JSTOR

I.  Introduction

The secondary loan market for loans 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 sale market in which a bank subsequently sells off an existing loan (or part of a loan). While a number of reasons for seasoned loan sales have been identified in the previous literature, there have been no empirical studies of the effects of such sales on the returns of those borrowers whose loans are sold or on the banks selling the loans.1

Conventional wisdom has long held that loan sales by banks—especially loans of customers who have established long‐term customer relationships with that bank—would have a negative information effect regarding the borrowing firm. This effect would result from the special, or unique, role of banks as “insiders” to the borrowing firms, such that a decision to sell a customer’s loan would be taken as revealing to the market hitherto private (negative) information regarding a borrower’s financial condition.2 Indeed, while the effect of loan sales on borrowers has been untested prior to this article, such an effect might be expected given the findings of James (1987), Lummer and McConnell (1989), Best and Zhang (1993), and Billett, Flannery and Garfinkel (1995), among others, that new loans and loan renewals carry (positive) private information to the outside equity market about a borrowing firm’s financial condition.3 In a recent paper, James and Smith (2000) provide a comprehensive review of the past and recent research on the special nature of bank loan financing. Overall, they show that research to date finds a robust, favorable impact of bank loan announcements on borrowers’ stock returns in contrast with the insignificant or negative response of investors to the announcement of most other forms of new security issuance (e.g., public debt and equity).

While the positive impact of news announcing the formation of a bank lending relationship is well established, there is a paucity of studies examining the impact of the termination (or reduction) of a banking relationship on a borrower’s stock returns and the borrowers' long‐run performance in the postsale period.4 In this article we employ a previously unutilized information source to identify loan sale events. In particular, we focus on sales of seasoned subpar loans, where the information effects of bank sales are likely to be highest, and test the effects of such sales on borrowers’ returns. We conducted two set of tests. First, we tested for and find a significant negative impact of loan sale announcements on borrowers' stock returns, which complements the established finding that the announcement of new lending relationships (or their continuation) has a positive effect on a borrower’s stock returns. Our finding is both consistent with and extends the existing literature on this dimension. Second, we examined whether the negative information contained in the loan sale announcement is validated by the long‐term performance of firms whose loans are sold. We find that almost half of those firms whose loans are sold file for bankruptcy within 3 years of the loan sale announcement. Interestingly, these firms are not the worst‐performing firms in their industry at the time of the loan sale, suggesting that publicly available information alone might have been insufficient for outside investors to preidentify the degree of weakness of the firms whose loans were sold and that the bank’s loan sale announcement contained valuable negative private information.

We are also interested in the factors that influence a bank’s decision to sell its loans. One possibility identified in the theoretical literature is that bank loan sales are motivated by a bank’s desire to mitigate “regulatory taxes” such as capital requirements (see, e.g., Pennacchi 1988). Moreover, loan sales may reflect the loan origination and distribution abilities of a bank. Consequently, we examine the motivation for a bank to sell loans as well as the effect of a loan sale announcement on the selling bank’s equity returns.

The outline of the article is as follows: Section II discusses data sources and sample selection. Section III presents the results of tests on the effects of loan sales on borrower returns and the link between loan sales and the long‐run viability of the firm. Section IV presents tests analyzing the effects of loan sales on the selling bank’s returns and the characteristics of those banks that sell loans. Section V presents some robustness checks. Section VI summarizes and concludes.

II.  Data and Sample Selection

The bank loan sales market is an over‐the‐counter, wholesale market in which transactions are arranged through a network of dealers. Historically, trades took many weeks to complete; however, dealers now work for completion of trades with a day horizon.5 Typical sellers of bank loans are large, wholesale money centers (e.g., Citibank) and overseas banks (e.g., ING Barings). Typical buyers of bank loans are smaller regional banks, foreign banks, and vulture funds (including hedge funds), as well as insurance companies (see, e.g., Miller 1998). Key sources of information about the market are trade newsletters and screen services such as Bloomberg that list and identify loan sales. Our test design requires identifying loans that are traded in the secondary market and locating the date of actual sale of a loan by a bank in order to conduct the event study. We use two market newsletters, Bank Letter and Gold Sheets, as the primary data sources. To identify the secondary market sale of a bank loan by one of the lenders, we check for loan sale news stories in Bank Letter. Bank Letter is a weekly publication, produced by Institutional Investor, Inc., which publishes a number of other well‐known newsletters such as Bond Week, Derivatives Week, and Corporate Financing Week.6 Bank Letter is the primary “trade rag” that carries news of the sales of bank loans by banks, and the date of Bank Letter publication provides us with the event date to conduct the event study. Gold Sheets is a publication of Loan Pricing Corporation, which provides information about new loan originations and the terms of these loans. Gold Sheets is widely used by practitioners, and recently by some academics. While it carries no news stories of loan sales, it does carry secondary market loan price quotes (not the actual trades) of a large number of loans. Thus, Gold Sheets allows us to generate the universe of borrowers whose loans attract a price quote in the secondary market. From our discussion with traders, these two publications are the most widely followed newsletters in the secondary loan market.

The main period of our study is 1995–98. (In Sec. VI, we look at an augmented sample period 1995–2000.) For the 1995–98 period in Gold Sheets, we found a total of 215 U.S. borrowers whose bank loans attracted a secondary market price quote that was published for the first time. Gold Sheets report the price quotes for the loans under two distinct categories. The first category is classified as “Par Loans” and includes all loans that are quoted at or very close to 100% of face value. The second category, termed “distressed loans” or “subpar loans,” comprises loans where the bid quote is significantly below the face value (usually 80% or below of the face value) or loans where the the borrower is in default.

Of the 215 loans, 162 were par loans and the rest (53 loans) were subpar or distressed loans. While Gold Sheets provides us with a fairly good estimate of the total universe of loans being traded in the secondary market, we were unable to identify the exact date on which the loan sale took place for the first time, as the newsletter does not carry news of loan sale announcements. The loan sale announcement dates were obtained from reading the weekly issues of Bank Letter. We found 101 news stories of loan sale announcements for U.S. corporate borrowers (an additional six news stories involved non‐U.S. borrowers for a total of 107 loan sale news announcements). This list also included multiple sale announcements for the same borrower. For example, there were 11 separate loan sale announcements for Mobilemedia in the period from August 12, 1996, to August 18, 1997. Taking the multiple announcements into account, the 101 announcements involved 58 different borrowers. To examine the information impact of the loan sale announcement, we carried out an event study on borrowers’ stock returns around the date of loan sale announcements. As a first step, all borrowers for which announcements of loan sales were published in Bank Letter were identified. We then searched the Center for Research in Security Prices (CRSP) daily price database for a match to those firms whose loans had been sold, so as to determine the availability of an equity price history around the date of announcement of a loan sale. This procedure reduced the sample of 58 borrowers to 29 firms for which sufficient stock price data existed. Since these loan sales are not classified as par or nonpar, we used the following procedure to classify these borrowers as either par or subpar. The loans were classified subpar if Gold Sheets listed the loan as nonpar within 2 months of the loan sale. Similarly if Gold Sheets listed it as par within 2 months of loan sale, we classified it as par. If the loan sold did not appear in Gold Sheets within 2 months of sale, we looked at the price range mentioned in the loan sale story in Bank Letter, and if the price was below 80% of the face value, we classified it as subpar, and par otherwise. Finally, if no price is mentioned, we tried to determine from news stories if the firm was facing repayment difficulties, and we classified it as subpar if such repayment difficulties were reported. Of the 29 firms we were able to classify 13 as par loans and 15 as subpar loans. (For one firm—Checkers Drive In Restaurants—no information was available, and we were unable to assign it as either par or subpar, so we excluded it from the analysis.) Appendix table A1 provides the details of the main sample.

As a second step we conducted an event study on the whole sample of 29 loan sale announcements to examine the impact of such announcements on a borrower’s stock return. Next, we split this sample into par and subpar loans since we expected bank informational effects to be stronger in subpar firms. However, comparing our sample with the total number of traded loans listed in Gold Sheets, it seems clear that our sample captures only a small proportion of all par loan sales (13 as compared with 162 par loans mentioned in the Gold Sheets) but does capture a larger proportion of subpar loan sales (15 as compared with 53 subpar loans in Gold Sheets). Since the informational effects for which we were looking are likely to be strongest in subpar loan sales, where we have a more representative sample of such traded loans, we focus primarily on these firms in this article. Hereafter, the 15 subpar firms whose loan sale announcements appeared in Bank Letter will be referred to as the Bank Letter sample, and the 53 subpar firms that generated a secondary market bid price quote will be referred to as the Gold Sheets sample. Of the 15 firms in the Bank Letter sample, 13 are also in the Gold Sheets sample; thus, the Bank Letter sample is effectively a subsample of the Gold Sheets sample of subpar firms.

In order to examine the operating performance of the firms in the period before the loan sale was made, we computed financial characteristics (performance measures) of these firms using data for the year prior to the year in which the loan sale took place. We calculated three operating rations for the borrowing firm. Return on assets was calculated by dividing earnings before interest, taxes, depreciation, and amortization (EBITDA; Compustat data item no. 13) by the book value total assets (Compustat data item no. 6). The second operating ratio that we used was total leverage, which is computed by dividing book value of current liabilities and long‐term debt (sum of Compustat item nos. 5 and 9) by the book value of total asset. Finally, we calculated investment intensity, which is the ratio of capital expenditure (Compustat data item no. 128) divided by the book value of total assets.

The borrowing firms whose loans were sold come from a number of industries; in order to account for industry‐wide effects, we adjusted each borrowing firm’s performance measures by median industry performance. For example, to calculate the industry‐adjusted return on assets, we calculated this ratio for all firms in the Compustat files that had the same four‐digit Standard Industrial Classification (SIC) code as the sample firm and took the median of these ratios.This median return on assets for the industry was then subtracted from the return on assets for the sample firm (for the same year). This procedure was carried out for all the nonpar firms in both samples so as to compute the industry‐adjusted return on assets for every firm.

To analyze the long‐term performance of the sample of loan sale firms (in the postloan sale period), we focused on the survival rate of these firms after an announcement of sale of their loan was made. To determine if any of the subpar firms in the Bank Letter as well as the Gold Sheets samples filed for bankruptcy after a bank announced the sale of a firm’s loan, we followed a two‐step procedure. As a first step we matched all the firms whose loans were sold against the list of firms filing for Chapter 11 in the Bankruptcy Datasource.7 This step allowed us to identify the firms that went bankrupt subsequent to the loan sale announcement. As a second step, we searched the Dow Jones News Retrieval Service for any stories that contained the sample firm’s name and the words “Chapter 11” or “bankruptcy” to confirm if the news of the bankruptcy was reported in the public media. This step is a robustness check to ensure that we identified all firms that filed for bankruptcy after a loan sale. While for the event study our sample is limited to subpar loans, where we have both the loan sale date as well as the CRSP stock prices, we can feasibly measure long‐term performance for the entire sample of subpar loans appearing in Gold Sheets, from the date that the first loan price appears. We report the long‐run survival rate for the firms in both the Bank Letter and the Gold Sheets samples. We find that the survival rates are similar across the two samples.

III.  Test Methodology and Results

We employ the event study methodology as outlined in Mikkelson and Partch (1986) to estimate the impact of a bank loan sale announcement on the stock return of the borrowing firm. The abnormal returns are computed around the date of publication of Bank Letter in which the loan sale announcement first appeared. The abnormal return for common stock of borrower j on day t is defined as Expression R_jt is the continuously compounded rate of return for borrower j on day t, and R_mt is the continuously compounded rate of return for the CRSP’s dividend‐inclusive, value‐weighted index for New York Stock Exchange/American Stock Exchange/NASDAQ stocks. The coefficients α_j and β_j are estimated by regressing R_jt for the period from 200 trading days before the event date (defined as the date of publication of Bank Letter announcing the loan sale) to 51 trading days before the event date on R_mt. The abnormal returns are computed for each day in the event period that begins 50 trading days before the announcement and ends 30 trading days after the announcement.

The average abnormal return on event day t for a sample of size N is

Tests of significance are based on standardized abnormal returns. Standardized errors for firm j on day t are defined as where

Expression V_j² is the residual variance for borrower j from the market model regression in equation 1, ED is the estimation period (150 trading days) used in the market model regression, R_mt is the market return on day t, and is the mean market return over the estimation period. The average standardized abnormal return for day t is

Under the assumption that individual daily abnormal returns are distributed normally, SAR_jt follows a Student‐t distribution with degrees of freedom. Cumulative abnormal returns ( ) are the sum of abnormal returns for the event window beginning with trading day T₁ and ending with T₂, and they are given by The test statistic for the null hypothesis that is given by where the average standardized cumulative abnormal return ( ) is calculated as follows:

The results for the event study are reported in table 1. Panel A reports the cumulative abnormal returns (CAR) for the entire sample of 29 loan sale announcements around the date of first announcement of a loan sale. Panel B reports the results of the event study for the 15 borrowers that we classify as subpar, while panel C describes the same for the 13 borrowers classified as par. (One borrower could not be classified as either par or nonpar and is excluded for the analysis.) As can be seen, for various event windows (7, 5, and 3 days) the results reported in table 1 provide strong evidence of a negative news effect surrounding a bank loan sale announcement. From table 1, panel A, for a 3‐day window surrounding the full sample of 29 loan sale announcements, the average abnormal return for the borrowers whose loans were sold was −1.74%, which is significant at the 1% level. However, this negative news impact is not the same across subpar and par loan sales. As reported in panel B, the impact of the news of a loan sale is much larger and more significant if the loan sale involves a subpar borrower. This result holds true across all event windows. Moreover, as shown in panel C, the negative abnormal return effect is statistically insignificant at conventional levels for the loan sale announcements involving par borrowers. Overall, these results are consistent with a bad news (information) effect arising from loan sales, in particular those of subpar borrowers, which is the converse of the good news effect of new loan announcements or renewals (see, e.g., James 1987).

Table 1 Cumulative Abnormal Returns for the Firms on the Announcement of Sale of Their Loan by Their Bank

Open New Window

Our next test examines whether the negative information that the market surmises from a loan sale announcement is validated by the borrowing firm’s subsequent performance. If the market perceives that the bank’s inside information about the firm’s future prospects is unfavorable, since the bank has decided to sell the loan rather than continue its lending relationship with the firm at its current level, then a logical consequence would be that such a firm’s performance would worsen subsequent to the loan sale.

Perhaps the starkest and simplest measure of poor performance is whether a firm goes bankrupt. We collect data to examine if and when a firm filed for bankruptcy (under Chapter 11 of the U.S. bankruptcy code) for a period of 3 years subsequent to the date of the loan sale announcement. This test has the additional advantage of allowing us to use the entire sample of 53 subpar loan sales obtained from Gold Sheets as well as the 15 subpar loan sales firms from Bank Letter used in the event study tests above. Overall, we find that a large number of our sample firms (close to half of the firms for whom loan sales occurred) file for bankruptcy within 3 years of the date of the first loan sale announcement.8 Table 2, panel A, shows that 32% of the firms whose loans were quoted as subpar filed for bankruptcy within 1 year of the first bid price quote reported by Gold Sheets, an additional 8% within 2 years, and a further 2% within 3 years. In aggregate, 42% of the Gold Sheets subpar sample firms filed for bankruptcy within 3 years of their loan first being announced for sale.

Table 2 Incidence of Financial Distress for the Sample of Firms for Which an Outstanding Bank Loan Attracted a Subpar Bid between January 1995 and December 1998

Open New Window

Although the bankruptcy filing rate appears to be high, there may be a concern that this rate reflects industry or economy wide factors. To control for such effects, we carry out a robustness test. For each firm whose loan attracted a price quote, we obtain the four‐digit SIC classification from Compustat. If the firm is not listed in the Compustat database, we search the Dealscan database for any loans made to that borrower. The Dealscan database is maintained by Loan Pricing Corporation and contains details of over 60,000 private loans made to U.S. borrowers. Apart from loan‐specific details such as amount, terms, and maturity, the database also provides borrower‐specific information, including primary SIC code. We were able to obtain information on 47 of our 53 borrowers.

Next we compiled a list of these firms that included their SIC classification and the year in which the first loan price quote for that borrower appeared in Gold Sheets. Since some of the borrowers are from the same industry or had their first loan price quoted in the same year, there is some duplication in this list. We eliminated multiple observations of the same industry by including each industry only once. If there were multiple borrowers from an industry, we picked the year in which an industry borrower’s loan was quoted in Gold Sheets for the first time. This year is defined as the base year for that industry.

Next, for each industry observation, we generated the list of all firms with the same four‐digit SIC code that are in the Compustat database for the year immediately preceding the base year. If there were fewer than three firms that matched the borrower’s industry group, we generated all the firms with same three‐digit SIC code. This step generates a plausible universe of all firms in the same industry as the firm that attracted the first loan price quote in that industry. We trace all firms in each industry for any incidence of Chapter 11 filing in the 3 years subsequent to the year of loan sale activity. This exercise allows us to generate the bankruptcy incidence in the industries to which the loan sale borrowers belonged. The results presented in table 2, panel B, show that the frequency of bankruptcy is 3% in the year in which the first loan sale price quote appeared, and for the year following the loan sale activity it is 1.6%. This is substantially lower then the results reported in table 2, panel A.

One can argue that the bankruptcy rate is likely to be higher among firms that are showing poor financial performance. Hence, we repeated the same exercise with a subset of those firms that fall in the bottom quartile of their industry as ranked by their operating performance. We calculated the ratio of EBITDA to total assets for all firms. Those firms for which this ratio is in the bottom 25% of the industry were included in the sample of poorly performing firms in that industry. As expected, the bankruptcy filing rate is higher for this subsample, with the frequency of bankruptcy being 4.7% in the year in which the first loan price quote appeared and 2.6% for the first year following the loan sale activity. As the results in table 2, panel C, show, the bankruptcy rate is substantially lower than the rates reported in panel A.9

These results of subsequent poor performance, combined with our results of the negative stock price reaction at the time of the seasoned nonpar loan sale announcement, are further evidence consistent with the view that banks play a special, or unique, role as monitors (or corporate insiders) and that announcements about bank lending decisions convey hitherto private information to the capital market at large (see, e.g., Fama 1985; James 1987).

A natural question, however, is whether the ex post performance of these firms is correlated to their ex ante performance prior to the loan sale, that is, whether we could have anticipated the poor performance of these firms even without the announcement of the sale of a loan. Related to this is another question: What are the (publicly available) financial characteristics of those firms whose loans are sold at the time of the loan sale? To evaluate this question, we collated information on the financial characteristics of those firms whose loans were sold in the year prior to the loan sale.

Table 3 provides some financial and operating performance measures for the sample of borrowing firms whose loans were sold over the 1994–98 period. Panel A documents these measures for the sample of nonpar firms obtained from Gold Sheets, while panel B presents similar results for the Bank Letter sample. These performance measures are reported on an industry‐adjusted (median) basis. Specifically, financial ratios were calculated for all firms with the same four‐digit SIC code as that of the borrowing firm whose loan was sold, and the median financial ratio (for the industry) was then subtracted from the financial ratio of the borrowing firm. Three such financial ratios are shown in table 3. As can be seen, firms whose loans were sold appear to have performed below the industry median in the year preceding the loan sale. In particular, they had a lower return on assets, a higher level of debt (as measured by the ratio of book liabilities to total assets), and a lower degree of investment intensity (measured by capital expenditures to total assets) compared with the median firm in the same industry.

Table 3 Industry‐Adjusted Performance Characteristics for the Sample of Subpar Loan Firms

Open New Window

Nevertheless, we wished to examine further whether these are the poorest performing firms in their respective industry, that is, whether, on an ex ante basis, we would have expected these firms to have poor future viability (i.e., file for bankruptcy). For example, if our loan sale sample firms were in the bottom quartile of their industry, an astute investor might reasonably anticipate that some of these firms would go bankrupt or exit the industry regardless of a bank’s decision to sell the firm’s loans. Table 4 compares the financial performance of the two nonpar loan samples to their respective peer industry groups. Specifically, for the year immediately preceding the loan sale announcement, we calculated a set of financial ratios (as publicly available proxy measures of performance) for all firms that have the same four‐digit SIC codes as each borrower whose loan was sold. This allows us to generate a distribution for each financial ratio within the loan sale firm’s industry. Surprisingly, we find that the firms whose loans were sold are not the worst performers in their respective industries across all performance measures, nor were they always concentrated in the bottom decile or even the bottom 25% (quartile). As table 4 shows, for two of the three accounting measures, return on assets and investment intensity, the majority of the sample firms lie in the second quartile (twenty‐fifth to fiftieth percentile). The results are similar across both the sample of nonpar borrowers obtained from Gold Sheets (panel A) and the sample obtained from Bank Letter (panel B). This suggests that publicly available financial information alone may have been insufficient for outside investors to clearly distinguish, or preidentify, the degree of weakness of the firms whose loans were sold and that publicly announced loan sale decisions by banks appear to have provided valuable (hitherto private) information to outside investors regarding the true financial condition of these firms.

Table 4 Within‐Industry Distribution of Selected Financial Characteristics for the Two Samples of Subpar Loan Sales

Open New Window

IV.  The Effect of Loan Sales on Bank Stock Returns

The decision to sell a loan may also contain information about the quality of bank loan portfolios. Indeed, loan sales may be interpreted favorably by the market as a reflection that the average quality of a bank’s remaining portfolio will improve—given its incentive to sell off, or divest, its poorer quality loans. However, in selling such loans the bank has to assess the potential cost of such sales, which might harm its relationship with the borrower whose loan is sold (as well as potential borrowers who may be concerned that their loans will be sold in the future)10 or its reputation with investors who buy the loans sold by the bank (should they deteriorate further in quality). Plus, a loan sale might signal that the bank’s management in general has exhibited poor judgment in its lending decisions or that its capital position is weak. To examine the net effects of a loan sale on a selling bank, we conducted an event study for those banks announcing the sale of loans. We are limited by the fact that Bank Letter news stories announcing loan sales do not always mention the identity of the selling bank. The quote below provides a typical loan sale announcement: “A Musicland lender auctioned off $11 million in bank debt late last week, with bids in the high 70s, according to market sources” (Bank Letter 1997).

For the total of 107 loan sale announcements originally collected from Bank Letter, we were able to identify the loan‐selling institution in 58 cases. Of these, 25 sale announcements were made by foreign financial institutions and 33 were made by U.S. financial institutions. The lenders mentioned in these 33 U.S. financial institution announcements were then matched to the CRSP daily stock price database. Two lenders did not have stock price information,11 and two announcements involving Fleet Bank were made on the same day and, thus, were included as a single announcement, leaving us with 30 bank loan sale announcement dates. Finally, we removed the three loan sales made by investment banks, leaving 27 U.S. commercial bank loan sellers.12 The results of the event study on selling banks’ returns are reported in table 5, panel A. We also looked at the subset of loan sale announcements involving the 15 nonpar borrowers that were examined in the “borrower”‐related tests described in tables 1–5. This subset yielded 15 clearly identifiable U.S. bank lenders. The loan sale announcement effects for these 15 banks are reported in table 5, panel B.

Table 5 Cumulative Abnormal Returns for the Banks on the Announcement of Sale of Their Loan

Open New Window

The results of table 5 indicate that, on average, the sale of loans by banks does not appear to have any net (new) impact on the selling banks' stock returns (i.e., there is no evidence of any net costs or benefits to the selling banks’ shareholders). This result holds true for the whole sample for which the bank seller could be identified (panel A) as well as the subsample of nonpar loan sales (panel B).

Nevertheless, it is of interest to examine the characteristics that differentiate between those banks that engage in loan sales and those that do not, especially as some have argued that an important motivation for loan sales is to improve a bank’s solvency position, such as the bank’s regulatory capital ratio (see, e.g., Pennacchi 1988). As discussed above, for the event study examining a bank’s own share price reaction to a sale of a loan, we were able to identify 27 unique loan sale event dates. Since some banks announced multiple sales in the same year, these multiple loan‐selling banks were treated as a single observation in the same year. This resulted in a final sample of 19 distinct bank‐loan sale years. We obtained key financial characteristics for each of the selling banks for the year before the loan sale announcement from a variety of sources, including Bank Compustat, annual reports, and 10‐K filings. These included size denoted by total assets (Bank Compustat data item no. 36), tier‐1 capital (data item no. 48), reserves for bad loans (data item no. 78), and net income (data item no. 161). We use these to construct the operating measures described in table 6. For each bank sale year we generated the list of all banks in the Bank Compustat by using the following four‐digit SIC codes: 6020 (commercial banks), 6021 (national commercial banks), 6022 (state commercial banks), and 6029 (commercial banks). We then calculated the ratios described in table 6 for these banks, giving us the distribution for each ratio of the banking industry as a whole. Banks involved in loan sales are large banks, with 18 out of 19 selling banks to be found in the top 25% of the banking firms ranked by total assets. Most of the loan‐selling banks also fall in the bottom quartile of banks ranked by their tier‐1 capital, which is consistent with Pennacchi's (1988) hypothesis that a prime incentive for loan sales is to boost a bank’s capital ratios (although it is worth noting that none of the banks were below the Bank for International Settlements statutory 4% minimum requirement). In addition, more than three‐quarters of the banks fall in the top 50% of banks ranked according to the bad loan reserves to total asset ratio. Finally, the selling banks are evenly distributed across the four quartiles if the ranking is done according to net income to total assets ratio. Overall, it appears that loan‐selling banks are, on average, of poorer quality than are nonselling banks.

Table 6 Within‐Industry Distribution of Selected Financial Characteristics for the Two Samples of Subpar Loan Sales

Open New Window

V.  Additional Robustness Tests

Our main sample consists of all news stories from 1995–98. Our sample ended in 1998 so that we could look at the long‐term performance of those firms whose loans were sold. However, while we cannot examine the long‐term performance of more recent loan sales, it is possible to extend our event study to include more recent loan sales (so as to examine the robustness of our event study results). Accordingly, we gathered additional data on loan sale news stories for 1999 and 2000 from Bank Letter/Loan Market Week.

Another robustness check is to examine whether there were other events taking place at the same time as the loan sale that may have affected the event study results. Accordingly, we did a news search on ABI/Inform for all news events in the event‐study window. We ruled out observations where there was news of losses or defaults, a cut or suspension of dividends, a credit rating cut, a CEO selling a major equity stake, a lawsuit being filed, and takeover‐target‐related news. As a result of these checks, we dropped three observations from our original sample and added 18 new observations from the 1999–2000 period. This yielded a total sample of 44 loan sales, of which 25 are subpar, 18 are par loan sales, and one cannot be classified.

With respect to the event study for the augmented sample (1995–2000) for borrowers for whom the loan sale is announced, our results are very similar to those shown in table 1 for 1995–98. The augmented results are reported in table 7. Overall, we find a significantly negative stock price reaction to loan sale announcements, which is larger and more significant for subpar loan sales. Moreover, the signs and scale of abnormal returns closely mirror those in table 1.

Table 7 Cumulative Abnormal Returns for the Firms on the Announcement of Sale of Their Loan by Their Bank for the Augmented Sample from 1995 to 2000

Open New Window

By adding 1999 and 2000, we were able to augment our original sample of 27 by an additional six banks that were clearly identifiable and for which CRSP data were available. Thus, we had a larger sample of 33 banks. We continued to find insignificant abnormal returns for selling banks with the larger sample.13

Finally, it might be of interest to examine the wealth effects for the “buyers” of the bank loans that are sold. As shown in the appendix, only in a limited number of cases do we know the identity of the buyer. Specifically, for the augmented sample of 44 loan sales, we have information (and CRSP data) for the buyers in eight cases. We conducted an event study to test for cumulative abnormal returns for buyers on the loan sale announcement. Perhaps not surprisingly, given the small sample size, we found insignificant cumulative abnormal returns for this group.

VI.  Summary and Conclusions

This is the first article to evaluate empirically the effects of secondary loan sales by banks on both borrowers as well as bank stock returns. Using loan sales announcements reported in the publication Bank Letter, we find evidence of a negative effect of loan sales by banks on borrowers’ returns. This effect is strongest for the sale of subpar loans. This result is consistent with a strong certification effect of loan sales. Specifically, the sale of a loan appears to carry a signal to the capital market that is the mirror image of the findings of James (1987), Lummer and McConnell (1989), and others regarding the (positive) news effect of loan initiations. This result adds further support to the view that banks play a special, or unique, role in diffusing hitherto private information to outside investors regarding the performance of borrowing forms. We also find that 42% of the firms whose loans are sold file for bankruptcy within 3 years of the announcement of a loan sale by their bank lender. While cross‐sectional tests confirmed that borrowers whose loans are sold are generally in a weaker financial and operating condition than a matched control group of firms whose loans are not sold, the sale of a loan appears to convey to the market additional (material) information about the relative degree of weakness of the borrowing firm whose loan is sold.

Finally, we examined the effects of a loan sale on the selling bank itself, since the sale of a loan may convey new information to the market regarding the quality of a selling bank’s loan portfolio as well as affect, among other things, its reputation and relationships with other loan customers and investors. Interestingly, the sale of a loan by a bank carried no significant impact on its own stock return, although loan sales appear to be made by generally weaker banks.