- Email: [email protected]
Insider trading behavior prior to Chapter 11 bankruptcy announcements
Journal of Business Research 54 (2001) 63 – 70
Insider trading behavior prior to Chapter 11 bankruptcy announcements Yulong Ma* Department of Finance, Real Estate and Law, College of Business Administration, California State University, Long Beach, 1250 Bellflower Boulevard, Long Beach, CA 90840, USA Accepted 18 December 1999
Abstract This study uses non-parametric methods to examine the difference in trading behavior between Chapter 11 bankruptcy firms and nonbankruptcy firms traded on New York Stock Exchange (NYSE)/American Stock Exchange (AMEX) exchanges. It documents new evidence about the anomaly of insider purchases rather than sales prior to Chapter 11 bankruptcy announcements. Insiders of Chapter 11 bankruptcy firms purchase significantly fewer shares than insiders of the control firms before the bankruptcy announcement. Although insider trading volume declines long before the announcement, the decline is statistically significant only during the 3-month period before the announcement. The study, however, finds no significant difference in trading behavior between insiders of firms that ceased to trade on and those that continue to trade on the NYSE/AMEX within 5 years after the bankruptcy announcements. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Insider trading; NYSE/AMEX; Chapter 11 bankruptcy
Empirical research on insider trading has shown that the significance of abnormal insider trading activities around corporate announcements may depend on the specific corporate event being investigated. Numerous studies such as Penman (1985), Hirschey and Zaima (1989), Karpoff and Lee (1991), and Pettit et al. (1996) have documented significant abnormal insider trading activities around corporate announcements. Other studies, however, find no strong associations between insider trading and corporate announcements. For example, Givoly and Palmon (1985) report no significant abnormal insider trading activity before the announcements of 11 types of corporate events. Further, Reinganum (1987) argues that insiders tend not to trade their own firms’ stocks prior to large stock price increases. Since Chapter 11 bankruptcy announcements cause large stock price declines, it is important to examine how insiders of these bankruptcy firms trade. There exists only limited empirical research on insider trading activities and corporate bankruptcies. Loderer and Sheehan (1989) study this issue by investigating whether insiders of bankrupt firms reduce their stockholdings in
* Tel.: +1-562-985-4563; fax: +1-562-985-1754. E-mail address: [email protected] (Y. Ma).
comparison with those of firms that do not file for bankruptcy. Their results show little evidence of abnormal insider trading behavior by insiders of bankrupt firms prior to the bankruptcies. Another study by Gosnell et al. (1992) uses actual insider trading transaction data to examine insider trading activities before the announcement of bankruptcy filings. They find that insiders of OTC firms increase their sales significantly over a period of 5 months before the bankruptcy announcement while the exchange-listed firms did not. A more recent study by Seyhun and Bradley (1997) examines the trading activities by insiders of New York Stock Exchange (NYSE) and American Stock Exchange (AMEX) bankrupt firms over the time period of 1975 to 1992. Their results show significant insider selling before the bankruptcy announcement with more intense selling by top executives. Should we expect abnormal insider trading activities before firms announce bankruptcy filings? By the Securities and Exchange Commission’s (SEC) definition, insiders are chairmen, directors, officers, etc., and principal shareholders with 10% or more of any equity class of securities. Corporate insiders refer to those insiders who hold managerial positions in the firm. The prevalent belief is that insiders possess superior information about their firms’ future prospects and may trade either to make profits or to convey private information to the market. Financial
0148-2963/00/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved. PII: S 0 1 4 8 - 2 9 6 3 ( 0 0 ) 0 0 11 5 - 6
64
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
agency theory suggests that insiders will be more likely to engage in self-serving activities when a firm is in financial distress. Although equity values decline significantly long before firms filing for bankruptcy, firms announcing Chapter 11 bankruptcy still experience significantly negative abnormal returns over the announcement period (e.g., Altman and Brenner, 1981; Clark and Weinstein, 1983; Morse and Shaw, 1988). Therefore, corporate insiders may be motivated to sell more stocks prior to the bankruptcy announcement. As such, insiders, theoretically, can certainly minimize their losses by reducing purchases or increasing sales of their own firms’ stock prior to the bankruptcy filings. However, trading by insiders based on material private information is against government regulations and corporate policies. Illegal insider trading may incur both high civil and criminal costs, which may deter insiders from trading before corporate announcements, especially with large price effects. This may explain the argument made by Reinganum (1987) that insiders tend not to trade their own firms’ stocks prior to large stock price increases. Meulbroek (1992) presents a detailed discussion and analysis on government regulations against illegal insider trading. Hence, whether insiders trade abnormally prior to the announcement of Chapter 11 bankruptcy may be an empirical issue. In contrast to events with large price increases, two aspects associated with insiders of bankrupt firms deserve emphasis. First, the public is usually aware of the financial difficulty of the firm before it declares bankruptcy, and its stock price usually began to decline long before the announcement. This means that insiders of these firms could use their firm’s publicly known financial distress as an excuse for selling before the bankruptcy announcement to avoid possible allegations of illegal trading. Second, since insiders in the financially distressed firms are more willing to take risks, insiders of Chapter 11 bankruptcy firms should have more incentive to engage in abnormal trading activities before the announcement. Therefore, even facing stringent government regulations, insiders of bankruptcy firms may still try to change their trading patterns in such a way so as to reduce their equity losses. The purpose of this study is to examine whether insiders of Chapter 11 bankruptcy firms change their trading patterns/behavior prior to the bankruptcy announcements to reduce their financial losses. Unlike Loderer and Sheehan (1989), who use changes in shareholdings in the proxy statements, this study uses the actual insider trading transaction data from the SEC. It differs from the study by Gosnell et al. (1992) in the following two ways. First, this paper investigates a final sample of 89 NYSE and AMEX firms over a period of 10 years. Their final sample includes only 18 NYSE and AMEX firms over a 3-year period. Second, this study focuses exclusively on firms filing Chapter 11 bankruptcy. Our study also differs from Seyhun and Bradley (1997) in sample selection and testing method. In their study,
sample firms consist of those that filed a voluntary bankruptcy petition over the period of 1975 –1992. This research includes only Chapter 11 bankruptcy firms. Further, we employ non-parametric methods to test the significance of abnormal insider trading rather than parametric methods. Seyhun (1990) argues that insider trading does not follow a normal distribution. The focus on Chapter 11 bankruptcy firms is motivated by the following reasons. First, Chapter 11 bankruptcy process, as a result of the 1978 Bankruptcy Act, gives the stockholders substantial protection against creditors. Morse and Shaw (1988) argue that the new changes may favor more reorganizations. The increasing number of firms seeking protection under Chapter 11 process is in supportive of this point of view. Franks and Torous (1989) also indicate that Chapter 11 bankruptcy process has been described as favoring corporate management and can be strategically used by managers as a new measure in its armory when the going gets tough rather than when the firm is insolvent. Therefore, the new attitude towards using Chapter 11 protection by the management may have a significant impact on insider trading behavior. Second, from an empirical point of view, the existence and distinction of Chapter 11 bankruptcy from other bankruptcy forms warrant an empirical investigation of the trading behavior by insiders of Chapter 11 bankruptcy firms. The empirical results provide new evidence of abnormal trading behavior by insiders of Chapter 11 bankruptcy firms. The finding shows that most of the significant abnormal insider trading occurred within 6 months before the bankruptcy announcement for insider purchases rather than sales. It also shows that total insider trading volume reduced significantly during the 3-month period prior to announcement. Efforts are also made to examine whether separating corporate insiders (holding managerial positions) from all insiders sample provides additional information. The results show very similar trading patterns but stronger significance levels for the corporate insiders sample in comparison with all insiders sample.
1. Data source and sample description The stock returns and the returns on the equally weighted portfolio of all NYSE and AMEX stocks are obtained from the Center for Research on Security Prices (CRSP) daily stock tapes. The insider trading data are retrieved from the Ownership Reporting System (ORS) master tape and the ORS accumulative tape. The master tape covers the period of January 1980 to August 1987, and the accumulative tape covers the period of July 1986 to March 1991. Although differing in data structure, these two tapes provide all trading transactions by insiders of the NYSE/AMEX firms over the respective time periods. Insider trading information is also available from the Official Summary of Securities Transactions and Holdings published by the SEC.
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
The insider trading tapes contain all types of transactions by insiders, that is, open market purchases and sales, private purchases and sales, exercises of options and warrants, shares acquired and disposed of through a plan, stock dividends, etc. For the purpose of this study, only open market and private purchases and sales are included in the empirical investigation. This approach is also consistent with the method used by the Value Line Investment Survey and the Wall Street Journal. The information for firms announcing filings of Chapter 11 bankruptcy petitions is collected from the Predicasts F&S Index of Corporate Changes under the section of Bankruptcy. Firms that filed petitions during the period of 1982 through 1990 constitute the initial sample. The initial sample consists of 162 announcements made by 157 firms over the 10-year period. Of all the initial sample firms, there are only 147 announcements made by 147 firms matched with the 1995 CRSP monthly file. That is, the stocks of these 147 firms were traded on the New York and American Stock Exchanges during some time in the 10-year time period. Of these firms, 42 firms have announcement dates either before their beginning trading or after their ending trading dates on the CRSP file; 12 firms have no matches with the firms on the insider trading tape (ORS). In addition, there are three firms that match the insider trading tape, but have no insider transaction data; finally, there is one firm that has abnormal insider trading pattern. The final sample of 89 firms is the result of the screening process mentioned above. Table 1 provides some of the basic descriptive statistics for the sample firms and the control firms. Panel A of the table shows the firm number and firm size distributions over the investigation period for the bankruptcy and control samples. Panel B of the table shows the announcement period information effect of the bankruptcy firms. The abnormal returns are estimated using the traditional market model in which the estimation period is 1 year before the announcement date. Over the period of 12 trading days (10, +1) prior to the announcement, the sample firms experience significantly negative cumulative abnormal returns of more than 25%. In comparison, the cumulative abnormal returns for the control sample are less than 1% and insignificant over the same 12-day period (10, +1) prior to the announcement. To examine the trading behavior by insiders of the Chapter 11 firms, the control sample is constructed to estimate the normal level of insider trading activities using the following procedure. Each of the 89 control firms is selected from the CRSP monthly file. A random search is made to match the industry and size of the control firms with those of the Chapter 11 firms as closely as possible. For each bankruptcy firm, a search is made from the beginning of the CRSP monthly file. When a firm meets the industry code, it then is screened for the size requirement. If the first run does not generate a matched firm, another run is made with wider upper and lower bounds
65
Table 1 Basic sample descriptive statistics Panel A: Firm size and distribution Year
Initial
Final
Chapter 11
Control
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Total
1 26 13 18 16 22 15 16 9 26 162
0 18 7 11 10 15 5 11 8 7 89
N.A. 171.531 165.994 149.700 125.195 207.870 62.399 136.229 43.739 259.697 161.003
N.A. 171.860 155.116 139.352 138.340 223.227 159.864 134.002 35.401 255.093 167.412
Panel B: Announcement period cumulative abnormal returns Period
Cumulative abnormal return
t-Statistics
(50, 11) (10, 2) (1, 0) +1
1.938 7.529 17.597 0.815
1.140 8.116*** 43.473*** 1.316
Panel A shows the number of firms in the initial and final samples over the 10-year period. The market value of a firm’s total equity is used as a proxy of firm size. Panel B presents the announcement period abnormal returns using the OLS market model. *** Significant at the 1% level.
for the firm size requirement. The industry is identified by the first two digits of the four-digit SIC code for the Chapter 11 firm. The market value of the firm’s total equity is used as a proxy of firm size. For the bankruptcy sample, stock prices 2 years prior to the announcements of Chapter 11 bankruptcy filings are used; while for the control sample, stock prices from the monthly file and just before the announcements are used. The average size of the Chapter 11 firms is about $161.003 million and the average size of the control firms is $167.412 million.
2. The methodology 2.1. Insider trading measures Insider trading is measured in terms of the number of transactions, the number of insiders, and the dollar amount of the transactions. The reason that these measures are employed is that in the literature of insider trading, there is still no one measure that is generally preferred to others. In addition, under each of the three measures, insider trading activities are examined separately for purchases, sales, and jointly for net purchases. The purpose for using separate analyses for purchases and sales is to examine how insiders act upon new information through purchasing and selling transactions, especially given the penalties on illegal insider trading. For instance, it is easy to argue that insiders are trading illegally if they purchase abnormally more of their own firm’s stock prior to the announcement of favorable
66
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
news. However, it is hard to make a case against insiders who sell abnormally less of their own firm’s stock prior to the announcement. The same argument holds true for the announcement of unfavorable news. The insider trading measures are computed on a monthly basis over the time period of January 1980 to March 1991. For each firm, PNit is used to denote the number of purchases made by insiders in the firm i during month t. Similarly, SNit is used to express insider sales for firm i in month t. The net purchase measures of insider trading are simply calculated as the difference between the purchase and sale measures, NPNit = PNit SNit. Given the time period covered in this study, the largest number of monthly insider trading measures is 135. These insider trading measures for Chapter 11 bankruptcy firms are compared with those for control firms over a period of 24 months before the announcements of Chapter 11 bankruptcies.
i = 1,2, . . . , N. These absolute values are ranked with the highest rank N assigned to the pair with the largest absolute difference. For pairs with the same absolute values, the average of the ranks is assigned to each of these pairs. The method used here in handling zero differences follows the one suggested by Pratt (1959). Wilcoxon (1949) discards all the zero differences before these differences are ranked. Conover (1980) provides more detailed descriptions about these procedures. The test statistic is then computed based on the signed-rank variable, Ri, which is determined in accordance with the following rule: Ri = the rank assigned to (MPNib MPNic) if Di is positive; Ri = the negative of the rank assigned to (MPNib MPNic) if Di is negative; Ri = 0 if Di = 0. The test statistic is then defined in Eq. (2) as: N P Ri i¼1 T ¼ sffiffiffiffiffiffiffiffiffiffiffi ð2Þ N P 2 Ri i¼1
2.2. Statistical methods This paper mainly relies on non-parametric approaches to examine the abnormal level of insider trading activities for Chapter 11 bankruptcy firms. Seyhun (1990) argues that the insider trading activity is non-normal, infrequent, and highly skewed in magnitude. Therefore, it seems more appropriate to use non-parametric methods. The non-parametric tests do not require the strong assumption of normality in the underlying distribution of the research variables. To compare insider trading measures of bankruptcy firms with those of control firms, the Wilcoxon signed-rank test is employed. To study the trading behavior by insiders of firms with negative and non-negative announcement period CARs, the Wilcoxon sum-rank test is used. The Wilcoxon signed-rank test is not available in SAS. The following provides a brief description of its testing procedure. For any firm i in the bankruptcy sample of N firms, the insider trading measure, PNit, is calculated for the 24 months prior to the announcement. That is, t ranges from 24 months to 1. The 24 months are split into eight non-overlapping quarterly periods. The mean values of insider trading measures for the bankruptcy sample over these quarterly periods are compared with their corresponding values for the control sample. For example, to compare insider trading measures in the first quarter for these two samples, the paired values for any sample firm are computed as in Eq. (1). MPNib ¼
1 X t¼3
PNit =3;
MPNic ¼
1 X
PNit =3
ð1Þ
t¼3
MPNib is the mean value for bankruptcy firm i, and MPNic is the mean value for the corresponding firm i in the control sample. The absolute differences of these paired values can be computed for each of the N paired-firms of these two samples. That is, |Di| = |MPNib MPNic|, where
This test statistic follows approximately a normal distribution when the number N of the paired observation is large (N > 29 in statistical sense). For our samples, the N is 89.
3. Empirical results The results are reported for both corporate insiders and all insiders. As mentioned earlier, corporate insiders refer to those insiders who possess managerial positions in the firm; while all insiders include all insider categories as defined by the SEC. The purpose of these separate analyses is to see whether corporate insiders as a group are more privy to firms’ private information will trade differently from all insiders as a whole. On the one hand, corporate insiders may be expected to trade more actively by taking advantage of their managerial positions in the firm. On the other hand, corporate insiders may be more inclined to constrain their trading activities, knowing that they are under closer scrutiny, particularly with regard to Chapter 11 bankruptcy announcements. From the investing public’s perspective, it is useful and important to see, empirically, whether trading patterns by corporate insiders convey significantly better information than those by all insiders. Although all insider trading analyses are conducted for eight quarterly periods prior to the announcements, however, only the testing results for the first four quarters are reported due to the insignificance of results from other quarters. Table 2 presents the results of the Wilcoxon signed-rank test that compares insider trading patterns between the bankruptcy sample and its control sample. All figures in the table represent the test statistics for the difference in insider trading between these two samples. Panel A of the table reports the results for corporate insiders and Panel B reports the results for all insiders. For both corporate
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
67
Table 2 Insider trading comparison between the Chapter 11 sample and the control sample Number of transactions Period
Buys
Sells
Number of insiders Net buys
Panel A: Insider trading measures including only transactions QTR 4 0.407 0.085 0.345 QTR 3 1.222 0.316 0.123 QTR 2 1.960** 0.298 1.277 QTR 1 2.143** 1.001 0.167
Buys
Sells
by corporate insiders 0.556 0.839 1.162 0.234 1.891* 0.012 2.176** 1.400
Panel B: Insider trading measures including transactions by all insiders QTR 4 0.182 0.294 0.372 0.059 QTR 3 0.597 0.182 0.242 0.663 QTR 2 1.882* 0.764 1.620 1.762* QTR 1 1.960** 0.793 0.085 2.150**
0.588 0.048 0.737 1.245
Dollar amount of trades ($1,000) Net buys
Buys
Sells
Net buys
0.031 0.134 1.245 0.205
0.787 1.293 1.830* 2.226**
1.185 0.723 0.175 1.824*
0.146 0.363 0.501 0.719
0.070 0.096 1.475 0.175
0.288 0.649 1.764* 2.139**
0.931 0.837 0.780 1.444
0.117 1.132 0.799 0.274
The figures in the table are Wilcoxon signed-rank test statistics for the difference in insider trading between Chapter 11 bankruptcy firms and controlledsample firms. The Wilcoxon test is applied for all nine difference insider trading measures over a four-quarters period prior to the announcement of Chapter 11 filings. Panel A shows the results for only corporate insiders who have a managerial position in the firm while Panel B provides the results including all insiders defined by the SEC. * Significant at the 10% level. ** Significant at the 5% level.
insiders and all insiders, the results show that there is no significant difference in insider trading between these two samples during the third and fourth quarters before Chapter 11 announcements. During the first and second quarters, only the testing results for insider buys are significant. The Wilcoxon test statistics are significant at the 5% level for insider buys measures in the first quarter, and are significant at the 10% level for insider buys measures in the second quarter. The negative value means that the insider trading measure for the bankruptcy sample is smaller than that for the control sample. As shown in the table, all of the statistics in the Buys column are negative. These results indicate that during the two quarters prior to the Chapter 11 announcement, insiders in the bankruptcy firms purchased significantly less stock than those in the control firms. This implies that insiders may in fact have already incorporated the upcoming unfavorable news in trading on their own firm’s stock. However, this argument does not carry over to insider sales and net buys measures. In contrast to the popular perception that insiders of bankruptcy firms may sell their stock before the announcement, none of the test statistics are significant at a reasonable level. The result seems to suggest that insiders of bankrupt firms do not engage in active trading schemes in anticipation of the announcement of Chapter 11 bankruptcies. Rather, they adopt passive trading tactics by restraining their purchasing activities. Harlow and Howe (1993) find similar results in their study. A comparison of Panel A with Panel B shows that all statistics except one for all insiders are larger in absolute value than those for corporate insiders though the differences are very small. This shows that although corporate insiders have higher significant levels, they, in general, probably do not behave significantly differently in trading from the rest of insiders before the firm files for Chapter 11 bankruptcy.
To further examine insider trading behavior, the total volume of insider purchases and sales from the bankruptcy sample is compared with that from the control sample using the Wilcoxon signed-rank test. The total volume is defined as the sum of insider purchase and insider sale measures. Table 3 contains the testing results. All of the figures except one in the table are negative, indicating that the trading volume is lower for the bankruptcy sample than for the control sample over the four quarters before the Chapter 11 announcements. However, basically, only those numbers in the first quarter are statistically significant at the 1% or 5% levels (one of the exceptions is in Panel B).
Table 3 Trading volume comparison between the Chapter 11 and the control samples Period
Transactions
Insiders
Dollar amount
Panel A: Transactions including only corporate insiders QTR 4 0.259 1.391 QTR 3 0.823 0.764 QTR 2 1.685* 1.446 QTR 1 2.178** 2.823***
1.900* 1.210 0.713 3.223***
Panel B: Transactions including all insiders QTR 4 0.367 0.724 QTR 3 0.598 0.219 QTR 2 1.079 1.098 QTR 1 1.553 2.494**
1.439 1.439 0.969 2.698***
The Wilcoxon signed-rank test is conducted for the difference in trading volume between firms in the Chapter 11 and the control samples. Trading volumes are measured as the sum of buys and sells in terms of the number of transaction, the number of insiders, and the dollar amount. The test statistics are shown for trading activities made by corporate insiders (Panel A) only and by all insiders (Panel B). * Significant at the 10% level. ** Significant at the 5% level. *** Significant at the 1% level.
68
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
In both Panel A and Panel B, insider trading volume is measured in numbers of transactions, numbers of insiders, and dollar amount. Overall, there is not much difference between Panel A and Panel B, although the test statistics are slightly higher in absolute value for corporate insiders than for all insiders. The only significant difference between these two panels occurs when the Wilcoxon statistic in terms of number of transactions turns from significant for corporate insiders to insignificant for all insiders. These results appear to indicate that insiders of bankrupt firms reduce their trading volume long before the bankruptcy announcements. However, the decline in trading volume is statistically significant only in the quarter just prior to the announcements when compared with that by insiders of the control firms. The findings are consistent with those from Table 2. That is, insiders do not increase their selling activities, but rather reduce their purchasing activities prior to the Chapter 11 announcements. Table 4 provides some evidence of the relationship between insider trading activities and announcement information effect among the Chapter 11 bankruptcy firms. The main empirical issue is whether insiders of firms with different levels of announcement period abnormal returns trade differently before the announcements. The purpose is to see whether we can differentiate Chapter 11 bankruptcy firms based on their insider trading activities. Among the 89 sample firms, 74 firms have negative announcement period abnormal returns over a 12-day announcement period (10, +1). The Wilcoxon sum-rank test is used to test the mean difference in insider trading between firms with non-negative and negative announcement period CARs. Panel A of the table contains the test statistics for corporate insiders and Panel B contains those for all insiders. During QTR 4 and QTR 3, there are more figures in Panel A significant at the 10% level than in Panel B (only
one test). During QTR 1, only two statistics under Net Buys are significant at the 10% level for Panel A, and at the 5% level for Panel B. This trading pattern seems to suggest that corporate insiders’ trading behavior is not quite the same as non-corporate insiders’. A positive value in the table indicates that the insider trading measure in the quarter for firms with non-negative CARs is large than that for firms with negative CARs. For a negative value, the opposite holds true. These findings indicate that among Chapter 11 bankruptcy firms, insiders seem to control their trading activities based on their private information contained in the forthcoming announcements. In particular, correctly anticipating the market reaction, insiders of firms with negative announcement effects seem to purchase less than insiders of firms with positive announcement effects. An effort has also been made to determine whether insiders of firms that re-emerge from the Chapter 11 process trade differently from those of firms that do not. In their recent study, Rimbey et al. (1995) find that the market can discriminate between firms that are ultimately worthless and those that may retain some shareholder value. The bankruptcy sample is split into two groups. One group contains all firms that ceased to trade on the NYSE/AMEX either shortly after or within 5 years after the announcement of Chapter 11 bankruptcies. The remaining firms constitute the second group. There are 65 firms in the first group and 24 firms in the second group. The Wilcoxon sum-rank test is used to test the mean difference in insider trading between insiders in the first group and those in the second group. The test fails to provide significant results to support the argument since, basically, none of the Wilcoxon statistics are significant at a reasonable level. The evidence presented above is, in general, consistent with previous studies on insider trading, and it also helps shed new light on trading behavior by insiders of Chapter 11
Table 4 Insider trading comparison between positive and negative announcement period CAR firms Number of transactions Period
Buys
Sells
Number of insiders Net buys
Buys
Sells
Dollar amount of trades ($1,000) Net buys
Buys
Sells
Panel A: Insider trading measures including only transactions by corporate insiders QTR 4 1.646 0.740 1.480 1.623 0.777 QTR 3 1.798* 0.593 2.070** 1.808* 0.487 QTR 2 0.016 0.780 0.312 0.086 0.864 QTR 1 0.729 1.454 1.851* 0.766 1.401
1.674* 1.701* 0.283 1.787*
1.809* 1.845* 0.102 0.748
0.449 0.559 0.839 1.326
Panel B: Insider trading measures including transactions by all insiders QTR 4 1.258 0.360 0.631 1.360 QTR 3 1.313 0.718 1.661* 1.276 QTR 2 0.652 0.313 0.006 0.545 QTR 1 1.120 1.634 2.179** 1.127
1.311 1.351 0.387 2.161**
1.414 1.237 0.551 1.092
0.067 0.665 0.286 1.494
0.380 0.614 0.354 1.596
Net buys 0.808 1.684* 0.359 1.184
0.140 1.256 0.186 1.587
The Wilcoxon rank-sum test is used to compare insider trading behavior between firms with positive and negative announcement period cumulative abnormal returns. A 12-day announcement period (10, +1) is used to compute firms’ CARs using the market model. Out of the total sample firms, 15 firms have positive CARs and 74 firms have negative CARs. The test statistics for both corporate insiders and all insiders are respectively shown in Panel A and Panel B. * Significant at the 10% level. ** Significant at the 5% level.
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
bankruptcy firms. In two recent studies by Pettit and Venkatech (1995) and Pettit et al. (1996), the authors find that abnormal insider trading patterns are mainly caused by insider selling activities around corporate events with positive abnormal returns. That is, insiders attempt to make use of their private information through their trading, but in a rather passive manner. Similarly, this study shows that insiders of Chapter 11 bankruptcy firms may try to reduce their losses by purchasing less rather than by selling more in the months before the bankruptcy announcements. The above results exhibit clear patterns in the trading activities by insiders of Chapter 11 bankruptcy firms prior to the bankruptcy announcement. That is, insiders of the bankruptcy firms purchase less before the announcement than those of non-bankruptcy firms; and tests seem to be stronger for corporate insiders than for all insiders. The insignificance of many of the test statistics complements the study by Reinganum (1987), which finds little abnormal insider trading behavior before large stock price increases. However, the significant abnormal trading activities documented in this research provide new evidence about insider trading and bankruptcy announcements. The evidence supports the argument that insiders of Chapter 11 bankruptcy firms do try to take advantage of their knowledge about the forthcoming announcement through trading on their own firms’ stocks. The passive nature of their behavior suggests that insiders follow a constrained trading strategy in light of the stringent government regulations against illegal insider trading.
4. Conclusions This paper provides new evidence on the effect of Chapter 11 bankruptcy announcements on insider trading behavior. Empirical analyses are conducted for a final sample of 89 NYSE/AMEX firms that filed for Chapter 11 bankruptcies during the period of 1982 through 1990. In contrast to earlier studies, this study finds significant abnormal insider purchases between Chapter 11 bankruptcy firms and the controlled firms. Specifically, insiders of the bankrupt firms reduce their purchases significantly in the 6-month period before the bankruptcy announcement in comparison with those of the control firms. However, this type of abnormal trading behavior does not carry to the insider trading measures in sales and net purchases. Insider trading volume is observed to decline at least 1 year before the bankruptcy announcement, but the decline is statistically significant only within the 3-month period before the announcement. In addition, the empirical results show a clear pattern in trading between the bankrupt firms with negative announcement CARs and those with non-negative CARs. Compared with insiders in the non-negative CARs firms, insiders in the negative CARs firms tend to purchase less and sell more prior to the bankruptcy announcement. Efforts are also made to ex-
69
amine the trading difference between (1) corporate insiders and all insiders, and (2) insiders of firms which re-emerge from the Chapter 11 process and those which do not. In both cases, the testing results show little significant abnormal insider trading behavior. These findings provide interesting insights into the rationale and behavior of insider trading before corporate announcements. The fact that insiders of bankrupt firms appear to trade differently long before, and only significantly differently just before the bankruptcy announcement suggests that insiders do use their private information through trading but may not possess such information long before the announcement. Further, even when insiders do anticipate the market reaction to the announcement, they tend to follow a less aggressive trading strategy in light of potential penalties on illegal insider trading. The fewer significant testing results for measures in sales and net purchases are supportive of the above argument. The evidence also seems to suggest that insider trading is influenced more by the short-term announcement effect than by the long-term re-emergence issue.
Acknowledgments The author is very grateful to the editors and especially the two anonymous referees whose helpful and insightful comments and suggestions greatly improved the paper. All errors are my own responsibility.
References Altman EI, Brenner M. Information effects and stock market response to signs of firm deterioration. J Financ Quant Anal 1981;16:35 – 47. Clark TA, Weinstein MI. The behavior of the common stock of bankrupt firms. J Finance 1983;38:489 – 504. Conover WJ. Practical Nonparametric statistics. 2nd ed. New York, NY: John Wiley & Sons, 1980. Franks JR, Torous WN. An empirical investigation of U.S. firms in reorganization. J Finance 1989;44:747 – 69. Givoly D, Palmon D. Insider trading and the exploitation of insider information: some empirical evidence. J Bus 1985;58:69 – 87. Gosnell T, Keown AJ, Pinkerton JM. Bankruptcy and insider trading: difference between exchange-listed and OTC firms. J Finance 1992;47: 349 – 62. Harlow WV, Howe JS. Leveraged buyout and insider nontrading. Financ Manage 1993;22(Spring):109 – 18. Hirschey M, Zaima JK. Insider trading, ownership structure, and the market assessment of corporate sell-offs. J Finance 1989;44:571 – 95. Karpoff JM, Lee D. Insider trading before new issue announcements. Financ Manage 1991;20:18 – 26. Loderer CF, Sheehan DP. Corporate bankruptcy and managers’ self-serving behavior. J Finance 1989;44:1059 – 75. Meulbroek LK. An empirical analysis of illegal insider trading. J Finance 1992;47:1661 – 70. Morse D, Shaw W. Investing in bankrupt firms. J Finance 1988;43: 1193 – 1207. Penman SH. A comparison of the information content of insider trading and management earnings forecasts. J Financ Quant Anal 1985;20:1 – 17.
70
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
Pettit RR, Venkatech PC. Insider trading and long-run return performance. Financ Manage 1995;24:88 – 103. Pettit RR, Ma Y, He J. Do corporate insiders circumvent insider trading regulations: the case of stock repurchases? Rev Quant Finance Account 1996;7:81 – 96. Pratt JW. Remarks on zeros and ties in the Wilcoxon signed rank procedures. J Am Stat Assoc 1959;54:655 – 67. Reinganum M. The anatomy of a stock market winner. Financ Anal J 1987;43:16 – 28.
Rimbey JN, Anderson SC, Born JA. Shareholder wealth responses to bankruptcy filing announcements under the chandler and reform acts. Financ Rev 1995;30:1 – 22. Seyhun HN. Do bidder managers knowingly pay too much for target firms. J Bus 1990;63:439 – 64. Seyhun HN, Bradley M. Corporate bankruptcy and insider trading. J Bus 1997;70:189 – 216. Wilcoxon F. Some rapid approximate statistical procedures. Stamford, CT: American Cyanamid, Stamford Research Laboratories, 1949.
Insider trading behavior prior to Chapter 11 bankruptcy announcements Yulong Ma* Department of Finance, Real Estate and Law, College of Business Administration, California State University, Long Beach, 1250 Bellflower Boulevard, Long Beach, CA 90840, USA Accepted 18 December 1999
Abstract This study uses non-parametric methods to examine the difference in trading behavior between Chapter 11 bankruptcy firms and nonbankruptcy firms traded on New York Stock Exchange (NYSE)/American Stock Exchange (AMEX) exchanges. It documents new evidence about the anomaly of insider purchases rather than sales prior to Chapter 11 bankruptcy announcements. Insiders of Chapter 11 bankruptcy firms purchase significantly fewer shares than insiders of the control firms before the bankruptcy announcement. Although insider trading volume declines long before the announcement, the decline is statistically significant only during the 3-month period before the announcement. The study, however, finds no significant difference in trading behavior between insiders of firms that ceased to trade on and those that continue to trade on the NYSE/AMEX within 5 years after the bankruptcy announcements. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Insider trading; NYSE/AMEX; Chapter 11 bankruptcy
Empirical research on insider trading has shown that the significance of abnormal insider trading activities around corporate announcements may depend on the specific corporate event being investigated. Numerous studies such as Penman (1985), Hirschey and Zaima (1989), Karpoff and Lee (1991), and Pettit et al. (1996) have documented significant abnormal insider trading activities around corporate announcements. Other studies, however, find no strong associations between insider trading and corporate announcements. For example, Givoly and Palmon (1985) report no significant abnormal insider trading activity before the announcements of 11 types of corporate events. Further, Reinganum (1987) argues that insiders tend not to trade their own firms’ stocks prior to large stock price increases. Since Chapter 11 bankruptcy announcements cause large stock price declines, it is important to examine how insiders of these bankruptcy firms trade. There exists only limited empirical research on insider trading activities and corporate bankruptcies. Loderer and Sheehan (1989) study this issue by investigating whether insiders of bankrupt firms reduce their stockholdings in
* Tel.: +1-562-985-4563; fax: +1-562-985-1754. E-mail address: [email protected] (Y. Ma).
comparison with those of firms that do not file for bankruptcy. Their results show little evidence of abnormal insider trading behavior by insiders of bankrupt firms prior to the bankruptcies. Another study by Gosnell et al. (1992) uses actual insider trading transaction data to examine insider trading activities before the announcement of bankruptcy filings. They find that insiders of OTC firms increase their sales significantly over a period of 5 months before the bankruptcy announcement while the exchange-listed firms did not. A more recent study by Seyhun and Bradley (1997) examines the trading activities by insiders of New York Stock Exchange (NYSE) and American Stock Exchange (AMEX) bankrupt firms over the time period of 1975 to 1992. Their results show significant insider selling before the bankruptcy announcement with more intense selling by top executives. Should we expect abnormal insider trading activities before firms announce bankruptcy filings? By the Securities and Exchange Commission’s (SEC) definition, insiders are chairmen, directors, officers, etc., and principal shareholders with 10% or more of any equity class of securities. Corporate insiders refer to those insiders who hold managerial positions in the firm. The prevalent belief is that insiders possess superior information about their firms’ future prospects and may trade either to make profits or to convey private information to the market. Financial
0148-2963/00/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved. PII: S 0 1 4 8 - 2 9 6 3 ( 0 0 ) 0 0 11 5 - 6
64
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
agency theory suggests that insiders will be more likely to engage in self-serving activities when a firm is in financial distress. Although equity values decline significantly long before firms filing for bankruptcy, firms announcing Chapter 11 bankruptcy still experience significantly negative abnormal returns over the announcement period (e.g., Altman and Brenner, 1981; Clark and Weinstein, 1983; Morse and Shaw, 1988). Therefore, corporate insiders may be motivated to sell more stocks prior to the bankruptcy announcement. As such, insiders, theoretically, can certainly minimize their losses by reducing purchases or increasing sales of their own firms’ stock prior to the bankruptcy filings. However, trading by insiders based on material private information is against government regulations and corporate policies. Illegal insider trading may incur both high civil and criminal costs, which may deter insiders from trading before corporate announcements, especially with large price effects. This may explain the argument made by Reinganum (1987) that insiders tend not to trade their own firms’ stocks prior to large stock price increases. Meulbroek (1992) presents a detailed discussion and analysis on government regulations against illegal insider trading. Hence, whether insiders trade abnormally prior to the announcement of Chapter 11 bankruptcy may be an empirical issue. In contrast to events with large price increases, two aspects associated with insiders of bankrupt firms deserve emphasis. First, the public is usually aware of the financial difficulty of the firm before it declares bankruptcy, and its stock price usually began to decline long before the announcement. This means that insiders of these firms could use their firm’s publicly known financial distress as an excuse for selling before the bankruptcy announcement to avoid possible allegations of illegal trading. Second, since insiders in the financially distressed firms are more willing to take risks, insiders of Chapter 11 bankruptcy firms should have more incentive to engage in abnormal trading activities before the announcement. Therefore, even facing stringent government regulations, insiders of bankruptcy firms may still try to change their trading patterns in such a way so as to reduce their equity losses. The purpose of this study is to examine whether insiders of Chapter 11 bankruptcy firms change their trading patterns/behavior prior to the bankruptcy announcements to reduce their financial losses. Unlike Loderer and Sheehan (1989), who use changes in shareholdings in the proxy statements, this study uses the actual insider trading transaction data from the SEC. It differs from the study by Gosnell et al. (1992) in the following two ways. First, this paper investigates a final sample of 89 NYSE and AMEX firms over a period of 10 years. Their final sample includes only 18 NYSE and AMEX firms over a 3-year period. Second, this study focuses exclusively on firms filing Chapter 11 bankruptcy. Our study also differs from Seyhun and Bradley (1997) in sample selection and testing method. In their study,
sample firms consist of those that filed a voluntary bankruptcy petition over the period of 1975 –1992. This research includes only Chapter 11 bankruptcy firms. Further, we employ non-parametric methods to test the significance of abnormal insider trading rather than parametric methods. Seyhun (1990) argues that insider trading does not follow a normal distribution. The focus on Chapter 11 bankruptcy firms is motivated by the following reasons. First, Chapter 11 bankruptcy process, as a result of the 1978 Bankruptcy Act, gives the stockholders substantial protection against creditors. Morse and Shaw (1988) argue that the new changes may favor more reorganizations. The increasing number of firms seeking protection under Chapter 11 process is in supportive of this point of view. Franks and Torous (1989) also indicate that Chapter 11 bankruptcy process has been described as favoring corporate management and can be strategically used by managers as a new measure in its armory when the going gets tough rather than when the firm is insolvent. Therefore, the new attitude towards using Chapter 11 protection by the management may have a significant impact on insider trading behavior. Second, from an empirical point of view, the existence and distinction of Chapter 11 bankruptcy from other bankruptcy forms warrant an empirical investigation of the trading behavior by insiders of Chapter 11 bankruptcy firms. The empirical results provide new evidence of abnormal trading behavior by insiders of Chapter 11 bankruptcy firms. The finding shows that most of the significant abnormal insider trading occurred within 6 months before the bankruptcy announcement for insider purchases rather than sales. It also shows that total insider trading volume reduced significantly during the 3-month period prior to announcement. Efforts are also made to examine whether separating corporate insiders (holding managerial positions) from all insiders sample provides additional information. The results show very similar trading patterns but stronger significance levels for the corporate insiders sample in comparison with all insiders sample.
1. Data source and sample description The stock returns and the returns on the equally weighted portfolio of all NYSE and AMEX stocks are obtained from the Center for Research on Security Prices (CRSP) daily stock tapes. The insider trading data are retrieved from the Ownership Reporting System (ORS) master tape and the ORS accumulative tape. The master tape covers the period of January 1980 to August 1987, and the accumulative tape covers the period of July 1986 to March 1991. Although differing in data structure, these two tapes provide all trading transactions by insiders of the NYSE/AMEX firms over the respective time periods. Insider trading information is also available from the Official Summary of Securities Transactions and Holdings published by the SEC.
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
The insider trading tapes contain all types of transactions by insiders, that is, open market purchases and sales, private purchases and sales, exercises of options and warrants, shares acquired and disposed of through a plan, stock dividends, etc. For the purpose of this study, only open market and private purchases and sales are included in the empirical investigation. This approach is also consistent with the method used by the Value Line Investment Survey and the Wall Street Journal. The information for firms announcing filings of Chapter 11 bankruptcy petitions is collected from the Predicasts F&S Index of Corporate Changes under the section of Bankruptcy. Firms that filed petitions during the period of 1982 through 1990 constitute the initial sample. The initial sample consists of 162 announcements made by 157 firms over the 10-year period. Of all the initial sample firms, there are only 147 announcements made by 147 firms matched with the 1995 CRSP monthly file. That is, the stocks of these 147 firms were traded on the New York and American Stock Exchanges during some time in the 10-year time period. Of these firms, 42 firms have announcement dates either before their beginning trading or after their ending trading dates on the CRSP file; 12 firms have no matches with the firms on the insider trading tape (ORS). In addition, there are three firms that match the insider trading tape, but have no insider transaction data; finally, there is one firm that has abnormal insider trading pattern. The final sample of 89 firms is the result of the screening process mentioned above. Table 1 provides some of the basic descriptive statistics for the sample firms and the control firms. Panel A of the table shows the firm number and firm size distributions over the investigation period for the bankruptcy and control samples. Panel B of the table shows the announcement period information effect of the bankruptcy firms. The abnormal returns are estimated using the traditional market model in which the estimation period is 1 year before the announcement date. Over the period of 12 trading days (10, +1) prior to the announcement, the sample firms experience significantly negative cumulative abnormal returns of more than 25%. In comparison, the cumulative abnormal returns for the control sample are less than 1% and insignificant over the same 12-day period (10, +1) prior to the announcement. To examine the trading behavior by insiders of the Chapter 11 firms, the control sample is constructed to estimate the normal level of insider trading activities using the following procedure. Each of the 89 control firms is selected from the CRSP monthly file. A random search is made to match the industry and size of the control firms with those of the Chapter 11 firms as closely as possible. For each bankruptcy firm, a search is made from the beginning of the CRSP monthly file. When a firm meets the industry code, it then is screened for the size requirement. If the first run does not generate a matched firm, another run is made with wider upper and lower bounds
65
Table 1 Basic sample descriptive statistics Panel A: Firm size and distribution Year
Initial
Final
Chapter 11
Control
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Total
1 26 13 18 16 22 15 16 9 26 162
0 18 7 11 10 15 5 11 8 7 89
N.A. 171.531 165.994 149.700 125.195 207.870 62.399 136.229 43.739 259.697 161.003
N.A. 171.860 155.116 139.352 138.340 223.227 159.864 134.002 35.401 255.093 167.412
Panel B: Announcement period cumulative abnormal returns Period
Cumulative abnormal return
t-Statistics
(50, 11) (10, 2) (1, 0) +1
1.938 7.529 17.597 0.815
1.140 8.116*** 43.473*** 1.316
Panel A shows the number of firms in the initial and final samples over the 10-year period. The market value of a firm’s total equity is used as a proxy of firm size. Panel B presents the announcement period abnormal returns using the OLS market model. *** Significant at the 1% level.
for the firm size requirement. The industry is identified by the first two digits of the four-digit SIC code for the Chapter 11 firm. The market value of the firm’s total equity is used as a proxy of firm size. For the bankruptcy sample, stock prices 2 years prior to the announcements of Chapter 11 bankruptcy filings are used; while for the control sample, stock prices from the monthly file and just before the announcements are used. The average size of the Chapter 11 firms is about $161.003 million and the average size of the control firms is $167.412 million.
2. The methodology 2.1. Insider trading measures Insider trading is measured in terms of the number of transactions, the number of insiders, and the dollar amount of the transactions. The reason that these measures are employed is that in the literature of insider trading, there is still no one measure that is generally preferred to others. In addition, under each of the three measures, insider trading activities are examined separately for purchases, sales, and jointly for net purchases. The purpose for using separate analyses for purchases and sales is to examine how insiders act upon new information through purchasing and selling transactions, especially given the penalties on illegal insider trading. For instance, it is easy to argue that insiders are trading illegally if they purchase abnormally more of their own firm’s stock prior to the announcement of favorable
66
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
news. However, it is hard to make a case against insiders who sell abnormally less of their own firm’s stock prior to the announcement. The same argument holds true for the announcement of unfavorable news. The insider trading measures are computed on a monthly basis over the time period of January 1980 to March 1991. For each firm, PNit is used to denote the number of purchases made by insiders in the firm i during month t. Similarly, SNit is used to express insider sales for firm i in month t. The net purchase measures of insider trading are simply calculated as the difference between the purchase and sale measures, NPNit = PNit SNit. Given the time period covered in this study, the largest number of monthly insider trading measures is 135. These insider trading measures for Chapter 11 bankruptcy firms are compared with those for control firms over a period of 24 months before the announcements of Chapter 11 bankruptcies.
i = 1,2, . . . , N. These absolute values are ranked with the highest rank N assigned to the pair with the largest absolute difference. For pairs with the same absolute values, the average of the ranks is assigned to each of these pairs. The method used here in handling zero differences follows the one suggested by Pratt (1959). Wilcoxon (1949) discards all the zero differences before these differences are ranked. Conover (1980) provides more detailed descriptions about these procedures. The test statistic is then computed based on the signed-rank variable, Ri, which is determined in accordance with the following rule: Ri = the rank assigned to (MPNib MPNic) if Di is positive; Ri = the negative of the rank assigned to (MPNib MPNic) if Di is negative; Ri = 0 if Di = 0. The test statistic is then defined in Eq. (2) as: N P Ri i¼1 T ¼ sffiffiffiffiffiffiffiffiffiffiffi ð2Þ N P 2 Ri i¼1
2.2. Statistical methods This paper mainly relies on non-parametric approaches to examine the abnormal level of insider trading activities for Chapter 11 bankruptcy firms. Seyhun (1990) argues that the insider trading activity is non-normal, infrequent, and highly skewed in magnitude. Therefore, it seems more appropriate to use non-parametric methods. The non-parametric tests do not require the strong assumption of normality in the underlying distribution of the research variables. To compare insider trading measures of bankruptcy firms with those of control firms, the Wilcoxon signed-rank test is employed. To study the trading behavior by insiders of firms with negative and non-negative announcement period CARs, the Wilcoxon sum-rank test is used. The Wilcoxon signed-rank test is not available in SAS. The following provides a brief description of its testing procedure. For any firm i in the bankruptcy sample of N firms, the insider trading measure, PNit, is calculated for the 24 months prior to the announcement. That is, t ranges from 24 months to 1. The 24 months are split into eight non-overlapping quarterly periods. The mean values of insider trading measures for the bankruptcy sample over these quarterly periods are compared with their corresponding values for the control sample. For example, to compare insider trading measures in the first quarter for these two samples, the paired values for any sample firm are computed as in Eq. (1). MPNib ¼
1 X t¼3
PNit =3;
MPNic ¼
1 X
PNit =3
ð1Þ
t¼3
MPNib is the mean value for bankruptcy firm i, and MPNic is the mean value for the corresponding firm i in the control sample. The absolute differences of these paired values can be computed for each of the N paired-firms of these two samples. That is, |Di| = |MPNib MPNic|, where
This test statistic follows approximately a normal distribution when the number N of the paired observation is large (N > 29 in statistical sense). For our samples, the N is 89.
3. Empirical results The results are reported for both corporate insiders and all insiders. As mentioned earlier, corporate insiders refer to those insiders who possess managerial positions in the firm; while all insiders include all insider categories as defined by the SEC. The purpose of these separate analyses is to see whether corporate insiders as a group are more privy to firms’ private information will trade differently from all insiders as a whole. On the one hand, corporate insiders may be expected to trade more actively by taking advantage of their managerial positions in the firm. On the other hand, corporate insiders may be more inclined to constrain their trading activities, knowing that they are under closer scrutiny, particularly with regard to Chapter 11 bankruptcy announcements. From the investing public’s perspective, it is useful and important to see, empirically, whether trading patterns by corporate insiders convey significantly better information than those by all insiders. Although all insider trading analyses are conducted for eight quarterly periods prior to the announcements, however, only the testing results for the first four quarters are reported due to the insignificance of results from other quarters. Table 2 presents the results of the Wilcoxon signed-rank test that compares insider trading patterns between the bankruptcy sample and its control sample. All figures in the table represent the test statistics for the difference in insider trading between these two samples. Panel A of the table reports the results for corporate insiders and Panel B reports the results for all insiders. For both corporate
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
67
Table 2 Insider trading comparison between the Chapter 11 sample and the control sample Number of transactions Period
Buys
Sells
Number of insiders Net buys
Panel A: Insider trading measures including only transactions QTR 4 0.407 0.085 0.345 QTR 3 1.222 0.316 0.123 QTR 2 1.960** 0.298 1.277 QTR 1 2.143** 1.001 0.167
Buys
Sells
by corporate insiders 0.556 0.839 1.162 0.234 1.891* 0.012 2.176** 1.400
Panel B: Insider trading measures including transactions by all insiders QTR 4 0.182 0.294 0.372 0.059 QTR 3 0.597 0.182 0.242 0.663 QTR 2 1.882* 0.764 1.620 1.762* QTR 1 1.960** 0.793 0.085 2.150**
0.588 0.048 0.737 1.245
Dollar amount of trades ($1,000) Net buys
Buys
Sells
Net buys
0.031 0.134 1.245 0.205
0.787 1.293 1.830* 2.226**
1.185 0.723 0.175 1.824*
0.146 0.363 0.501 0.719
0.070 0.096 1.475 0.175
0.288 0.649 1.764* 2.139**
0.931 0.837 0.780 1.444
0.117 1.132 0.799 0.274
The figures in the table are Wilcoxon signed-rank test statistics for the difference in insider trading between Chapter 11 bankruptcy firms and controlledsample firms. The Wilcoxon test is applied for all nine difference insider trading measures over a four-quarters period prior to the announcement of Chapter 11 filings. Panel A shows the results for only corporate insiders who have a managerial position in the firm while Panel B provides the results including all insiders defined by the SEC. * Significant at the 10% level. ** Significant at the 5% level.
insiders and all insiders, the results show that there is no significant difference in insider trading between these two samples during the third and fourth quarters before Chapter 11 announcements. During the first and second quarters, only the testing results for insider buys are significant. The Wilcoxon test statistics are significant at the 5% level for insider buys measures in the first quarter, and are significant at the 10% level for insider buys measures in the second quarter. The negative value means that the insider trading measure for the bankruptcy sample is smaller than that for the control sample. As shown in the table, all of the statistics in the Buys column are negative. These results indicate that during the two quarters prior to the Chapter 11 announcement, insiders in the bankruptcy firms purchased significantly less stock than those in the control firms. This implies that insiders may in fact have already incorporated the upcoming unfavorable news in trading on their own firm’s stock. However, this argument does not carry over to insider sales and net buys measures. In contrast to the popular perception that insiders of bankruptcy firms may sell their stock before the announcement, none of the test statistics are significant at a reasonable level. The result seems to suggest that insiders of bankrupt firms do not engage in active trading schemes in anticipation of the announcement of Chapter 11 bankruptcies. Rather, they adopt passive trading tactics by restraining their purchasing activities. Harlow and Howe (1993) find similar results in their study. A comparison of Panel A with Panel B shows that all statistics except one for all insiders are larger in absolute value than those for corporate insiders though the differences are very small. This shows that although corporate insiders have higher significant levels, they, in general, probably do not behave significantly differently in trading from the rest of insiders before the firm files for Chapter 11 bankruptcy.
To further examine insider trading behavior, the total volume of insider purchases and sales from the bankruptcy sample is compared with that from the control sample using the Wilcoxon signed-rank test. The total volume is defined as the sum of insider purchase and insider sale measures. Table 3 contains the testing results. All of the figures except one in the table are negative, indicating that the trading volume is lower for the bankruptcy sample than for the control sample over the four quarters before the Chapter 11 announcements. However, basically, only those numbers in the first quarter are statistically significant at the 1% or 5% levels (one of the exceptions is in Panel B).
Table 3 Trading volume comparison between the Chapter 11 and the control samples Period
Transactions
Insiders
Dollar amount
Panel A: Transactions including only corporate insiders QTR 4 0.259 1.391 QTR 3 0.823 0.764 QTR 2 1.685* 1.446 QTR 1 2.178** 2.823***
1.900* 1.210 0.713 3.223***
Panel B: Transactions including all insiders QTR 4 0.367 0.724 QTR 3 0.598 0.219 QTR 2 1.079 1.098 QTR 1 1.553 2.494**
1.439 1.439 0.969 2.698***
The Wilcoxon signed-rank test is conducted for the difference in trading volume between firms in the Chapter 11 and the control samples. Trading volumes are measured as the sum of buys and sells in terms of the number of transaction, the number of insiders, and the dollar amount. The test statistics are shown for trading activities made by corporate insiders (Panel A) only and by all insiders (Panel B). * Significant at the 10% level. ** Significant at the 5% level. *** Significant at the 1% level.
68
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
In both Panel A and Panel B, insider trading volume is measured in numbers of transactions, numbers of insiders, and dollar amount. Overall, there is not much difference between Panel A and Panel B, although the test statistics are slightly higher in absolute value for corporate insiders than for all insiders. The only significant difference between these two panels occurs when the Wilcoxon statistic in terms of number of transactions turns from significant for corporate insiders to insignificant for all insiders. These results appear to indicate that insiders of bankrupt firms reduce their trading volume long before the bankruptcy announcements. However, the decline in trading volume is statistically significant only in the quarter just prior to the announcements when compared with that by insiders of the control firms. The findings are consistent with those from Table 2. That is, insiders do not increase their selling activities, but rather reduce their purchasing activities prior to the Chapter 11 announcements. Table 4 provides some evidence of the relationship between insider trading activities and announcement information effect among the Chapter 11 bankruptcy firms. The main empirical issue is whether insiders of firms with different levels of announcement period abnormal returns trade differently before the announcements. The purpose is to see whether we can differentiate Chapter 11 bankruptcy firms based on their insider trading activities. Among the 89 sample firms, 74 firms have negative announcement period abnormal returns over a 12-day announcement period (10, +1). The Wilcoxon sum-rank test is used to test the mean difference in insider trading between firms with non-negative and negative announcement period CARs. Panel A of the table contains the test statistics for corporate insiders and Panel B contains those for all insiders. During QTR 4 and QTR 3, there are more figures in Panel A significant at the 10% level than in Panel B (only
one test). During QTR 1, only two statistics under Net Buys are significant at the 10% level for Panel A, and at the 5% level for Panel B. This trading pattern seems to suggest that corporate insiders’ trading behavior is not quite the same as non-corporate insiders’. A positive value in the table indicates that the insider trading measure in the quarter for firms with non-negative CARs is large than that for firms with negative CARs. For a negative value, the opposite holds true. These findings indicate that among Chapter 11 bankruptcy firms, insiders seem to control their trading activities based on their private information contained in the forthcoming announcements. In particular, correctly anticipating the market reaction, insiders of firms with negative announcement effects seem to purchase less than insiders of firms with positive announcement effects. An effort has also been made to determine whether insiders of firms that re-emerge from the Chapter 11 process trade differently from those of firms that do not. In their recent study, Rimbey et al. (1995) find that the market can discriminate between firms that are ultimately worthless and those that may retain some shareholder value. The bankruptcy sample is split into two groups. One group contains all firms that ceased to trade on the NYSE/AMEX either shortly after or within 5 years after the announcement of Chapter 11 bankruptcies. The remaining firms constitute the second group. There are 65 firms in the first group and 24 firms in the second group. The Wilcoxon sum-rank test is used to test the mean difference in insider trading between insiders in the first group and those in the second group. The test fails to provide significant results to support the argument since, basically, none of the Wilcoxon statistics are significant at a reasonable level. The evidence presented above is, in general, consistent with previous studies on insider trading, and it also helps shed new light on trading behavior by insiders of Chapter 11
Table 4 Insider trading comparison between positive and negative announcement period CAR firms Number of transactions Period
Buys
Sells
Number of insiders Net buys
Buys
Sells
Dollar amount of trades ($1,000) Net buys
Buys
Sells
Panel A: Insider trading measures including only transactions by corporate insiders QTR 4 1.646 0.740 1.480 1.623 0.777 QTR 3 1.798* 0.593 2.070** 1.808* 0.487 QTR 2 0.016 0.780 0.312 0.086 0.864 QTR 1 0.729 1.454 1.851* 0.766 1.401
1.674* 1.701* 0.283 1.787*
1.809* 1.845* 0.102 0.748
0.449 0.559 0.839 1.326
Panel B: Insider trading measures including transactions by all insiders QTR 4 1.258 0.360 0.631 1.360 QTR 3 1.313 0.718 1.661* 1.276 QTR 2 0.652 0.313 0.006 0.545 QTR 1 1.120 1.634 2.179** 1.127
1.311 1.351 0.387 2.161**
1.414 1.237 0.551 1.092
0.067 0.665 0.286 1.494
0.380 0.614 0.354 1.596
Net buys 0.808 1.684* 0.359 1.184
0.140 1.256 0.186 1.587
The Wilcoxon rank-sum test is used to compare insider trading behavior between firms with positive and negative announcement period cumulative abnormal returns. A 12-day announcement period (10, +1) is used to compute firms’ CARs using the market model. Out of the total sample firms, 15 firms have positive CARs and 74 firms have negative CARs. The test statistics for both corporate insiders and all insiders are respectively shown in Panel A and Panel B. * Significant at the 10% level. ** Significant at the 5% level.
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
bankruptcy firms. In two recent studies by Pettit and Venkatech (1995) and Pettit et al. (1996), the authors find that abnormal insider trading patterns are mainly caused by insider selling activities around corporate events with positive abnormal returns. That is, insiders attempt to make use of their private information through their trading, but in a rather passive manner. Similarly, this study shows that insiders of Chapter 11 bankruptcy firms may try to reduce their losses by purchasing less rather than by selling more in the months before the bankruptcy announcements. The above results exhibit clear patterns in the trading activities by insiders of Chapter 11 bankruptcy firms prior to the bankruptcy announcement. That is, insiders of the bankruptcy firms purchase less before the announcement than those of non-bankruptcy firms; and tests seem to be stronger for corporate insiders than for all insiders. The insignificance of many of the test statistics complements the study by Reinganum (1987), which finds little abnormal insider trading behavior before large stock price increases. However, the significant abnormal trading activities documented in this research provide new evidence about insider trading and bankruptcy announcements. The evidence supports the argument that insiders of Chapter 11 bankruptcy firms do try to take advantage of their knowledge about the forthcoming announcement through trading on their own firms’ stocks. The passive nature of their behavior suggests that insiders follow a constrained trading strategy in light of the stringent government regulations against illegal insider trading.
4. Conclusions This paper provides new evidence on the effect of Chapter 11 bankruptcy announcements on insider trading behavior. Empirical analyses are conducted for a final sample of 89 NYSE/AMEX firms that filed for Chapter 11 bankruptcies during the period of 1982 through 1990. In contrast to earlier studies, this study finds significant abnormal insider purchases between Chapter 11 bankruptcy firms and the controlled firms. Specifically, insiders of the bankrupt firms reduce their purchases significantly in the 6-month period before the bankruptcy announcement in comparison with those of the control firms. However, this type of abnormal trading behavior does not carry to the insider trading measures in sales and net purchases. Insider trading volume is observed to decline at least 1 year before the bankruptcy announcement, but the decline is statistically significant only within the 3-month period before the announcement. In addition, the empirical results show a clear pattern in trading between the bankrupt firms with negative announcement CARs and those with non-negative CARs. Compared with insiders in the non-negative CARs firms, insiders in the negative CARs firms tend to purchase less and sell more prior to the bankruptcy announcement. Efforts are also made to ex-
69
amine the trading difference between (1) corporate insiders and all insiders, and (2) insiders of firms which re-emerge from the Chapter 11 process and those which do not. In both cases, the testing results show little significant abnormal insider trading behavior. These findings provide interesting insights into the rationale and behavior of insider trading before corporate announcements. The fact that insiders of bankrupt firms appear to trade differently long before, and only significantly differently just before the bankruptcy announcement suggests that insiders do use their private information through trading but may not possess such information long before the announcement. Further, even when insiders do anticipate the market reaction to the announcement, they tend to follow a less aggressive trading strategy in light of potential penalties on illegal insider trading. The fewer significant testing results for measures in sales and net purchases are supportive of the above argument. The evidence also seems to suggest that insider trading is influenced more by the short-term announcement effect than by the long-term re-emergence issue.
Acknowledgments The author is very grateful to the editors and especially the two anonymous referees whose helpful and insightful comments and suggestions greatly improved the paper. All errors are my own responsibility.
References Altman EI, Brenner M. Information effects and stock market response to signs of firm deterioration. J Financ Quant Anal 1981;16:35 – 47. Clark TA, Weinstein MI. The behavior of the common stock of bankrupt firms. J Finance 1983;38:489 – 504. Conover WJ. Practical Nonparametric statistics. 2nd ed. New York, NY: John Wiley & Sons, 1980. Franks JR, Torous WN. An empirical investigation of U.S. firms in reorganization. J Finance 1989;44:747 – 69. Givoly D, Palmon D. Insider trading and the exploitation of insider information: some empirical evidence. J Bus 1985;58:69 – 87. Gosnell T, Keown AJ, Pinkerton JM. Bankruptcy and insider trading: difference between exchange-listed and OTC firms. J Finance 1992;47: 349 – 62. Harlow WV, Howe JS. Leveraged buyout and insider nontrading. Financ Manage 1993;22(Spring):109 – 18. Hirschey M, Zaima JK. Insider trading, ownership structure, and the market assessment of corporate sell-offs. J Finance 1989;44:571 – 95. Karpoff JM, Lee D. Insider trading before new issue announcements. Financ Manage 1991;20:18 – 26. Loderer CF, Sheehan DP. Corporate bankruptcy and managers’ self-serving behavior. J Finance 1989;44:1059 – 75. Meulbroek LK. An empirical analysis of illegal insider trading. J Finance 1992;47:1661 – 70. Morse D, Shaw W. Investing in bankrupt firms. J Finance 1988;43: 1193 – 1207. Penman SH. A comparison of the information content of insider trading and management earnings forecasts. J Financ Quant Anal 1985;20:1 – 17.
70
Y. Ma / Journal of Business Research 54 (2001) 63 – 70
Pettit RR, Venkatech PC. Insider trading and long-run return performance. Financ Manage 1995;24:88 – 103. Pettit RR, Ma Y, He J. Do corporate insiders circumvent insider trading regulations: the case of stock repurchases? Rev Quant Finance Account 1996;7:81 – 96. Pratt JW. Remarks on zeros and ties in the Wilcoxon signed rank procedures. J Am Stat Assoc 1959;54:655 – 67. Reinganum M. The anatomy of a stock market winner. Financ Anal J 1987;43:16 – 28.
Rimbey JN, Anderson SC, Born JA. Shareholder wealth responses to bankruptcy filing announcements under the chandler and reform acts. Financ Rev 1995;30:1 – 22. Seyhun HN. Do bidder managers knowingly pay too much for target firms. J Bus 1990;63:439 – 64. Seyhun HN, Bradley M. Corporate bankruptcy and insider trading. J Bus 1997;70:189 – 216. Wilcoxon F. Some rapid approximate statistical procedures. Stamford, CT: American Cyanamid, Stamford Research Laboratories, 1949.