness. H2: Any loss of bank specialness impacts low-rated borrowers more than high-rated borrowers. H3: Secondary market trading in loans is valuable to the equity investors of a borrower. Hypothesis H1 follows from our reasoning in Section 1 that a secondary market for bank 8

10 loans can potentially reduce the incentives of a bank to monitor its loans to a borrower. The extent of a reduction in monitoring incentives may vary from none to all. For example, the lead bank, which typically holds the largest share of a syndicated loan rarely sells its share of a loan in order to preserve its banking relationship with the borrower and other syndicate members. This view suggests that a lead bank continues to monitor its loans to the borrower even in the presence of an active secondary market for loans. However, at the other extreme, a bank (such as a particular bank in a syndicate) may use the secondary market to sell its lending stake in a borrower. If so, such a loan sale could potentially eliminate that bank s incentives to monitor the borrower. 10 Hypothesis H2 follows from a well-known fact that low-rated borrowers, such as distressed borrowers ex ante are likely to benefit the most from bank monitoring and loan renewals (see Billett, Flannery and Garfinkel (1995)). Consequently, distressed borrowers may lose the most if they fear that there are reduced incentives to monitor due to the existence of a secondary market for loans. In other words, any loss of bank monitoring is likely to affect distressed borrowers much more adversely than non-distressed borrowers. Finally, hypothesis H3 follows from our argument in Section 1 that a secondary market for a borrower s loans provides an additional source of information to the information that may substitute or complement information that is traditionally gathered by banks through monitoring of loans. As a result (after controlling for any liquidity effect resulting from the secondary market trading in loans), we expect a positive borrower stock price reaction when a borrower s loans trade for the very first time as long as the loss of monitoring incentives are not large enough to fully offset the stock price effect attributable to loan trading and its associated information production Test methodology To be consistent with the prior literature on bank loan specialness, we employ the event study methodology as outlined in Mikkelson and Partch (1986) to estimate the impact of 10 See footnote 2 for empirical evidence that supports this view. 9

11 a bank loan announcement on the stock return of the borrowing firm. 11 For every loan, Dealscan mentions several different dates, such as active date, signing date, closed date, etc. We take the earliest of these dates for a loan as a proxy for the loan announcement date. The abnormal returns are computed around the loan announcement date (day 0). The abnormal stock return or prediction error for borrower j over day t is defined as P E jt = R jt ( ˆα j + ˆβ j R mt ), (1) where R jt is the rate of return for the common stock of firm j on day t, and R mt is the rate of return on CRSP s dividend-inclusive value-weighted market index (of NYSE, AMEX, and NASDAQ stocks) on day t. The coefficients ˆα j and ˆβ j are estimated by regressing R jt on R mt for the period [-200,-51], i.e., from 200 trading days before the event date (day 0) to 51 trading days before the event date. The prediction errors are computed for each day in the event period [-50,+30], i.e., that begins 50 trading days before the event date and ends 30 days after the event date. The daily prediction errors are averaged across all firms to produce a daily portfolio average prediction error: AP E t = 1 N N P E jt, (2) j=1 where N is the number of firms in the sample. Tests of statistical significance are based on standardized prediction errors. The standardized prediction error for firm j on day t (SP E jt ) is defined as: where SP E t = P E jt S jt, (3) 11 Our results in this study (not reported here) are qualitatively similar if we use the event study methodology as outlined in Brown and Warner (1985). 10

12 S jt = { V 2 j [ ED + (R mt R m ) 2 ]} 1/2 ED k=1 (R mk R, (4) m ) 2 and V 2 j is the residual variance of the market model regression for firm j in equation (1), ED is the estimation period (150 days) used in the market model regression, R m is the mean market return over the estimation period, i.e., [-200,-51]. The average standardized prediction error for day t is given by: ASP E t = 1 N N SP E jt. (5) j=1 Under the assumption that individual daily prediction errors are distributed normally, SP E jt follows a Student t distribution with ED-2 degrees of freedom. Cumulative abnormal returns (CAR T1,T 2 ) are the sum of the prediction errors for the event window beginning with trading day T 1 and ending with T 2, and are given by: CAR T1,T 2 = 1 N N j=1 T 2 t=t 1 P E jt. (6) The test statistic is distributed asymptotically normal under the null hypothesis that CAR T1,T 2 = 0 and is calculated as follows: Z T1,T 2 = N(ASCAR T1,T 2 ), (7) where the average standardized cumulative abnormal return (ASCAR T1,T 2 ) is given by: ASCAR T1,T 2 = 1 N N T 2 SP E jt /( T 2 T 1 + 1). (8) t=t 1 j=1 In the next section, we present results of econometric tests designed to empirically test the three hypotheses outlined in Section Empirical results 11

13 We start our analysis with documenting borrower stock price reactions (i.e., cumulative abnormal returns) to (a) loan announcements and (b) when a borrower s loans trade on the secondary market for the first time. We contrast this evidence with what is known from the existing literature and what we expect to find based on the test hypotheses from Section 5.1. We present the univariate results of the three hypotheses in Section 6.1. In multivariate regression analysis in Section 6.2, we control for some of the well-known determinants of the borrower stock price reaction from prior literature, such as those based on loan-specific and borrower-specific characteristics. We also construct a proxy for the frequency of trading of loans to better understand whether the information from secondary market loan trading serves as a substitute or as a complement to the information that is traditionally gathered by banks through loan monitoring Univariate results The univariate results of the hypothesis H1, H2 and H3 are discussed in sections 6.1.1, 6.1.2, and respectively. We then proceed to a discuss the results of multivariate regression analysis in Section Does a secondary market for loans reduce bank specialness? Hypothesis H1 suggests that since a secondary market for bank loans can potentially reduce the incentives of a bank to monitor its loans to a borrower, it can result in a possible reduction in bank specialness. In this section, we test this hypothesis in two ways. First, through a comparison of pre-trade versus post-trade announcement effects. Consistent with H1, we expect post-trade announcement effects to be smaller than pre-trade announcement effects for the same borrower. That is, we ask if the announcement effect corresponding to new loan announcements subsequent to the first trading day of a borrower s existing loans is smaller than the announcement effect associated with new loan offerings to the same borrower prior to any of its existing loans trading in the secondary market. In Table 2 (Panel A) we find an average two-day [-1,0] abnormal return of 0.56% (z-stat 12

14 2.45) on the loan announcement date for a sample of 602 post-trade loans (i.e., loans whose announcement dates do not precede the first trading day of loans of the same borrower). 12 Moreover, the two-day [-1,0] post-trade CAR of 0.56% is more than double that of the corresponding estimate of pre-trade loans of 0.21% (see Table 2 Panel B) of the same borrowers. 13 While the post-trade abnormal return of 0.56% is economically larger than the pre-trade abnormal return of 0.21%, the difference is not statistically significant at reasonable levels of significance. 14 Results based on longer event windows yield qualitatively similar results. Second, we test H1 through an examination of announcement effects of traded versus nontraded borrowers. Specifically, from LPC Dealscan, we construct a control group of borrowers whose loans never trade, i.e., those borrowers who are not in our loan price dataset. We refer to these borrowers as non-traded borrowers to distinguish from traded-borrowers (whose loans trade during the sample period, such as those in Table 2). Consistent with hypothesis H1, we expect the announcement effect of loans to be lower for traded borrowers relative to non-traded borrowers. Interestingly, the average two-day [-1,0] abnormal return of 0.28% (z-stat 2.46) from Table 2 (Panel C) for traded borrowers is higher than the corresponding estimate of 0.14% (z-stat 4.02) for non-traded borrowers from Table 3. This difference, however is not statistically significant at any meaningful levels of significance. 15 Results based on [-1,1] yield similar results, whereas the other windows show economically insignificant differences. We interpret a positive post-trade abnormal return that is at least as large as the pretrade abnormal return and an abnormal return of traded borrowers that is at least as large 12 Where a borrower has multiple loans, the first trading day of a borrower s loans equals the minimum of the first trading day of all loans of the same borrower. For example, if a firm has three loans whose first-day of trading as evidenced in the loan price dataset are 1/1/01, 1/2/01, and 1/3/01 respectively, the first trading day of a borrower s loans is determined as the minimum of these dates. 13 Our two-day [-1,0] pre-trade CAR of 0.21% is comparable to the estimate of 0.32% from Best and Zhang (1993), albeit much smaller than estimates from prior studies see Table 1 for a summary of estimates of loan announcement effects from prior studies on bank specialness that use data from the 1970s and 1980s, a time period during which a well-developed secondary market for loans did not exist. 14 In subsequent analysis in Section 6.2, when we control for some of the well-known determinants of the abnormal returns in a regression framework, we do find that post-trade abnormal returns are higher. 15 We revisit this issue in a regression framework in Section 6.2 where we control for some of the well-known determinants of abnormal returns. 13

15 as the abnormal return of non-traded borrowers as evidence of continued specialness of bank loans in the presence of a secondary market for loans. One possible reason for a larger (rather than smaller, as hypothesized) announcement effect for borrowers whose loans trade relative to others is that a secondary market for loans provides additional information on perceived borrower credit worthiness that offsets any loss in bank monitoring and monitoring incentives, an issue that we analyze further in Section 6.3. Alternatively, this could be due to an improvement in the liquidity of a loan resulting in cost savings to borrower s equity holders, an issue that we investigate further in Section Are distressed borrowers adversely affected by a secondary market for loans? We test hypothesis H2 by examining whether a secondary market for trading of loans affects distressed borrower loans more than non-distressed borrower loans. The rationale for this hypothesis is that distressed borrowers are likely to be more sensitive to any loss of bank monitoring incentives that may arise as a result of the secondary market trading of loans. We start our analysis with a sample of traded borrowers. That is, all borrowers in the loan price dataset. Specifically, we segment the sample of 371 traded borrowers into those with distressed loans (loans selling at below 90% of face value) and those with par loans (loans selling at 90% or more of face value). Interestingly, the average stock price reaction for borrowers with distressed loans is 9.21% on the start of secondary market trading (see Panel A of Table 4), which is significantly higher than the 0.34% (see Panel B of Table 4) for borrowers with par loans, and the difference is statistically significant at the 1% level. The above evidence suggests that the availability of trading of loans for the first time appears to be interpreted as good news rather than bad news by equity investors of distressed borrowers, because loans serve as an alternative and additional source of information to the information that is traditionally gathered by banks through loan monitoring (see Section 6.1.3) as well as making a relatively illiquid/low quality asset more liquid. 14

16 Is loan trading valuable to equity investors? In this section, we test hypothesis H3 by examining borrower stock price reaction on the first day of trading of any of its existing loans. Specifically, we conduct an event study around the first trading day of a borrower s previously non-traded loans. We view such an event study as providing direct evidence on whether equity investors interpret the availability of trading in bank loans of a borrower to be good news (or bad news) in the face of reduced monitoring incentives of original lenders and the alternative source of information about the borrower that secondary loan markets provide. Panel C of Table 4 summarizes the results. We find a positive announcement effect of 0.77% (z-stat 2.60) on the first trading day of a borrower s loans for a sample of 371 borrowers. Multi-day, event windows present similar results. Interestingly, the positive announcement effects tabulated here are larger than the announcement effects associated with new loan offerings documented by 3 of the 4 studies in Table 1. That is, the announcement effects associated with the first trading date of a borrower s loans are statistically and economically significant, comparable to new loan announcement effects. The above evidence suggests that equity investors view the availability of trading of the loans of a borrower for the first time as good news, even in the face of reduced monitoring incentives of original lenders. This finding is consistent with a secondary loan market providing information about the borrower that complements information traditionally gathered by banks through monitoring of loans. That is, if the information provided by the secondary loan market merely substitutes (either partially or fully) for the loss of monitoring incentives, then we would not expect to see a positive borrower stock price reaction on the first day of trading of a borrower s existing loans. 16 In summary, based on the univariate results, we find that banks continue to be special even in the presence of a secondary market for bank loans, and that the bank monitoring function and the secondary market for bank loans are complementary rather than substitute 16 In results not reported here, we find that subsequent loans (i.e., loans whose first trading day is greater than the minimum of first trading day of all loans of the same borrower) yield, not surprisingly, statistically insignificant stock price reaction on the first trading day of the subsequent loan. In other words, the market capitalizes the benefits of loan trading the very first time a loan of that borrower is traded. 15

17 sources of information about borrowers. Our analysis so far has not controlled for other well-known determinants of loan-announcement borrower stock price effects from prior studies. We also have not controlled for any favorable liquidity effects resulting from the secondary market trading in loans. We next turn our attention to the results from a multivariate regression analysis where we control for these additional factors Multivariate results The dependent variable is the two-day [-1,0] CAR where day 0 refers to the event date, e.g., a loan announcement date. The independent variables that control for loan-specific and borrower-specific characteristics used in the OLS regressions are: SDPE: Standard deviation of the prediction errors (i.e., borrower s stock return residual) during the estimation period. BETA: Borrower s market model beta calculated over the estimation period. RUNUP: Cumulative return of the borrower s stock during the estimation period. MATURITY: Stands for the maturity of a loan at issuance, measured in years. LN(AMOUNT): Stands for the natural log of one plus the amount of the loan (in $ millions). SENIORITY: A dummy variable that takes a value of one if a loan is senior. SECURED: A dummy variable that takes a value of one if a loan is secured Discussion of variables Several of these variables have been used in previous studies on bank specialness. For example, the shareholders in a riskier firm (as proxied by the standard deviation of the prediction errors during the estimation period) might value a lender s assessment or monitoring ability more highly as compared to that of a less risky firm (see Best and Zhang (1993)). In addition, as suggested by Billett, Flannery and Garfinkel (1995), one could also use the stock beta to capture a borrower s systematic risk, which is distinct from the idiosyncratic 16

18 risk (SDPE). 17 Consequently, the combination of SDPE and BETA can thus be viewed as measuring the total variability of a borrower s stock return. Based on the above reasoning, we expect SDPE and BETA to have a positive coefficient in a regression where the dependent variable is the borrower stock CAR. We also include a stock price runup variable based on Korajczyk, Lucas, and McDonald (1991). Their study shows that firms tend to sell new equity claims following a runup. Hence, if bank loans tend to be similarly announced in the wake of other good news, RUNUP should be inversely related to the extent to which a loan announcement was a surprise. We expect MATURITY to have a negative coefficient since longer-maturity issues are potentially subject to greater interest-rate exposure than shorter-maturity issues, and can have a higher default risk (see, Flannery, (1986)). With respect to LN(AMOUNT), we expect larger loans to have more publicly available information generated about the borrower, and the value of bank monitoring is likely to be less for such borrowers. This predicts an inverse relationship between LN(AMOUNT) and the borrower equity CAR. Finally, the priority structure of a loan, such as seniority and collateral is expected to be positively or negatively related with the borrower s loan announcement effects to the extent a higher priority either enhances or lowers the effectiveness of bank monitoring The secondary loan market and bank specialness (H1) In this section, we conduct two sets of empirical tests for further examining hypothesis H1. First, we interpret the univariate evidence from Section as to whether pre-trade loan announcement effects are larger in terms of abnormal returns than post-trade loan announcement effects by controlling for well-known determinants of borrower stock price reaction associated with loan announcements. Table 5 summarizes the results. The POST TRADE variable (that takes a value of one if a loan announcement date does not precede the minimum of first trading day of any loan of the same firm, and zero otherwise) is positive and statistically significant suggesting 17 We also considered other variables used by them, such as DBIGFIRM, TOBQ, OIBD, LEVERAGE (see pages of Billett, Flannery and Garfinkel (1995) for descriptions of these variables) but did not include them since none of these variables were statistically significant in any of the regressions in their study. 17

19 that post-trade loan announcement effects are significantly larger (rather than smaller) than pre-trade loan announcement effects. 18 Second, in interpreting evidence from Section on whether loan announcement effects of traded borrowers are smaller than those for non-traded borrowers, we control for both wellknown determinants of borrower stock price reaction associated with loan announcements as well as additionally including data from a control group of non-traded borrowers in the regressions. These results are presented in Table 6. In column 1 of Table 6, we run the same specification as the one in column 2 in Table 5 for the augmented dataset. In column 2 we include both the PRE TRADE and POST TRADE variables. The PRE TRADE variable is not statistically significant in any of the regressions in Table 6 19, while the POST TRADE variable is positive and statistically significant at the 5% level. That is, we find that the post-trade abnormal returns are higher than the pre-trade abnormal returns The secondary loan market and distressed borrowers (H2) In this section, as in the previous section, we conduct two sets of multivariate tests to reexamine hypothesis H2. First, in interpreting the univariate evidence from Section 6.1.2, as to whether a secondary market for trading of loans affects distressed borrower loans more than non-distressed borrower loans, we now control for well-known determinants of borrower stock price reaction associated with loan announcements. Table 7 summarizes the results. First, we confirm the result from our univariate analysis in Section that distressed borrowers are not more adversely affected by a secondary market for loans. Specifically, distressed borrowers have significantly higher positive abnormal returns at the time of the first trading of their loans in the secondary market. That 18 The explanatory power of the regressions in Table 5, while low, are comparable to those in Billett, Flannery, and Garfinkel (1995). They document that a loan from a lender of higher credit quality elicits a more positive borrower stock price reaction than that associated with a lender of lower credit quality (Adjusted R 2 are approximately 2% for such regressions see Table IV of their paper). 19 We could not include both variables in Table 5 due to linear dependence. That is, such a PRE TRADE = 1-POST TRADE, and together they will be the same as the intercept in the regression. However, since we augmented non-traded borrower data in Table 6, PRE TRADE and POST TRADE do not sum to one. 18

20 is, when we regress the two-day [-1,0] CAR surrounding the first trading day of loans of a borrower on the DISTRESSED variable (which takes a value of one if a loan is trading at below 90 percent of its face value on the first trading day and zero otherwise) and the control variables described above, the coefficient on the DISTRESSED variable is positive (rather than negative) and statistically significant at the 5% level in both specifications in Table 7. Second, we analyze the extent to which the fact that post-trade loan announcement effects are larger than the pre-trade loan announcement effects found in Table 5 are driven by the reaction of investors to news about loan sales of distressed borrowers. To measure this effect, we interact the POST TRADE variable with the DISTRESSED variable. The regression results are presented in Table 8. As can be seen in the POST TRADE x DISTRESSED variable is positive and statistically significant coefficient (at the 1% level) indicating that a significant component of the post-trade announcement reaction found in Table 5 is due to the announcement of distressed loan trading Complementarity or substitutability of information from loan trading (H3) Our univariate results from Section clearly indicate that the additional information produced by loan trading in the secondary loan market about a borrower is of value to the borrower s equity investors. We next examine whether this information from loan trading in the secondary market for loans complements or substitutes for the information traditionally gathered by banks through monitoring of loans. Towards this objective, we analyze the effect of frequency of secondary market loan trading on abnormal returns. A positive relationship suggests a complementary relationship whereas a zero or negative relationship is consistent with a substitute relationship. A positive relationship suggests that the value of a secondary market in loans to borrower s equity investors increases with the frequency of loan trading which inturn implies a greater level of information production in the secondary loan market and hence is not contingent on the extent of monitoring incentives associated with that borrower s type. 19

21 Since we do not have transaction-level data on loan trading, we follow Lesmond, Ogden, and Trzcinka (1999) and construct a proxy for the frequency of trading (which they refer to as proxy for liquidity in their study) as the percentage of daily returns that are non-zero. Specifically, we examine whether the stock price effects associated with the availability of trading of loans for the first time for a borrower are higher for loans that trade more. Column 1 of Table 9 summarizes the results as to whether the stock price effects associated with loans trading for the first time are higher for loans that trade more frequently (measured ex-post). We find some evidence of the complementary nature of information from secondary loan market trading, i.e., stock price effects are larger for loans that trade more frequently. Specifically, the coefficient of TRADING FREQUENCY variable (the percentage of daily loan returns that are non-zero during the first six months of trading of a loan) is positive and statistically significant at the 10% level. 20 This is consistent with the secondary market for loans providing additional information on the perceived credit worthiness of a borrower which is complementary to the information traditionally gathered by banks through the loan monitoring process Liquidity versus information An alternative explanation for our results (relating to hypothesis H1 and H2), for example, that the post-trade loan announcement effects are larger than the pre-trade loan announcement effects is a potential improvement in liquidity rather than any additional information from secondary market trading of loans. Moreover, the TRADING FREQUENCY variable may proxy for both the information from secondary market loan trading as well as an improvement in liquidity of a loan, such as a lowering of non-informational transaction costs. Since one cannot easily partition the TRADING FREQUENCY variable into two components: a non-informational component (i.e., improved liquidity), and an informational component, we construct a separate proxy for the non-informational component of a transaction cost, namely the ROLL SPREAD (see Roll (1994) and Stoll (2003)). That 20 This measure is highly correlated with LN(AMOUNT). Hence we can only keep one of these variables at a time as an independent variable in a regression. 20

22 is, we estimate ROLL SPREAD as the square root of negative serial covariance of daily loan returns and lagged loan returns during the first six months of trading, measured as a fraction of the average bid-ask spread of the same loan during the same period. Specifically, we augment the regression in Column 1 of Table 9 with the ROLL SPREAD variable. The results are reported in Column 2 of Table 9. The ROLL SPREAD variable has a negative coefficient (statistically significant at the 10% level), consistent with a view that a lowering of non-informational transaction costs due to improved liquidity is associated with an increase in loan announcement abnormal returns. Importantly, the TRADING FRE- QUENCY variable continues to be positive (and statistically significant at the 1% level) suggesting that a secondary market for loans provides additional information on the perceived credit worthiness of a borrower in addition to a potential improvement in liquidity Monitoring incentives and lead lender bank stake In Section 5.1, we argued that the lead lender bank typically holds the largest share of a syndicated loan and rarely sells its share of a loan in order to preserve its banking relationship with the borrower and other syndicate members. Consequently, the lead lender bank continues to monitor its loans to the borrower even in the presence of an active secondary market for loans. In this section, we examine whether there is any empirical evidence supporting this argument. That is, whether a loan where the lead lender bank has a large share of a syndicated loan is associated with a stronger positive loan announcement abnormal returns. We obtain the information on lead lender bank stake from Dealscan. In particular, we take as the lead lender the bank with the largest stake in a loan at the time of offering. We construct a BANK STAKE variable, the maximum lending share of a lending bank in the syndicate at the time of loan origination, and we augment the regressions in Table 5 with the BANK STAKE variable. The results are reported in Table 10. The BANK STAKE variable is positive and statistically significant at the 5% level suggesting that the monitoring incentives are higher when the lead lender has a large stake in a loan. The POST TRADE variable continues to be positive and statistically significant (at the 10% level) consistent 21

23 with a view that banks continue to be special even in the presence of a secondary market for bank loans. 7. Conclusions We find that new loan announcements are associated with a positive announcement effect on the borrower s stock price even when a borrower s loans trade on the secondary market. Interestingly, we find that distressed borrowers, for whom bank monitoring is likely to be most valuable, are not adversely affected by the presence of a secondary market for loans. Moreover, when a borrower s existing loans trade for the first time in the secondary loan market, it elicits a positive announcement effect on the borrower s stock price. Our results are also robust to the effects of increased liquidity of loans as a result of the secondary market for trading in such loans. Overall, we conclude that banks continue to be special even in the presence of a secondary market for bank loans, and that the bank monitoring function and the secondary market for bank loans are complementary rather than substitute sources of information about borrowers. Consequently, our results have implications for the broader debate on the comparative advantages of banks versus markets as information producers, and for the substitutability and complementarity of markets for banks as monitors and information producers. 22

24 References Allen, F., Stock markets and resource allocation. In: C. and X. Vives, eds., Capital Markets and Financial Intermediation, Cambridge University Press, Cambridge, MA, Allen, F., Gale, D., A welfare comparison of intermediaries and financial markets in Germany and the US. European Economic Review, 39, Allen, F., Gale, D., Comparing Financial Systems. MIT Press, Cambridge, MA. Best, R., Zhang, H., Alternative information sources and the information content of bank loans. Journal of Finance 48, Billett, M., Flannery, M., Garfinkel, J., The effect of lender identity on a borrowing firm s equity return. Journal of Finance 50, Billett, M., Flannery, M., Garfinkel, J., Are Bank Loans Special? Evidence on the Post-Announcement Performance of Bank Borrowers. Journal of Financial and Quantitative Analysis, Forthcoming. Brown, S. J., Warner, J. B., Using daily stock returns the case of event studies. Journal of Financial Economics 14, Dahiya, S., Puri, M., Saunders, A., Bank borrowers and loan sales: New evidence on the uniqueness of bank loans. Journal of Business 76(4), Dahiya, S., Saunders, A., Srinivasan, A., Financial distress and bank lending relationships. Journal of Finance 58,

25 Diamond, D. W., Financial intermediation and delegated monitoring. The Review of Economic Studies 51, Dow, J., Gorton, G., Stock market efficiency and economic efficiency: Is there a connection? Journal of Finance 52, Fama, E. F., What s different about banks? Journal of Monetary Economics 15, Flannery, M. J., Asymmetric information and risky debt maturity choice. Journal of Finance 41, James, C. M., Some evidence on the uniqueness of bank loans. Journal of Financial Economics 19, James, C. M., The use of loan sales and standby letters of credit by commercial banks. Journal of Monetary Economics 22, James, C. M., Smith, D. C., Are banks still special? New evidence on their role in the capital-raising process. Journal of Applied Corporate Finance 13, Korajczyk, R., Lucas, D. J., McDonald, R., L., The effect of information releases on the pricing and timing of equity issues. Review of Financial Studies 4, Levine, R., Bank-based or market-based financial systems: Which is better? Journal of Financial Intermediation 11(4), Lesmond, D. A., Ogden, J. P., Trzcinka, C., A., A new estimate of transaction costs. 24

26 Review of Financial Studies 12(5), Lummer, S. L., McConnell, J. J., Further evidence on the bank lending process and the capital-market response to bank loan agreements. Journal of Financial Economics 25, Mikkelson, W., Partch, M., Valuation effects of security offerings and the issuance process. Journal of Financial Economics 15, Pennacchi, G., Loan sales and cost of bank capital. Journal of Finance 43, Ramakrishnan, R., Thakor, A., Information reliability and a theory of financial intermediation. Review of Economic Studies 51, Rajan, R. G., Insiders and outsiders: The choice between informed and arms length debt. Journal of Finance 47(4), Roll, R., A simple implicit measure of the effective bid-ask spread in an efficient market. Journal of Finance 39, Saunders, A., Financial Institutions Management: A Modern Perspective. Irwin Publishers, Fourth edition. Shleifer, A., Vishny, R. W., Large shareholders and corporate control. Journal of Political Economy 96(3), Slovin, M. B., Shuska, M. A., Polonchek, J., A., The value of bank durability: Borrowers as the bank stakeholders. Journal of Finance 48,

27 Stiglitz, J. E., Credit markets and the control of capital. Journal of Money, Credit and Banking 17(2), Stoll, H. R., Market microstructure. In: G. M. Constantinides, M. Harris, R. M. Stulz (Eds.), Handbook of Economics of Finance Vol 1A, , Elsevier B.V., Amsterdam, The Netherlands. White, H., A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48,

28 TABLE 1 Summary of results from prior studies This table summarizes the event study results from prior studies. The Z statistics of Average Cumulative Abnormal Returns (ACARs) in the event window (shown in parentheses) reports by these studies are based on the methodology of Mikkelson and Partch (1989) that considers both the time-series and cross-sectional dependence, and unequal variances in returns. The superscripts for Z statistics a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test. Sample Event Name of the study period window ACAR (%) Z-stat N James (1987) [-1,0 ] a 80 Lummer and McConnell (1989) [-1,0 ] a 728 Best and Zhang (1993) [-1,0 ] b 491 Billett, Flannery and Garfinkel (1995) [0 ] a

29 TABLE 2 Average cumulative abnormal returns to traded borrowers This table presents the average cumulative abnormal return (ACAR) to traded borrowers, i.e., firms receiving loans that have at least one loan traded in the secondary market during the sample period. In other words, all loans in the LPC Dealscan database corresponding to the borrowers in the loan price dataset. We use the earliest date of a loan in the LPC Dealcsan database as a proxy for the announcement date of the firm receiving that loan. Panel A presents the ACAR data for a firm s post-trade loans, i.e., loans whose announcement dates do not precede the minimum of the first trading day of all loans of the same firm. Panel B presents the ACAR for a firm s pre-trade loans, i.e., loans whose announcement dates precede the minimum of the first trading day of all loans of the same firm. Panel C presents the ACAR for all loans, i.e., both post-trade and pre-trade loans combined. The Z statistics of ACARs in the event window (shown in parentheses) are computed using the methodology of Mikkelson and Partch (1989) that considers both the time-series and cross-sectional dependence, and unequal variances in returns. The superscripts for Z statistics a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test. Panel A: Post-trade loans Event window ACAR (%) Z-stat N [0 ] [-1,0 ] b 602 [-1,1 ] b 602 [-2,2 ] c 602 Panel B: Pre-trade loans Event window ACAR (%) Z-stat N [0 ] b 2,338 [-1,0 ] a 2,338 [-1,1 ] a 2,338 [-2,2 ] a 2,338 Panel C: All loans Event window ACAR (%) Z-stat N [0 ] b 2,940 [-1,0 ] b 2,940 [-1,1 ] a 2,940 [-2,2 ] a 2,940 28

30 TABLE 3 Average cumulative abnormal returns to non-traded borrowers This table presents the average cumulative abnormal return (ACAR) to non-traded borrowers, i.e., firms receiving loans that never traded in the secondary market during the sample period. In other words, all loans in the LPC Dealscan database corresponding to the borrowers that are not included in the loan price dataset. We use the earliest date of a loan in the LPC Dealcsan database as a proxy for the announcement date of the firm receiving that loan. The Z statistics of ACARs in the event window (shown in parentheses) are computed using the methodology of Mikkelson and Partch (1989) that considers both the time-series and cross-sectional dependence, and unequal variances in returns. The superscripts for Z statistics a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test. Event window ACAR (%) Z-stat N [0 ] a 13,151 [-1,0 ] a 13,151 [-1,1 ] a 13,151 [-2,2 ] a 13,151 29

31 TABLE 4 Average cumulative abnormal returns surrounding the first trading day of loans for traded borrowers This table presents the average cumulative abnormal return (ACAR) to traded borrowers (i.e., borrowers in the loan price dataset) surrounding the first trading day of a borrower s loans. For borrowers with multiple loans, we use the minimum of the first trading day of all loans of the same borrower. Panel A and Panel B segment the sample into distressed loans, defined as loans that trade at below 90 percent of the face value on the first trading day, and par loans, defined as loans that trade at or above 90 percent of the face value on the first trading day. Panel C uses all loans. The Z statistics of ACARs in the event window (shown in parentheses) are computed using the methodology of Mikkelson and Partch (1989) that considers both the time-series and cross-sectional dependence, and unequal variances in returns. The superscripts for Z statistics a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test. Panel A: Distressed loans Event window ACAR (%) Z-stat N [0 ] a 29 [-1,0 ] a 29 [-1,1 ] a 29 [-2,2 ] c 29 Panel B: Par loans Event window ACAR (%) Z-stat N [0 ] [-1,0 ] [-1,1 ] [-2,2 ] Panel C: All loans Event window ACAR (%) Z-stat N [0 ] a 371 [-1,0 ] b 371 [-1,1 ] c 371 [-2,2 ] b

32 TABLE 5 Regression analysis of determinants of cumulative abnormal returns of firms receiving loans This table presents estimates from a linear regression analysis of the determinants of cumulative abnormal returns (CARs) of loans received by the borrowers in Table 2 as recorded in Dealscan. The dependent variable is the two-day [-1,0] CAR, measured as a percentage. Independent variables used in this table are: POST TRADE takes a value of one if a loan announcement date (proxied in Dealscan by the earliest date for a loan) does not precede the minimum of the first trading day of all loans of the same firm, and zero otherwise. MATURITY stands for the maturity of the loan in years at issuance. AMOUNT refers to the size of the loan, measured in millions of dollars. SENIOR takes a value of one if a loan is classified as senior in Dealscan, and zero otherwise. SECURED takes a value of one if a loan is classified as secured in Dealscan, and zero otherwise. BETA is the market model beta from the estimation window. RUNUP is the cumulative return of the borrower s stock price during the estimation window. SDPE is the standard deviation of the market model prediction error over the estimation period. The t ratios shown in parentheses are adjusted for heteroskedasticity using White s (1980) variance-covariance matrix (a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test). Dependent Variable: CAR[-1,0], % Variable Model 1 Model 2 INTERCEPT (0.59) (1.01) POST TRADE (1.85) c (1.97) b MATURITY (0.42) (0.67) LN(AMOUNT) (-0.23) (-0.43) SENIOR (-1.16) (-1.29) SECURED (0.21) (0.35) BETA 0.04 (0.21) RUNUP (-2.93) a SDPE (-0.70) Observations 2,275 2,275 Adjusted R

33 TABLE 6 Regression analysis of determinants of cumulative abnormal returns of traded and non-traded firms receiving loans This table presents estimates from a linear regression analysis of the determinants of cumulative abnormal returns (CARs) of loans received by the traded and non-traded borrowers in Table 2 as recorded in Dealscan. The dependent variable is the two-day [-1,0] CAR, measured as a percentage. Independent variables used in this table are: PRE TRADE (POST TRADE) takes a value of one if a loan announcement date proxied in Dealscan by the earliest date for a loan precedes (does not precede) the minimum of the first trading day of all loans of the same firm, and zero otherwise. MATURITY stands for the maturity of the loan in years at issuance. AMOUNT refers to the size of the loan, measured in millions of dollars. SENIOR takes a value of one if a loan is classified as senior in Dealscan, and zero otherwise. SECURED takes a value of one if a loan is classified as secured in Dealscan, and zero otherwise. BETA is the market model beta from the estimation window. RUNUP is the cumulative return of the borrower s stock price during the estimation window. SDPE is the standard deviation of the market model prediction error over the estimation period. The t ratios shown in parentheses are adjusted for heteroskedasticity using White s (1980) variance-covariance matrix (a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test). Dependent Variable: CAR[-1,0], % Variable Model 1 Model 2 INTERCEPT (3.11) a (0.03) PRE TRADE 0.69 (1.55) POST TRADE (2.22) b (2.52) b MATURITY (1.54) (1.16) LN(AMOUNT) (0.29) (1.51) SENIOR (-2.43) b (-2.53) b SECURED (-0.56) (-0.22) BETA (-8.19) a (-8.30) a RUNUP (-5.06) a (-5.04) a SDPE (0.21) (0.54) Observations 13,952 13,952 Adjusted R

34 TABLE 7 Regression analysis of determinants of cumulative abnormal returns surrounding the first trading day of loans This table presents estimates from a linear regression analysis of the determinants of cumulative abnormal returns (CARs) surrounding the first trading day of loans. The dependent variable is the two-day [-1,0] CAR, measured as a percentage. Independent variables used in this table are: DISTRESSED takes a value of one if a loan is trading at below 90 percent of its face value. MATURITY stands for the maturity of the loan in years at issuance. AMOUNT refers to the size of the loan, measured in millions of dollars. SENIOR takes a value of one if a loan is classified as senior in Dealscan, and zero otherwise. SECURED takes a value of one if a loan is classified as secured in Dealscan, and zero otherwise. BETA is the market model beta from the estimation window. RUNUP is the cumulative return of the borrower s stock price during the estimation window. SDPE is the standard deviation of the market model prediction error over the estimation period. The t ratios shown in parentheses are adjusted for heteroskedasticity using White s (1980) variance-covariance matrix (a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test). Dependent Variable: CAR[-1,0], % Variable Model 1 Model 2 INTERCEPT (-2.89) a (-2.89) a DISTRESSED (2.10) b (2.07) b MATURITY (-0.65) (-0.47) LN(AMOUNT) (2.84) a (3.08) a SENIOR (1.61) (1.62) SECURED (-1.87) c (-1.86) c BETA (-2.06) b RUNUP (-1.15) SDPE (0.51) Observations Adjusted R

35 TABLE 8 Regression analysis of determinants of cumulative abnormal returns of firms receiving loans segmented further by whether a loan is distressed This table presents estimates from a linear regression analysis of the determinants of cumulative abnormal returns (CARs) of loans received by the borrowers in Table 2 as recorded in Dealscan. The dependent variable is the two-day [-1,0] CAR, measured as a percentage. Independent variables used in this table are: POST TRADE x DISTRESSED is an interactive variable based on POST TRADE and DISTRESSED variables (see Tables 5 and 7 for definitions). MATURITY stands for the maturity of the loan in years at issuance. AMOUNT refers to the size of the loan, measured in millions of dollars. SENIOR takes a value of one if a loan is classified as senior in Dealscan, and zero otherwise. SECURED takes a value of one if a loan is classified as secured in Dealscan, and zero otherwise. BETA is the market model beta from the estimation window. RUNUP is the cumulative return of the borrower s stock price during the estimation window. SDPE is the standard deviation of the market model prediction error over the estimation period. The t ratios are shown in parentheses (a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test). Dependent Variable: CAR[-1,0], % Variable Model 1 Model 2 INTERCEPT (0.95) (1.34) POST TRADE x DISTRESSED (3.37) a (3.35) a MATURITY (-0.45) (-0.27) LN(AMOUNT) (-0.31) (-0.49) a SENIOR (-0.73) (-0.81) SECURED (0.28) (0.45) BETA 0.04 (0.29) RUNUP (-3.94) a SDPE (-1.38) Observations 2,275 2,275 Adjusted R

36 TABLE 9 Regression analysis of determinants of cumulative abnormal returns surrounding the first trading day of loans This table presents estimates from a linear regression analysis of the determinants of cumulative abnormal returns (CARs) surrounding the first trading day of loans. The dependent variable is the two-day [-1,0] CAR, measured as a percentage. Independent variables used in this table are: DISTRESSED takes a value of one if a loan is trading at below 90 percent of its face value. MATURITY stands for the maturity of the loan in years at issuance. SENIOR takes a value of one if a loan is classified as senior in Dealscan, and zero otherwise. SECURED takes a value of one if a loan is classified as secured in Dealscan, and zero otherwise. BETA is the market model beta from the estimation window. RUNUP is the cumulative return of the borrower s stock price during the estimation window. SDPE is the standard deviation of the market model prediction error over the estimation period. TRADING FREQUENCY stands for the percentage of daily loan returns that are non-zero during the first six months of trading of a loan. ROLL SPREAD is the square root of negative serial covariance of daily loan returns and lagged loan returns during the first six months of trading, measured as a fraction of the average bid-ask spread of the same loan during the same period. The t ratios shown in parentheses are adjusted for heteroskedasticity using White s (1980) variance-covariance matrix (a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test). Dependent Variable: CAR[-1,0], % Variable Model 1 Model 2 INTERCEPT (-0.28) (-0.89) DISTRESSED (2.09) b (1.23) MATURITY (-0.92) (0.37) SENIOR (1.74) c (1.41) SECURED (-2.36) b (-1.40) BETA (-1.86) c (-2.10) b RUNUP (-1.06) (-1.20) SDPE (0.04) (0.74) TRADING FREQUENCY (1.85) c (1.99) b ROLL SPREAD (-1.82) c Observations Adjusted R

37 TABLE 10 Regression analysis of determinants of cumulative abnormal returns of firms receiving loans This table presents estimates from a linear regression analysis of the determinants of cumulative abnormal returns (CARs) of loans received by the borrowers in Table 2 as recorded in Dealscan. The dependent variable is the two-day [-1,0] CAR, measured as a percentage. Independent variables used in this table are: POST TRADE takes a value of one if a loan announcement date (proxied in Dealscan by the earliest date for a loan) does not precede the minimum of the first trading day of all loans of the same firm, and zero otherwise. MATURITY stands for the maturity of the loan in years at issuance. AMOUNT refers to the size of the loan, measured in millions of dollars. SENIOR takes a value of one if a loan is classified as senior in Dealscan, and zero otherwise. SECURED takes a value of one if a loan is classified as secured in Dealscan, and zero otherwise. BETA is the market model beta from the estimation window. RUNUP is the cumulative return of the borrower s stock price during the estimation window. SDPE is the standard deviation of the market model prediction error over the estimation period. BANK STAKE is the maximum lending share of a lending bank in the syndicate at the time of loan origination. The t ratios shown in parentheses are adjusted for heteroskedasticity using White s (1980) variance-covariance matrix (a, b, and c stand for significance at the 1%, 5%, and 10% levels using a two-tailed test). Dependent Variable: CAR[-1,0], % Variable Model 1 Model 2 INTERCEPT (-0.44) (-0.05) POST TRADE (1.73) c (1.86) c MATURITY (0.57) (0.84) LN(AMOUNT) (-0.17) (-0.39) SENIOR (-0.93) (-1.04) SECURED (0.17) (0.33) BETA 0.05 (0.26) RUNUP (-3.00) a SDPE (-0.74) BANK STAKE (2.04) b (2.12) b Observations 2,275 2,275 Adjusted R

38 Source: Reuters LPC Traders Survey 37