Are Bank Loans Special Evidence from bank loan announcements

Are Bank Loans Special? Evidence from bank loan announcements Pankaj Kumar Maskara Gatton College of Business & Economics University of Kentucky pkmaskara@uky.edu 615-335-2825 FIRST DRAFT 17 Nov, 2006 ABSTRACT Prior studies have found that borrowing firms, on average, get a bank loan announcement return of 100 to 150 basis points. Despite some conflicting evidence (Peterson and Rajan, 2002; Billett, Flannery and Garfinkel, 2003; Thomas and Wang, 2004), the literature tends to interpret this positive bank loan announcement effect as the market’s response to lower information asymmetry regarding the borrowing firm caused by the certification role of the lending banks as quasi-insiders. We document that there was a strong selection bias in the prior studies. We show that less than a quarter of the loans made by banks are ever announced by the borrowing firm and the loans that are announced are systematically different from the loans that are never explicitly announced by the firms. Constrained firms (defined as firms with lower credit ratings) and distressed firms (defined as firms with low or negative operating earnings yet high interest costs) are more likely to announce their loans. We show that even though earlier there was a positive loan announcement effect as documented in the prior studies, loan announcements did not elicit positive announcement returns in the last ten years. Our evidence suggests that prior studies misinterpreted the easing of the credit constraints of the borrowing firms during a period of scarce credit as evidence in favor of the information asymmetry hypothesis. Are Bank Loans Special? Evidence from bank loan announcements I. Introduction Announcements regarding equity issuance have been known to elicit negative stock returns and those regarding debt issuance elicit zero or slightly negative stock returns. However, bank loan announcements have been shown to have positive announcement effects (Mikkelson and Partch, 1986; James, 1987; Lummer and McConnell, 1989). This has led researchers to believe that the bank loans are somehow special. The literature tends to treat banks as quasi-insiders. The positive bank loan announcement effect is normally justified as the market’s response to lowered information asymmetry regarding the borrowing firm. The act of making a loan by the bank is considered as a certification of the quality of the borrowing firm. Recent studies have questioned this justification of the positive announcement effect (Preece and Mullineaux, 1994; Peterson and Rajan, 2002; Billett, Flannery and Garfinkel, 1995, 2003; Thomas and Wang, 2004). Yet the information asymmetry hypothesis is the most widely accepted explanation for the positive loan announcement effect. In this study, we show that the positive loan announcement effect was a temporal phenomenon and it does not exist any more. Also, we show that prior studies suffered from a selection bias problem. Loans that are announced are systematically different from those that are not announced. Constrained firms (defined as firms with lower credit ratings) and distressed firms (defined as firms with low or negative operating earnings but high interest costs) are more likely to announce their loans than other firms. The positive loan announcement effect observed in prior studies was primarily driven by the reaction of the market to the announcement of loans by constrained and distressed firms. Announcement of a bank loan by such firms signaled easing of credit constraints. In a period of relatively scarce credit, this easing of credit constraints was welcomed by the market and was rewarded with significantly positive stock returns. Prior studies interpreted this positive stock return on loan announcements as an evidence for the information asymmetry hypothesis. They argued that the positive bank loan announcement effect was the market’s response to lower information asymmetry regarding the borrowing firm caused by the certification role of the lending banks as quasi-insiders. Though we do not question the certification role of the banks or the value of relationship lending, we argue that the positive loan announcement effect observed in the prior studies cannot be construed as evidence in support of the value of relationship lending. All the prior studies on loan announcement effect generated a database of bank loan announcements made in the study period and studied the market’s response to the announcements. They performed keyword search in the news databases like WSJ and DJ Newswire searching for keywords like “credit” and “loan”. Unfortunately, this methodology only ensured that loans that were explicitly announced in the media during the study period were captured by their studies. Some of these studies acknowledged that their methodology was exposed to loan reporting bias. As both lenders and borrowers are more likely to announce positive rather than negative information, their studies were more likely to be biased towards higher quality loans. Firms that were able to procure loans at favorable rates were more likely to announce their loans than those firms that had their loans renewed at unfavorable terms or were let to expire. To avoid this selection bias problem we use a different methodology in this paper. Rather than beginning with the loan announcements, we start with the loans themselves and then search for the announcement of randomly selected sample of loans. We find that less than a quarter of all the loans made in the period are ever announced by the company.1 We argue that if there were really a known positive announcement effect of bank loans as warranted by the information asymmetry hypothesis and suggested by prior studies, every firm would have an incentive to announce bank loans. Yet over three quarters of borrowing firms choose not to announce their loans. We show that the loans that are announced are systematically different from those that remain unannounced. Secondly, all the prior studies that found positive loan announcement effect used data from the 70s and the 80s. It needs to be noted that this was a period of relatively scarce credit. After the publication of these studies, several studies showed that the positive loan announcement effect was not universal across different size firms. They showed that the positive effect was limited to small firms and even within small firms, only such firms that announced renewal of loans on favorable terms experienced positive loan announcement returns. Yet this finding was construed as evidence in support of the information asymmetry hypothesis because small firms face higher information 1 We consider a loan to have been announced if any news story in any of the newswires covered by Factiva database explicitly mentions the loan in question. asymmetry than large firms and therefore, benefit more from the certification of quality by the lending banks. However, in this study we show that the positive loan announcement effect does not exist any more. As the reasons that make bank loans special have not disappeared with time over the last ten years, it implies that the positive bank loan announcement effect observed in earlier studies was not due to bank specialness but due to some other reasons. Our study makes the following contribution to the existing literature. It documents that less than 25% of bank loans made in a period are announced by the company and majority of the loans announced by the company are announced within five trading days of the loan start date, especially the day after the loan start date. We introduce a new methodology to account for selection bias in event studies. We show that prior studies suffered from selection bias. Firms that announce loans are systematically different from those firms that do not announce loans. We show that constrained firms, distressed firms, and small firms who take relatively large loans are more likely to announce loans. We also show that the positive loan announcement effect observed in prior studies was limited to their sample period. No such loan announcement effect existed during the last ten years. We call into question the asymmetric information argument as an explanation for positive loan announcement effect observed in earlier studies. We contend that the positive stock return observed earlier was primarily driven by the announcement of loans by distressed and constrained firms. The market reacted favorably to the announcement of such loans because, in a period of relatively scarce credit, the bank loans signaled easing of credit constraints for the borrowing firm. We do not observe any positive loan announcement effect in the last ten years because the credit availability in the last ten years has been relatively very high. The rest of the paper is organized as follows. We review the existing literature in section II of the paper. In Section III we discuss our methodology and data. Section IV discusses our results and tests the robustness of our results. We conclude in section V and elaborate on our calculation of the test statistics in the Appendix section of the paper. II. Literature Review Myers and Majluf (1984) argue that issuance of equity by a firm signals that managers of the firm consider it to be overvalued. Hence, announcement of a seasoned equity offering (SEO) results in negative returns of 2-3% (Asquith and Mullins (1986), and Masulis and Korwar (1986)). Announcements of public bond issues have been shown to generate zero or slightly negative returns by Eckbo (1986), and Howton, Howton, and Perfect (1998). Jensen and Meckling (1976) argue that announcement of debt by firms with free cash flow should result in positive returns because of the accompanying managerial control. Unlike public debt, loans from commercial banks have been shown to generate positive abnormal returns (Mikkelson and Partch (1986), James (1987), and Lummer and McConnell (1989). The positive abnormal returns on announcement of bank loans have been justified using asymmetric information argument and the monitoring benefits offered by a bank. Prior studies argue that banks capture “soft” and private information in their day-to-day dealing with a borrower (Billett, Flannery, and Garfinkel, 1995).2 Commercial banks possess the unique power to provide corporate demand deposit services. They capture valuable information from the deposit accounts of the borrower and have better estimates of the true risk of the borrower. They also have enhanced ability to monitor the borrower (Kane and Malkiel (1965), Black (1975), and Fama (1985)). Puri (1996) analyzed pricing of bank-underwritten securities and investment-house- underwritten securities in the preGlass-Steagall period and found that investors were willing to pay a higher price for securities underwritten by banks rather than investment houses. Their results support a certification role of commercial banks and the inability of investment houses to do so. Additionally, Billett, Flannery and Garfinkel (1995) argue that the market might make a positive inference about future borrower prospects from the fact that loan financing was obtained from a (constrained) commercial bank instead of a less constrained non-bank lender. However, there is some evidence to the contrary that suggests that commercial banks may not have superior private information to begin with. Berger, Rosen, and Udell (2005) state that any disadvantage for a large bank in lending to opaque borrowers using relationship lending is essentially offset by its advantage in transaction technologies. Peterson and Rajan (2002) define soft information as information that is hard to communicate to others, let alone capture in written documents. 2 Peterson and Rajan (2002) state that even if lenders do not have the rich soft information obtained by commercial banks from infrequent, but close, contact with the borrower, far more timely hard information about their creditworthiness is readily available. Their evidence along these lines suggests that commercial banks may not have information advantage over other non-bank financing firms. Billett, Flannery and Garfinkel (2003) find that earnings announcement returns for a borrowing firm are significantly more volatile post loan than pre loan. Also, the standard deviation of the price reactions to earnings announcement by borrowing firms is always higher than that of their peer firms. They interpret this finding as a reduction in earnings transparency and conclude that bank loans do not mitigate asymmetric information problems of the borrower. They report that bank borrowers have significantly lower operating performance as compared to their peers in the year before announcing their loan and this continues for three years after the announcement. Billett, Flannery and Garfinkel (1995), and Preece and Mullineaux (1994) investigate whether lender’s identity influences market’s reaction to a loan announcement and finds that there exists no significant difference between the market’s response to bank and non-bank loans.3 Thomas and Wang (2004) find that special role of bank as corporate quasi-insiders has been eroding and bond market liquidity factors affect bank loan pricing. The literature on bank loan announcement effect started with Mikkelson and Partch (1986). They performed a longitudinal study of 360 firms and analyzed the stock returns Both commercial banks and investment banks are considered banks by Billett, Flannery and Garfinkel (1995). 3 on these firms around announcement of various types of securities offering. They discovered positive stock returns around announcement of bank loans. Their results were then confirmed in a separate study by James (1987). James searched the Wall Street Journal Index for loan announcements by 300 firms over a period of ten years from 19741983. He found that there was positive announcement effect for bank loans and negative announcement effect for issuance of straight debt to repay bank loans. Lummer and McConnell (1989) added a new dimension to bank loan announcement effect and showed that there was positive loan announcement effect for loan renewals but not for loan initiations. However, bank loan announcement studies have been producing inconsistent results after Lummer and McConnell (1989). Slovin et al (1992) did not find any significant difference between the loan announcement returns of loan initiations and loan renewals. Aintablian and Roberts (2000) found positive returns for loan renewals and loan initiations in Canadian firms that are small or have lower credit rating. Slovin et al (1992) found significantly positive loan announcement returns for small capitalization stocks but did not find any such returns for large firms. Preece and Mullineaux (1994) found positive announcement effect even for non-bank loans. The results in aggregate suggest that the positive announcement effect might not after all be due to the asymmetric information but plainly due to the fact that the announcement tells the investors that the firms now have access to funds at a reasonable cost to meet their financial needs and to grow. Theoretical work in the field distinguishes private debt from arms-length borrowing and justifies positive announcement effect of bank loans by viewing institutional lenders as insiders who monitor firm performance and reduce information asymmetry (Fama (1985), Berlin and Loeys (1988), and Kwan and Carleton (1998)). Though syndicated loans do not necessarily fall into either of the two categories- private debt or arms-length borrowing - positive announcement effects have been documented for syndicated loans also, thereby making them comparable to bank loans. Preece and Mullineaux (1996) examine the relation between the number of lenders and market reaction to announcement of syndicated loans.4 They find that only the smallest syndicated group generates a positive and significant return. They document an inverse relationship between number of syndicate members and price reaction. Aintablian and Roberts (2000) find that syndicated loans in Canada result in lower excess returns than non-syndicated loans. They interpret their results as market’s positive reaction to higher contractual flexibility and lower free-rider issue with monitoring in single-bank loans as compared to syndicated loans. Gasbarro, Le, Schwebach, and Zumwalt (2004) find that announcement of syndicated loans elicits positive returns. However, they find that their results are driven by revolving credit agreements rather than term loans which they find to elicit significantly negative returns. Mikkelson and Partch (1986) and Lummer and McConnell (1989) find positive share response to revolving credit agreements but no significant response to term loans. Several articles have shown that market’s reaction to a loan announcement varies with borrower’s characteristics. Slovin, Johnson, and Glascock (1992) find that larger 4 We include all loans in our study, including syndicated loans. borrowers receive smaller announcement returns.5 Best and Zhang (1993) find that firms with negative recent earnings trends or greater market dispersion in expected earnings receive larger bank loan announcement returns. Wansley, Elayan, and Collins (1992) argue that credit announcement effect would be higher for firms that are more difficult to estimate. They find that firms with higher market to book ratio (i.e, more growth options) are associated with slightly larger equity returns. III. Methodology Loan announcement studies thus far have employed relatively similar methodology to arrive at their conclusions. We list the main papers in this literature in Table I and highlight the methodology and sample size of all the papers. It is interesting to note that all the studies perform a keyword search in one or more news databases for bank loan announcements. The downside to this methodology is that only those loans that are explicitly announced in the media are included in their sample. This introduces a loan reporting bias and an information noise bias in their study because borrowers and lenders are more likely to announce positive rather than negative information. Some of these studies have explicitly acknowledged this bias (Mosebach, 1999; Fery et al., 2003). Only Fery et al (2003) make any effort to handle this bias. They state that prior studies are likely to be biased towards higher quality loans and to address this bias they include unpublished loans also in their sample. They distinguish between published and unpublished loans made in Australia and find that market reacts positively to published 5 This is consistent with Fama (1995) who suggests that larger firms operate under the scrutiny of numerous external monitors. loans and unpublished loans fail to elicit any positive returns. Unfortunately, they do not address this issue any further and fail to take their study to any meaningful conclusion. As highlighted in Table I all the prior studies have dealt with a relatively small sample (between 100 and 750 observations) even though the total number of loans made in their sample period was at least 10 times larger.6 We note that Aintablian and Roberts (2000) have 137 observations in their sample as compared to 7500 loans that we have in the Loan Pricing Corporation (LPC) database for the same period. Data Our main source of information on the population of syndicated and non-syndicated loans made to borrowers in US is the Loan Pricing Corporation’s (LPC) Dealscan database. LPC maintains a database with detailed information on bank loans made to borrowers in different parts of the world from 1987 to the present. It contains both price and non-price terms of loans at origination. It also contains borrower specific information like its rating and sales at the time of origination of the loan. The data in Dealscan primarily comes from SEC filings, large loan syndicators, and a staff of reporters. LPC claims that its database contains most of the loans made to large publicly traded companies. It is one of the leading sources of data for research on loans worldwide. We use confirmed data in our study on US borrowers that are not government entities or utility companies between 1987 and 2004 (inclusive). Given that we are interested in the information content of Gasbarro, Le, Schwebach, and Zumwalt (2004) is an exception because they do not delete contaminated announcement and include in their study multiple announcements for the same loan. Yet they have 2061 observations only as compared to 9,669 loans present in the Loan Pricing Corporation’s Dealscan database for the same period. 6 bank loans as reflected in the movement of the stock price of the borrowing firms we delete all such observations for which there is no ticker available. We are left with 28,051 observations. Our database has facility level data for all loans. As we are interested in the announcement effect of the loan deal we need deal level data rather than facility level data. Therefore, we first code each deal as single-facility or a multi-facility deal by creating a dummy variable TRANCH that takes a unit value when the loan is a multifacility deal. After that we trim the data to contain deal level data only. We are thus left with 20,141 loan observations. This constitutes full sample of all loans – single bank loans and syndicated loans - made to public borrowers in US between 1987 and 2004. Table II shows the descriptive statistics of these loans. We then randomly pick 200 loan announcements without replacement from the total population of 20,141 loans. We analyze the characteristics7 of our sample loan and compare it with those of the population and find it representative of the population. We thoroughly search for the announcements of these loans in the Factiva database.8 We do not use a computer program to search for the announcement. We rather search for each announcement manually to minimize measurement error. We read through the text of thousands of news stories. We use possible combinations of the company name, ticker, bank names, loan amount, loan purpose, and any other possible keyword(s) to search for 7 Important characteristics include loan amount, loan maturity, distribution method, number of facilities in the loan, lender identity, borrower rating, borrower sales size, and the origination year of the loan. Factiva collects its information from 10,000 authoritative sources that include exclusive combination of The Wall Street Journal, The Financial Times, Dow Jones and Reuters newswires, and the Associate Press. 8 the loan announcement. Our search window is six months prior to and two months after the loan start date. Given that none of the prior studies have documented the average timing of a loan announcement relative to the loan start date, one of the contributions of our study is that we document the average time lapse between the loan start date and the loan announcement date. We are able to find some form of mention in the media for 57 of the 200 loan announcements in our sample. Of the 57 instances, 5 media reports talked about a borrower either seeking the loan or expecting to receive the loan or a lead bank inviting syndicate members to participate in a loan. The remaining 52 announcements confirmed that the loan was made. Of the 52 announcements, 37 announcements were made by the company and 1 was made by the lending bank. We considered a loan to have been announced by the company when the media report said “in a press release the company said” or “the company announced today”. We also considered an announcement to have been made by the company when the news story had a quote from the company’s top management. The source of information in ten announcements was either reporters or SEC filing and we could not ascertain the source of information in four announcements. We noted that majority of the loan announcements took place on or right after the loan start date. Of the 52 announcements only one announcement took place 17 days before the loan start date. All the other announcements took place within 15 days of the loan start date. Seven of the loan announcements were made on the day of the loan start date and twelve announcements were made the next trading day. Of the 52 announcements 34 were made within five days of the loan start date. Eleven loan announcements were made prior to the loan start date and seven were made after more than 5 days past the loan start date. We also capture the time of the loan announcements. After accounting for announcements that were made after 4:00 PM, 13 of the 52 announcements were effectively made the next trading day. As all the announcements in our sample of 200 firms except for one took place within 15 days of the loan start date we now randomly pick 600 more loans from our population of 20,141 loans and search for their announcements in a narrower window of +/- 15 days from the loan start date. Figure I shows the histogram for all the loan announcements in our sample of 800 loans relative to the loan start date.9 It shows that majority of the loan announcements took place on the day after the loan start day. Over two thirds of the announcements (159 of 232) took place within five days after the loan start day. In Figure II we show the histogram for loan announcements made by the company. The shape of the histogram in figure II is similar to that of figure I except for one major difference. In figure I, 41 of the total of 232 announcements (i.e., 17.6%) took place more than a day before the loan start date whereas only 12 of 168 announcements made by the company (i.e., 7.1%) took place more than a day before the loan start day. This shows that companies announce their loan very close to the loan start date whereas; reporters are more likely to announce loans based on rumor before their actual start date. Mosebach (1999) documents that large loans are usually captured through the Gold Sheets. Large banks inform the reporters at Loan Pricing Corporation of the large loans made during the week ending Thursday evening. Information on these loans is then usually distributed in the market through Gold Sheets. Hence, we are unable to find press release for any major loan above $1 billion. In our conversation with an officer at Loan Pricing Corporation it was confirmed that Gold Sheets capture very large and Mid cap loans only (usually over $500 million). 9 We also keep track of all news stories in the media regarding the borrowing firms in our sample around the loan start date and the loan announcement date. We find that 100 of the 232 loan announcements had contaminating news within three days of the loan announcement day. We consider the following news as contaminating: 1) ratings initiation, downgrade or upgrade, 2) Buyback of shares, 3) Creation of a subsidiary 4) Bagging or loosing a big order, 5) New exchange listing of a subsidiary or options/IPO, 6) Earnings / dividends announcements, 7) Union strike or failed union renegotiations, 8) Acquisition, spinoff, or tender offer, 8) Filing of major lawsuit or settlement, 9) Sale of a division, 10) Growth/expansion in a new market or introduction of a new product, 11) Appointment or resignation of a board member or senior management, 12) Announcement of other loans or securities. We keep track of all the important news stories in the media around the loan start date for all firms in our sample irrespective of whether the borrowing company made an announcement or not. We consider news to be contaminating if a material news story was published in the media within three days of the loan start date. We then get return data on the borrowing firms and the market from CRSP. Of the 800 loans we are able to find return data on the borrowing firms of 741 loans. We get the returns of the borrowing firms and the market on the announcement day and +/-1 trading day of the announcement day. When the announcement was made on a non-trading day, we consider the announcement to have been made on the next trading day. We also get the returns of the borrowing firms and the market on the loan start date and +/-1 trading day of the loan start date. For the purposes of calculating the t-stat we also get the returns of the borrowing firm and the market for a year ending thirty days before the loan start date. We then get financial data on the borrowing firm from Compustat. We get financial data for the year prior to the year in which the loan was made. We get the following data: 1) Total assets (DATA6), 2) Total long term debt (DATA9), 3) Operating income before depreciation (DATA13), 4) Interest expense (DATA15), 5) Price – calendar year – close (DATA24), 6) Common shares outstanding (DATA25). We are able to get financial data for the borrower of 735 loans out of the 800 loans in the sample. IV. Results To show that earlier studies suffered from a selection bias problem we first need to show that firms who announce loans are systematically different from those firms that do not announce loans. We therefore divide all of our sample firms based on their senior debt rating. We have rating information on the borrowers of 342 of the 800 loans in our sample. We find that in our sample none of the firms rated AA or higher announced their loan. Less than 20% of the firms rated BBB or higher announced their loans but over 40% of the loan made to firms rated BB or lower were announced (see Table III). This indicates that firms that are otherwise constrained because they are perceived to be less creditworthy consider their being able to get a bank loan as an important event and therefore, announce their bank loans. Similarly, in Table IV we show that firms that either had negative or zero operating earnings in the prior year were almost twice as likely to announce their loans as compared to other firms. Firms with lower or negative operating earnings are considered to be distressed and such firms find it difficult to prove their capability to repay loans. Therefore, when such firms are able to get bank loans, they are more likely to announce. Similarly, firms that had positive interest expense in the prior year but had negative earnings find it the toughest to raise new loans. Therefore, such firms are most likely to announce their loan when they are able to get one. As we see in Table IV only 25% of the firms with Interest to EBITDA ratio of less than 0.1 announced their loans but over 40% of the firms with negative EBITDA but positive interest expense announced their loans. Earlier studies have shown that small firms experience higher stock returns on bank loan announcement as compared to large firms. We show in Table V that small firms are also the most likely to announce their bank loans as compared to large firms. We show that 86% of the companies with a market capitalization of less than $100 million announced their loans but only 20% of the companies with market capitalization of over $10 billion announced their loans. A proponent of the information asymmetry hypothesis is likely to argue that small firms have least number of analysts following the firm and small firms are most likely to suffer from information asymmetry and therefore, they are also likely to benefit from lowered information asymmetry on announcement of bank loans. However, we argue that small firms are more likely to be constrained and distressed and are more likely to find it difficult to raise credit especially during periods of scarce credit availability. Hence, when they are able to raise credit it signals easing of credit constraints for such firms and thereby an announcement of a bank loan elicits positive stock return for such firms. Similarly, we show in Table VI that 48% of the firms with loan size to assets ratio of over 50 percent announce their bank loan as compared to 8% of the firms whose loan size to assets ratio is less than 5 percent. This implies that firms announce their loans when such loans signal availability of funds to meet growth needs. We use two measures of abnormal returns to capture the announcement effect of bank loans. The first measure of return that we report is the holding period return on the stock for the event date. This measure of abnormal return assumes that the expected return for a stock on a given day is zero. The second measure of abnormal return that we use in our study is the market adjusted return. This measure assumes that the return on the market on any given day is an unbiased estimate for the average return on a sample of firms. We calculate the abnormal return on the sample of firms as the average of the difference between the realized return on the sample of firms and the market on the given day. We use the value weighted return (including disbursements) on CRSP stocks as a proxy for the market. We report the average abnormal return of the borrowing firm on the event date in Table VII. We find that over our sample period the average abnormal return on the announcement day for our sample of borrowing firms was not significantly different from zero. However, pre 1995 the average holding period return on the announcement day for the sample of borrowing firms who did not have any other contaminating news in the media within three days of the announcement day was 132 basis points. The average market adjusted abnormal return on the announcement day for such firms was 113 basis points. These results are in conformity with the results of prior studies. However, post 1995, the average holding period return and the average market adjusted abnormal return on the announcement day for the sample of borrowing firms were not significantly different from zero. Also, we show in Table VIII that pre 1995 majority of the abnormal returns on bank loan announcement was due to distressed and constrained firms. Though our sample size for distressed and constrained firms in Table VIII is small, the message is very clear. Distressed and constrained firms are most likely to announce loans and announcement by such firms yielded higher returns during periods of scarce credit availability. As noted in Brown and Warner (1980) the standard deviation of returns is higher during the announcement periods. Therefore, we use the standard deviation of the daily abnormal return for the sample firms for the year ending thirty days before the announcement date. We elaborate on our methodology in the appendix. This method of calculating the test statistic assumes cross section independence of the excess returns. In order to account for the possible cross-sectional dependence we use the “portfolio” methodology of Brown and Warner (1985). We elaborate on this methodology also in the appendix. Our results (unreported) remain unchanged on using the portfolio methodology of test statistic calculation. Robustness check Prior studies have used an event window rather than an event date to measure the abnormal returns for the firms announcing their bank loans. Therefore, as a robustness test and for the sake of comparability we also calculate the abnormal returns for our sample firms over an event window. We calculate abnormal returns for our sample firms over the following three different event windows: (-1,0) days, (0,1) days, and (-1,1) days of the announcement date. Our results remain unchanged on using these different event windows. In table VIII we report the three day event window abnormal returns for our sample firms. For brevity, we do not report the results for the two-day event windows. Measurement of abnormal return requires that we calculate the difference between the realized return and the expected return. Even though realized returns can be measured precisely, expected returns cannot be. In our measures of abnormal return we assumed the expected return on a stock to be zero when we used holding period return as a measure of abnormal return and we assumed the expected return to be the market return when we used market adjusted return as a measure of abnormal return. These assumptions are subject to argument. Therefore, as a robustness test we use the beta excess return and the standard deviation excess return from the CRSP Eventus database as two additional measures of abnormal return. Beta excess return of a firm is the difference between the daily holding period return on the stock and the daily holding period return on a portfolio of stocks with the same beta measure as the stock. This measure adjusts for the systematic risk of the stock. Standard deviation excess return is the difference between the daily return on the stock and the daily return on a portfolio of stock with similar standard deviation of returns as the stock. This measure adjusts for the return volatility of the stock. We report the abnormal returns for the sample of borrowing firms in Table VII along with our other measures of abnormal return. The magnitude of average beta excess returns and average standard deviation excess returns for our sample firms is the same as that of their average holding period return and average market adjusted return. This suggests that our measure of abnormal returns is robust to alternative specification. We segregate our sample period into two sub periods in table VII to show that the positive loan announcement effect was a temporal phenomenon. We use the year 1995 as our cutoff year. As we contend that the positive loan announcement effect has disappeared over time due to the relative abundance of credit availability in the recent years, our results should not be highly sensitive to the choice of the cutoff year. We therefore perform our analysis using different years like 1993 and 1994 as the cutoff year and our results remain qualitatively unchanged (results unreported). Earlier studies have distinguished between syndicated loans and non-syndicated loans and have found that the positive loan announcement effect is lower for syndicated loans. It is possible that our results are driven primarily because of an increase in syndicate lending over the last 10 years. Therefore, we perform our analysis separately for syndicated loans and non-syndicated loans. The results are reported in Table IX. We note that even though pre 1995 syndicated loans earned lower announcement returns than nonsyndicated loans, post 1995 syndicated loans do not fare any worse than single lender loans. More importantly, our analysis shows that our results are not being driven by announcement of syndicated loans. Similarly, earlier studies have found that small firms experience positive loan announcement effect but large firms do not. Hence, it is possible that our results for the last ten years are being driven by large firms who are known not to elicit positive announcement effect. Therefore, we perform our analysis on small firms only but we find that small firms did not experience positive announcement returns in the last ten years. To attribute the returns of a stock on a particular day or over a particular window to a particular event it is important to ensure that no other important event occurred close to the event being studied. Therefore, in event studies, all observations that have other contaminating events take place close to the event in question are deleted from the sample. This decreases the number of observations in the study and reduces the power of the study. A researcher balances the cost of deleting observations to the benefits of having a clean sample. We deleted all observations that had any contaminating news published in the media within three days of the loan announcement. We test the sensitivity of our results to our selection of this event window of +/-3 days in categorization of news stories as contaminating. We now deem a news story to be contaminating only when it is published (-3,+1) days of the loan announcement. This new scheme of categorization does not alter our results. Our results remain unchanged even when we use (-2, +1) days scheme. Mosebach (1999) states that firms that take very large lines of credit (defined as lines for over $1 billion) do not announce their loans because the information of their lines of credit becomes readily available to the market via Gold Sheets. It is possible that over a period of time, due to improvements in communication technologies, the speed of information diffusion has increased to an extent that firms no longer consider it important to announce their loans. It is possible that the market becomes aware of the new loan as soon as the loan is closed. We therefore use the loan start date as the event date and calculate the abnormal returns for all firms in our sample that did not announce their loans. If the information asymmetry hypothesis and the hypothesis regarding the increased speed of information diffusion are true then we should observe positive return on the event date for the sample firms that did not have any other contaminating event take place within three days of the loan start date. However, we find no such positive return. V. Conclusion Prior studies have shown a positive bank loan announcement effect on the market value of the equity of the borrowing firm. However, these studies have relied on the data compiled by the researchers by performing keyword searches for loan announcements in news databases like the WSJI and DJ Newswire. We use an innovative methodology to show that even though firms that announced their bank loans did earn positive announcement returns in the early days there was no positive loan announcement effect in the last ten years. We also document that all firms do not announce their bank loans. This observation is completely at odds with existing literature that argues that bank loan announcements decrease information asymmetry problems and the banks reveal their certification of the value of the firm by giving them loans. This is because if all bank loans alleviated information asymmetry problems then all firms would have an incentive to announce their loans. Yet less than 25% of the firms choose to do so. We also document that majority of the companies that do announce their loans do so within a week after the loan start date. We show that earlier studies suffered from a major selection bias problem. We show that constrained and distressed firms are more likely to announce their bank loans and during periods of relatively scarce credit, announcement of bank loans by such firms is likely to elicit higher stock returns than other firms. This is because availability of credit to a distressed firm signals easing of credit constraints for such firm. The market rewards this easing of constraints and the availability of funds for future growth by bidding up the price of the borrowing firm thereby resulting positive stock returns on loan announcement. We argue that prior studies misinterpreted this response of the market as the market’s response to alleviated information asymmetry problem. Appendix AI Assuming cross-sectional independence of excess returns we calculate our test statistics as follows: Each excess return Ai,t is first divided by its estimated standard deviation to yield a standardized excess return, A’i,t: A't = Ai ,t / S ( Ai ,t ), i, Where S ( Ai ,t ) = t = −30 ^ t = −281 ( Ai ,t − Ai* ) 2 251 1 t = −30 A = Ai ,t 252 t = −281 * i The test statistic for t=0 is given by Nt i =1 A ' i ,t .( N t ) − 1 2 Where Nt is the number of sample securities at day t. By using the time-series of average excess returns (i.e., ‘portfolio’ excess return), the test statistic calculated as follows takes into account cross sectional dependence in the security-specific excess returns (Brown & warner, 1985). The test-statistics is equal to ˆ A / S ( At ) , where, Ai ,t = Ri ,t − E ( R i ,t ) At = 1 Nt Nt i =1 Ai ,t , t = −30 ˆ S ( At ) = A= t = −281 (A − A) t 2 251 1 t = −30 At 252 t = −281 Figure I All loan announcements 60 # of announcements 50 40 30 20 10 0 <-1 -1 0 1 2 3 4 5 6 7 >7 # of business days relative to loan start date Figure II Announcements made by the company 60 # of announcements 50 40 30 20 10 0 <-1 -1 0 1 2 3 4 5 6 7 >7 # of business days after loan start date Paper Journal of Financial Economics (1986) Journal of Financial Economics (1987) Journal of Financial Economics (1989) Journal of Banking and Finance (1992) Journal of Finance (1993) Journal of Financial Services Research (1994) Applied Financial Economics (1995) Journal of Finance (1995) Journal of Banking and Finance (2000) Quarterly Review of Economics and Finance (2003) Author Mikkelson and Partch Time Source Method Initial sample 360 Final sample Clean Sample James Lummer and McConnell Slovin, Johnson, and Glascock Best and Zhang Preece and Mullineaux 1974 1983 1976 1986 1980 1986 Jan 77Dec 89 Jan 80 - Dec 87 Jan 88 - Dec 91 1980 1989 1988 1995 Jan 83 - Dec 99 Wall Street Journal Index Wall Street Journal Index Dow Jones News Wire Wall Street Journal (WSJ) Search WSJI for all debt announcement for random sample of 300 firms Search for credit agreements in the WSJ Search for credit agreements in DJ Newswire Search for bank loan announcements in WSJ Search for credit agreements in WSJ Keyword search for syndicated loans in UK Keyword search for credit and loans keyword search for credit and loans by Canadian Firms Search for Australian firms reaching credit agreements 1145 676 117 728 273 491 117 728 273 491 * Wall Street Journal (WSJ) International Financing Review (IFR), Euroweek, Screen Insider Dow Jones News Retrieval Service Canadian Newswire, Canadian Corporate News, and Financial Post Database IFR Platinum Database of Thomson Financial Publishing 439 439 Armitage Billett, Flannery and Garfinkel Aintablian and Roberts Gasbarro, Fery, and Woodliff, and Zumwalt 659 574 1468 430 626 137 137 *** 196 Gasbarro, Song-Le, IFR Platinum Database keyword search for Journal of Financial and Schwebach, 1995 and Dow Jones Interactive "Launched", "Sold", Research (2004) and Zumwalt 2000 Index "Issued", or "Priced" * 207 Total financing announcements including straight debt ** This includes multiple announcements and they do not screen for contamination *** We have record of 7500 loans made during this time period **** we have 9669 records 2061 ** **** Table II Variable Ratings lenders Tenor d_amt AIS Sales Syndicated Tranch Label Population Descriptive Statistics N Mean Std Dev Senior debt rating 6,949 5.51 1.3 Number of Lenders 20,127 5.46 7.7 Tenor 18,011 44.78 34 amount of the loan deal 20,140 273 708 (millions) All-in-spread drawn Borrower's sale in the prior year (millions) Dummy variable =1 for syndicated loans Multifacility Loan dummy Minimum 1 1 1 0.05 -14 0 - Maximum 9 110 366 25,000 1,490 273,834 1 1 16,234 18,132 20,127 20,140 199.68 2,278 0.59 0.28 137 8,994 0.49 0.4 Variable Ratings lenders Tenor d_amt AIS Sales Syndicated Tranch Sample Descriptive Statistics Label N Mean Std. Dev. Senior debt rating 342 5.69 1.18 Number of Lenders 800 6.15 8.83 Tenor 715 41.46 32.72 amount of the loan deal 800 309.6 719.85 (millions) All-in-spread drawn Borrower's sale in the prior year (millions) Dummy variable =1 for syndicated loans Multifacility Loan dummy Minimum 3 1 0 0.2 6.32 1.22 0 0 Maximum 9 108 361 12,000 980 186,763 1 1 676 741 800 800 182 3,229 0.63 0.268 129.38 10,431 0.48 0.44 Table III Panel A Rating (Senior) AAA AA A BBB BB B CCC Unrated Total Total 1 12 80 106 80 55 8 458 800 Announced 0 0 9 20 32 28 6 137 232 Percent 0% 0% 11% 19% 40% 51% 75% 30% Company 0 0 4 19 18 14 2 111 168 Percent 0% 0% 5% 18% 23% 25% 25% 24% Panel B EBITDA / TA <=0 0=.20 No Data Total Total 107 161 186 162 119 65 800 Announced 44 47 56 42 26 17 232 Percent 41% 29% 30% 26% 22% 26% Company 40 29 40 30 16 13 168 percent 37% 18% 22% 19% 13% 20% Table IV Panel A Interest / EBITDA 0<=x<.10 .10<=x<.25 .25<=x<.50 x>=.50 +ve Interest but – ve or ‘0’ income No data Total 231 222 116 64 97 70 Announced company Others Total 40 17 57 37 21 58 22 12 34 17 6 23 38 14 3 5 41 19 Panel B Interest / EBITDA 0<=x<.10 .10<=x<.25 .25<=x<.50 x>=.50 +ve Interest but ve or '0' income No data Ann/Tot 25% 26% 29% 36% 42% 27% Comp/Ann 70% 64% 65% 74% 93% 74% Oth/Ann 30% 36% 35% 26% 7% 26% Table V Panel A Market Cap <=100M 100M10B No price Total Total 168 127 149 185 67 104 800 Announced company Others Total 51 41 39 20 1 16 168 8 11 16 18 4 7 64 59 52 55 38 5 23 232 Panel B As a percent <=100M 100M10B No price Ann/Tot 35% 41% 37% 21% 7% 22% Comp/Tot 86% 79% 71% 53% 20% 70% Oth/Tot 14% 21% 29% 47% 80% 30% Table VI Panel A Loan size to Asset size Announced Total 111 100 232 158 134 65 Company 5 11 44 48 47 13 Others 4 8 18 13 17 4 Total 9 19 62 61 64 17 x<=0.05 0.050.5 No data Panel B Loan size to Asset size x<=0.05 0.050.5 No data Ann/Tot 8% 19% 27% 39% 48% 26% Com/Ann 56% 58% 71% 79% 73% 76% Oth/Ann 44% 42% 29% 21% 27% 24% Table VII Panel A – Full sample period HPR Market Adj. Sigma Adj Beta Adj N 94 94 53 55 Mean -0.06% -0.39% -0.07% 0.05% Panel B – Pre 95 Abnormal Returns HPR Market Adj. Sigma Adj. Beta Adj. N 21 21 11 11 Mean 1.32% 1.13% 1.16% 1.35% * * Panel C – Post 95 Abnormal Returns HPR Market Adj. Sigma Adj. Beta Adj. N 73 73 42 44 Mean -0.46% -0.82% -0.39% -0.27% * significant at 5% level Table VIII Pre 95 Not distressed distressed N Mean N Mean 15 2.39% 5 6.13% 1.86% 5 5.22% 15 15 15 0.47% 0.35% 5 5 4.04% 3.68% 3-day CAR 3-day Mkt Adj. CAR one-day Abnormal return one-day mkt adj. ret 3-day CAR 3-day Mkt Adj. CAR one-day Abnormal return one-day mkt adj. ret N 45 45 45 45 Post 95 Mean N -0.97% 25 -1.02% 25 -0.09% -0.55% 25 25 Mean 0.81% 1.05% -0.89% -1.07% Table IX Non-syndicated Loans Pre-95 Post 95 (N=9) (N=26) total (N=35) One-day HPR One-day Mkt. Adj. 3-day CAR 3-day Mkt. Adj. 2.7% 2.5% 5.2% 4.4% -1.2% -1.5% 0.4% -0.2% -0.2% -0.5% 1.6% 1.0% syndicated Loans Pre-95 Post 95 (N=12) (N=47) total (N=59) One-day HPR One-day Mkt. Adj. 3-day CAR 3-day Mkt. Adj. 0.3% 0.1% 2.1% 1.6% 0.0% -0.5% -0.7% -0.4% 0.0% -0.3% -0.1% 0.0% REFERENCES Antablian, S., and G. S. Roberts, 2000, A Note on Market Response to Corporate Loan Announcements in Canada, Journal of Banking and Finance, 24, 381-393. Asquith, P., and D. Mullins, 1986, Equity Issues and Offering Dilution, Journal of Financial Economics, 15, 1/2, 61-89. Berger, A. N., Rosen, R. J., and G. F. Udell, 2005, Does Market Size Structure Affect Competition? The Case of Small Business Lending. Working Paper. Berlin, M., and J. Loeys, 1988, Bond Covenants and Delegated Monitoring, Journal of Finance, 43, 2, 397-412. Best, R., and H. Zhang, 1993, Alternative Information Sources and the Information Content of Bank Loans, The Journal of Finance, 48, 4, 1507-1522. Billett, M. T., Flannery, M. J., and J. A. Garfinkel, 1995, The Effect of Lender Identity on a Borrowing Firm’s Equity Return, The Journal of Finance, 50, 2, 699-718. Billett, M. T., Flannery, M. J., and J. A. Garfinkel, 2003, Are Bank Loans Special? Evidence on the Post-Announcement Performance of Bank Borrowers, Working Paper. Black, Fisher, 1975, Banks Funds Management in an Efficient Market, Journal of Financial Economics, 2, 323-340. Brown, S. J., and J. B. Warner, 1980, Measuring Security Price Performance, Journal of Financial Economics, 8, 205-258. Brown, S. J., and J. B. Warner, 1985, Using Daily Stock Returns: The Case of Event Studies, Journal of Financial Economics, 14, 3-31. Eckbo, Espen, 1986, Valuation Effects of Corporate Debt Offerings, Journal of Financial Economics, 15, 1/2, 119-152. Fama, Eugene, 1985, What’s Different About Banks, Journal of Monetary Economics, 15, 29-39. Fery, J., Gasbarro, D., Woodliff, D. R., and Zumwalt, J. K.,, 2003, Market Reaction to Published and Non-published Corporate Loan Announcements, Quarterly Journal of Economics and Finance, 43, 1-10. Gasbarro, D., Le, K. S., Schwebach, R. G., and J. K. Zumwalt, 2004, Syndicated Loan Announcements and Borrower Value, The Journal of Financial Research, 27, 1, 133-141. Harjoto, M., Mullineaux, D. J., and H. C. Yi, 2005, Loan Pricing at Investment versus Commercial Banks, University of Kentucky Working Paper. Howton, S. D., Howton, S. W., and S. B. Perfect, 1998, The Market Reaction to Straight Debt Issues: The Effects of Free Cash Flow, Journal of Financial Research, 21, 219-228. James, Chirstopher, 1987, Some Evidence on the Uniqueness of Bank Loans, Journal of Financial Economics, 19, 2, 217-236. Jensen, M. C. and W. H. Meckling, 1976, Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure, Journal of Financial Economics, 3, 305-360. Kane, E., and B. G. Malkiel, 1965, Bank Portfolio Allocation, Deposit Variability and the Availability Doctrine, Quarterly Journal of Economics, 79, 113-134. Kwan, S., and W. Carlton, 1998, Financial Contracting and the Choice Between Private Placement and Publicly Offered Bonds, Federal Reserve Bank of San Francisco Working Paper. Lummer, S., and J. McConnell, 1989, Further Evidence on the Bank Lending Process and the Capital Market Response to Bank Loan Agreements, Journal of Financial Economics, 25, 1, 99-122. Masulis, R. and A. Korwar, 1986, Seasoned Equity Offerings: An Empirical Investigation, Journal of Financial Economics, 15, 1/2, 91-118. Mikkelson, W., and M. Partch, 1986, Valuation Effects of Security Offerings and The Issuance Process, Journal of Financial Economics, 15, 1/2, 31-60. Mosebach, Michael, 1999, Market Reponse to Banks Granting Lines of Credit, Journal of Banking and Finance, 23, 1707-1723. Myers, S., and N. Majluf, 1984, Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have, Journal of Financial Economics, 13, 2, 187-221. Peterson, M. A., and R. G. Rajan, 2002, Does Distance Still Matter? The Information Revolution in Small Business Lending, The Journal of Finance, 57, 6, 2533-2570. Preece, D., and Mullineaux, D. J., 1994, Monitoring by Financial Intermediaries: Banks vs. Nonbanks, Journal of Financial Services Research, 4, 191-200. Preece, D., and D. J. Mullineaux, 1996, Monitoring, Loan Renegotiability, and Firm Value: The Role of Lending Syndicates, Journal of Banking and Finance, 20, 577-593. Puri, Manju, 1996, Commercial Banks in Investment Banking. Conflict of Interest or Certification Role? Journal of Financial Economics, 40, 373-401. Slovin, M. B., Johnson, S. A. and J. L. Glascock, 1992, Firm Size and the Information Content of Bank Loan Announcements, Journal of Banking and Finance, 16, 1057-1071. Thomas, H., and Z. Wang, 2004, The Integration of Bank Syndicated Loan and Junk Bond Markets, Journal of Banking and Finance, 28, 299-329. Wansley, J. W., F. A. Elayan, and M. C. Collins, 1992, Investment Opportunities and Firm Quality: An Empirical Investigation of the Information in Bank Lines of Credit, Working Paper, University of Tennessee.