Insider trading and performance of seasoned equity offering firms after controlling for exogenous trading needs

Transcription

1 The Quarterly Review of Economics and Finance 42 (2002) Insider trading and performance of seasoned equity offering firms after controlling for exogenous trading needs Inmoo Lee* Department of Business Administration, Korea University, Seoul , Korea Abstract Insiders trade not only because they have private information about their companies but also because of other exogenous reasons. Therefore, it is important to control for exogenous trading needs in empirical studies regarding insider trading. Lee (1997) shows that insider trading is not closely related to the long-term performance of primary seasoned equity offering firms. This paper examines whether the results hold after controlling for exogenous needs to trade by using an inequality test with instrumental-variables technique Board of Trustees of the University of Illinois. All rights reserved. JEL classification: G000, G140, G300 Keywords: Insider trading; Exogenous trading needs; Inequality test with instrumental variables 1. Introduction Insiders trade not only because they have private information about their companies future performance but also because of many other exogenous needs to trade (e.g., house down payments, college tuition payments for the kids, payments related to divorce settlements, seed money for a new business, and hostile takeover defenses). For example, Hulbert discusses the problems related to an investment strategy using inside trading information, and emphasizes that insider trading is motivated not only by inside information about the * Tel.: ; fax: address: inmoo@korea.ac.kr (I. Lee) /02/$ see front matter 2002 Board of Trustees of the University of Illinois. All rights reserved. PII: S (01)

2 60 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) firm but also by personal reasons not related to special knowledge about the firm [Forbes (August 15, 1994, page 145)]. However, most empirical studies in the insider trading literature [Jaffe (1974), Finnerty (1976), Elliot, Morse, and Richardson (1984), Givoly and Palmon (1985), Seyhun (1986), Rozeff and Zaman (1988), Lin and Howe (1990), John and Lang (1991), and Lee (1997) among others] do not control for exogenous trading needs. This is not because they do not recognize the importance of exogenous trading needs, but because it is hard to control for unobservable exogenous trading needs. Here, an inequality test with instrumental variable method (ITIV) is introduced as one way to control for exogenous trading needs in empirical studies regarding insider trading. As an example, the ITIV method is used to control for exogenous trading needs in the examination of the relation between pre-issue insider trading and the post-issue long-run performance of primary seasoned equity issuing companies. Lee (1997) shows that there is no close relation between insider trading and the long-run performance of primary seasoned equity issuing firms. Lee points out that the poor relation might be due to the fact that exogenous trading needs are not controlled for in the study. It is shown that the ITIV method can be used to examine this issue. The rest of the paper is organized as follows. Section 2 describes the data and summary statistics, and Section 3 presents the inequality test with instrumental variables methodology. Finally, Section 4 summarizes the results and presents some concluding remarks. 2. Data and summary statistics The sample consists of primary seasoned equity offerings during that have been recorded by the Securities Data Company (SDC). Similar to Lee (1997), the following firms are excluded from the original SEO sample: 1) firms which do not have price records on the CRSP daily NYSE/Amex or Nasdaq tapes on their offering date; 2) utility companies (SIC codes ), closed-end mutual funds (SIC codes ) and real estate investment trusts (REITs: SIC code 6798); 3) firms which do not have any insider trading for the 6 months ending on, and including, the issue date; and 4) firms that do not have book equity value for the fiscal year before their SEOs in any of the following data sources: COMPUSTAT, Moody s various manuals, and Standard and Poor s Stock Report. An SEO is also excluded if more than half of the total offering consists of secondary shares (shares sold by existing shareholders). Finally, any SEOs by the firms that have issued seasoned shares during the prior 36-month period are excluded. 1,281 SEOs are in the sample. Fig. 1A presents the annual volume of SEOs in the sample for each year during Note that there are large variations in the volume of SEOs during this sample period. The insider trading data is from the Securities and Exchange Commission s (SEC) Ownership Reporting System (ORS) data file that contains all transactions by insiders subject to disclosure according to the Securities and Exchange Act of Insiders are defined as officers, directors and controlling persons, and insider trading activity includes both open market and private transactions. Five prior return, size and book-to-market equity ratio-adjusted matching firms are used as a benchmark. The same selection procedures as those described in Lee (1997) are used.

3 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Fig. 1. The number of seasoned equity offerings (SEOs) during in the sample. The sample consists of 1,281 seasoned equity offerings (SEOs) of pure primary or combined primary and secondary shares (with the primary proportion at least 50% of the total) by firms listed on the CRSP NYSE/Amex and Nasdaq tapes. Offerings by utilities (SIC codes ), closed-end mutual funds, REITs, firms with any SEOs during the prior 3-year period, firms with no insider trading over the 6-month period ending on the issue date, and firms with no book value information in COMPUSTAT, Moody s manuals, or Standard and Poor s Stock Report, are excluded. The cut-off points for size are based on the last day of the previous month (one month before issue) market capitalization of all CRSP-listed NYSE/Amex firms and the cut-off points for BM, the book-to-market ratio of equity, are based on the last day of the previous month BM of all CRSP-listed NYSE/Amex and Nasdaq firms. Fig. 1B presents the distribution of SEO firms across size and book-to-market equity ratio (BM) quintiles. Notice that most issuing firms are in the lower BM quintiles (growth firms) and the middle size quintiles. 3-year buy-and-hold returns (BHRs) are used to measure the long-term performance. The

4 62 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) BHRs are calculated by compounding daily returns over either 756 trading days or the number of trading days from the offering date until the delisting date, whichever is smaller. The same holding periods are used to calculate the BHRs of matching firms. If a matching firm is delisted before the end of the 3-year anniversary or the SEO firm s delisting day, whichever is earlier, the CRSP value-weighted return 1 is spliced into the calculation of the BHR from the removal date. Table 1 reports summary statistics. There are 569 NYSE/Amex-listed firms and 712 Nasdaq-listed firms in the sample. NYSE/Amex-listed issuing firms average market capitalization ($1,299 million) is much larger than the average of Nasdaq-listed issuing firms ($237 million) and the average book-to-market equity ratio (BM) of NYSE/Amex-listed firms is higher than the average BM of Nasdaq-listed firms. The average one-year holding period return of SEO firms before issuing [86.2%: median (60.0%)] is extremely high, which is consistent with previous studies. 3-year buy-and-hold returns (BHR) of SEO firms and matching firms for different exchanges (for a given SEO firm, its matching firm may or may not be from the same exchange) are also reported in Table 1. Consistent with previous studies [e.g., Loughran and Ritter (1995) and Lee (1997)], the results show that SEO firms significantly underperform their matching firms. The wealth relatives (WR), defined as {1 average BHR of SEO firms} divided by {1 average BHR of matching firms}, are less than 1. We can observe noticeable differences in WRs between NYSE/Amex-listed companies and Nasdaq-listed companies. The average WR of NYSE/Amex-listed companies is 0.94 and that of Nasdaq-listed companies is The smaller average 3-year WR of Nasdaq-listed firms may result from the fact that more informational asymmetry is present in smaller and younger companies, most of which are listed on the Nasdaq exchange. 2 Table 2 reports the summary statistics of insider trading before and after SEOs by both issuing firms and matching firms. For the 6-month period prior to, and including issue dates, SEO firms insiders bought 1.81 times and sold 5.53 times, whereas the matching firms insiders bought 1.34 times and sold 3.73 times. While the average ratio of purchase transaction 3 [i.e., the number of purchases/(the number of purchases the number of sales)] of SEO firms (0.32) is larger than that of matching firms (0.29), the median of SEO firms (0.12) is smaller than the median of matching firms (0.22). The Wilcoxon rank-sum test result shows that the ratio of issuing firms is statistically significantly lower than the ratio of matching firms, implying that even though the prior annual returns of issuing firms and those of matching firms are similar by construction, the issuers purchase transactions proportion is significantly smaller than the matching firms purchase proportion. 4 This suggests that on average, the issuing firms insiders consider their shares as more overvalued than their matching firms insiders do. Still, the difference is surprisingly small. By comparing the prior and post insider trading of SEO firms, we notice that the average number of sell transactions has significantly dropped, while the average purchase ratio has significantly increased after issuing. Moreover, the insider trading patterns of SEO firms after issuing are not statistically significantly different from those of matching firms. Again, this is consistent with managers intentionally timing their issues. These results, however, raise the question of whether insiders really know how much their shares are overvalued. Both Loughran and Ritter (1995) and Spiess and Affleck-Graves (1995) show that even one year

5 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Table 1 Summary statistics: mean and median NYSE/Amex Nasdaq Total # of SEOs ,281 Market Capitalization 1, (in million) a [345] [106] [163] B/M ratio b [0.59] [0.38] [0.46] SEO Prior 78.2% 92.6% 86.2% 1-year Return c [58.6%] [60.7%] [60.0%] MF Prior 69.8% 80.9% 76.0% 1-year return d [53.2%] [56.7%] [55.1%] SEO Post 48.2% 24.8% 35.2% 3-year Return e [21.7%] [0.0%] [10.1%] MF Post 58.0% 47.4% 52.3% 3-year Return f [53.4%] [35.9%] [43.7%] Wealth Relatives [0.79] [0.74] [0.77] The sample consists of 1,281 seasoned equity offerings (SEOs) of pure primary or combined primary and secondary shares (with the primary proportion at least 50% of the total) during by firms listed on the CRSP NYSE/Amex and Nasdaq tapes. Offerings by utilities (SIC codes ), closed-end mutual funds, REITs, firms with any SEOs during the prior 3-year period, firms with no insider trading over the 6-month period before issues (including issue dates), and firms with no book value information in COMPUSTAT, Moody s manuals, or Standard and Poor s Stock Report, are excluded. At the time of a new issue, each SEO firm is matched with 5 seasoned firms which have the closest prior annual returns among the firms in the same size and BM quintiles. If any of these matching firms are delisted prior to the end of the corresponding SEO firms holding periods, the CRSP value-weighted returns are spliced into the calculation of the returns from the removal dates. The mean number is reported on top and the median number is in the bracket. a The market capitalization of an SEO firm is calculated by multiplying the number of shares outstanding after issue by the offering date closing price, and converted into dollars of 1993 purchasing power using the US consumer price index (CPI). b The book to market ratio (BM) ofanseofirm is calculated by dividing the book equity value after issue by the market value of equity at the close of the issue day. c The prior return of an SEO firm is the raw buy-and-hold return of the SEO firm for the 252 trading days ending on the day before the issue date, or the number of days between the first listed date and the day before the issue date, whichever is smaller. d The prior return of matching firms is the average of the raw buy-and-hold return of 5 matching firms for the same holding period as the corresponding SEO firm s. The median is the median of 1,281 mean prior returns. e The 3 year buy-and-hold % return (BHR) ofseofirms. If an SEO firm is delisted before its 3rd anniversary, the buy-and-hold return from the issue date to the delisting date is used as the BHR. f The 3 year buy-and-hold % returns of matching firms. The daily returns are compounded over the same holding periods as the corresponding SEOs holding periods. If any of matching firms are delisted prior to the end of the corresponding SEO firm s holding period, the CRSP value-weighted index returns are spliced into the calculation of the returns from the removal date. after issuing, the shares of SEO firms are substantially overvalued, indicating that the optimal insider purchase ratio should be lower even after issue. 5 Table 3 reports the different characteristics of SEO firms depending on the previous insider trading. As in Lee (1997), a pure insider trading criterion is used to classify SEOs into buyer vs. seller groups. Pure insider purchasing firms (PP) are companies with insiders only

6 64 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Table 2 Insider trading Summary statistics of insider trading SEO Average [median] ratio a Average [median] #of purchases Average [median] #of sales Matching Average [median] ratio Average [median] #of purchases Average [median] #of sales Prior [0.12] [1.00] [3.00] [0.22] [0.80] [2.80] Post [0.50] [0.00] [2.00] [0.29] [0.60] [1.60] Wilcoxon Rank Sum Test results: Z-statistics (two-sided P-values are in parentheses) Prior vs. post SEO vs. matching (ratio only) Ratio # of # of sales Prior Post purchases (0.00) (0.03) (0.00) (0.00) (0.23) The sample consists of 1,281 seasoned equity offerings (SEOs) of pure primary or combined primary and secondary shares (with the primary proportion at least 50% of the total) during by firms listed on the CRSP NYSE/Amex and Nasdaq tapes. Offerings by utilities (SIC codes ), closed-end mutual funds, REITs, firms with any SEOs during the prior 3-year period, firms with no insider trading over the 6-month period before issues (including issue dates), and firms with no book value information in COMPUSTAT, Moody s manuals, or Standard and Poor s Stock Report, are excluded. At the time of a new issue, each SEO firm is matched with 5 seasoned firms which have the closest prior annual returns among the firms in the same size and BM quintiles. Insiders are defined as officers, directors and controlling persons and insider trading activity includes both open market and private transactions. All insider trading during the 6-month period ending on the issue date is defined as the prior insider trading and all insider trading during the 6-month period starting on the day after the issue date is defined as the post insider trading. a Ratio is defined as (number of Purchases)/(number of Purchases number of Sales). If there is no insider trading, the ratio is set equal to 0.5. purchasing their shares without any selling for the previous 6-month period including issue dates. Pure insider selling firms (PS) are similarly defined. Firms that are neither PP nor PS are classified as mixed insider trading firms (MT). Even though the average prior annual return of buyers is high (63.4%), the average of sellers is much higher (97.1%). This is consistent with the conjecture that after a huge price run-up, insiders will want to sell more either to diversify their portfolios or to lock up the profits. On average, there is no significant relative (relative to their matching firms) performance differences among the firms in different insider trading groups. Fig. 2 shows the average annual returns of both SEO firms and their matching firms during the 4 years before and after issue. Panel A presents the average annual returns of buyers and Panel B presents those of sellers. Both buyers and sellers do not significantly underperform their matching firms during the first year, but they start to do so during the second year. The average annual return of buyers is 9% per year, compared to 14% for their matching firms,

7 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Table 3 The long-run performance and insider trading of SEO firms NYSE/Amex Nasdaq TOTAL PP MT PS PP MT PS PP MT PS SEO BHR a 33.7% 55.1% 47.4% 22.7% 33.2% 19.4% 27.2% 43.6% 31.5% MF BHR b 59.3% 60.7% 55.2% 37.5% 55.8% 45.1% 46.3% 58.1% 49.5% WR c SEO PR d 51.9% 76.3% 89.3% 71.2% 90.0% 103.0% 63.4% 83.4% 97.1% MF PR e 46.1% 69.5% 78.5% 65.0% 76.5% 90.4% 57.3% 73.2% 85.2% Market Cap 1,714 1, $ millions f # of SEOs The sample consists of 1,281 seasoned equity offerings (SEOs) of pure primary or combined primary and secondary shares (with the primary proportion at least 50% of the total) during by firms listed on the CRSP NYSE/Amex and Nasdaq tapes. Offerings by utilities (SIC codes ), closed-end mutual funds, REITs, firms with any SEOs during the prior 3-year period, firms with no insider trading over the 6-month period before issuing (including issue dates), and firms with no book value information in COMPUSTAT, Moody s manuals, or Standard and Poor s Stock Report, are excluded. Pure insider purchases firms (PP) are companies with only insider purchases and no insider sales for the previous 6-month period including issue dates. Pure insider sales firms (PS) are similarly defined. Firms without PP or PS are classified as mixed insider trading firms (MT). a 3-year buy-and-hold % return (BHR)ofSEOfirms. If an SEO is delisted before its 3-year anniversary, an SEO BHR is the buy-and-hold return from the issue date to the delisting date. b 3-year buy-and-hold % returns of matching firms. The daily returns are compounded over the same holding periods as the corresponding SEOs holding periods. At the time of a new issue, each SEO firm is matched with 5 seasoned firms (listed on the CRSP for at least five years) which have the closest prior annual returns among the firms in the same size and B/M quintiles. If any of these matching firms are delisted prior to the end of the corresponding SEO firm s holding period, the CRSP value-weighted index returns are spliced into the calculation of the returns from the removal date. The universe of firms from which matching firms are picked includes all operating companies (excluding utilities) listed on the NYSE/Amex and Nasdaq CRSP tapes, that have not conducted any equity issue during the prior 5-year period. c The wealth relatives (WR) are defined as [(1 average SEO firms BHR)/(1 average matching firms BHR)]. d The prior return of an SEO firm is the raw buy-and-hold return of the SEO firm for the 252 trading days ending on the day before the issue date, or the number of days between the first listed date and the day before the issue date, whichever is smaller. e The prior return of matching firms is the average of the raw buy-and-hold returns of 5 matching firms for the same holding period as the corresponding SEO firm s. f The market capitalization of an SEO firm is calculated by multiplying the number of shares outstanding after issue by the offering date closing price, and converted into dollars of 1993 purchasing power using the US consumer s price index (CPI). during the 3 years following the offering. For sellers, the average annual return is 11% per year, compared to 15% for their matching firms. The results indicate that there is no strong relation between insider trading during the 6 months ending at the time of the SEO and the long-run stock price performance of SEO firms. This poor performance of insider trading as a signal of the future prospects can be attributed to the fact that consumption shocks are not controlled for in this section. To examine this possibility, an inequality test with instrumental variables is used in the following section.

8 66 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Fig. 2. Annual returns for SEO firms and matching firms. The sample consists of 1,281 seasoned equity offerings (SEOs) of pure primary or combined primary and secondary shares (with the primary proportion at least 50% of the total) during by firms listed on the CRSP NYSE/Amex and Nasdaq tapes. Offerings by utilities (SIC codes ), closed-end mutual funds, REITs, firms with any SEOs during the prior 3-year period, firms with no insider trading over the 6-month period before issuing (including issue dates), and firms with no book value information in COMPUSTAT, Moody s manuals, or Standard and Poor s Stock Report, are excluded. Pure insider purchases firms are companies with only insider purchases and no insider sales for the previous 6-month period including issue dates. Pure insider sales firms are similarly defined. There are 224 pure purchasing SEOs and 589 pure selling SEOs in the sample. The offering date is the beginning of year 0. Matching firms are non-issuers chosen on the basis of offer date market capitalization, book-to-market equity ratio, and prior annual return. 3. Statistical tests with control for exogenous needs to trade In this section, an inequality test with instrumental variables methodology (ITIV) is introduced to examine whether or not the poor relation between insider trading and the

9 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) long-run performance of issuing firms is caused by the fact that exogenous trading needs are not controlled for in the previous section. Boudoukh, Richardson and Smith (1993: BRS hereafter) and the references therein describe this ITIV methodology in detail. BRS test the positivity of the ex ante risk premium by employing an instrumental variables approach to take into account the unobservability of expected returns. Similar to BRS, an instrumental variables approach is used to control for unobservable consumption shocks. In this section, it is examined whether buyers outperform their matching firms. If only those insiders in the SEO firms with good quality projects buy shares, the average performance of buyers will be better than their matching firms average performance. When we do not control for consumption shocks, the outperformance of SEO firms with prior insider purchases corresponds to the following hypothesis. H 0 : E SEO BHR MF BHR AR 0 H A : E SEO BHR MF BHR AR R where SEO BHR is the 3-year holding period return of a buyer and MF BHR is the average BHR of the matching firms. R represents the real domain. To control for consumption shocks, both sides of the equation are multiplied by nonnegative instrumental variables representing each factor of the consumption shocks. H 0 : E SEO BHR MF BHR Z AR Z 0 H A : E SEO BHR MF BHR Z AR Z R N Z is a vector with N nonnegative instrumental variables. Note that the inequality relation holds since nonnegative instrumental variables are multiplied in both sides. Instrumental variables that proxy for a big consumption shock are chosen because the insiders with big consumption shocks (i.e., exogenous needs to sell their shares) will not purchase their firms shares unless they have favorable inside information. By using instrumental variables, we can focus on the firms with insiders who are more likely to have bought the stocks because of the inside information regarding the future prospects of the company, rather than some exogenous needs to trade. The instrumental variables used in the test are listed in Table 4. First, the exogenous needs to sell will be related to diversification needs. As diversification needs increase, insiders current consumption needs also increase. The previous good performance of the company relative to the market would suggest that the portion of wealth invested into the company s stock increased more than was desired. Two instrumental variables, APR {SEO firm s prior annual return (SEO PR) CRSP value-weighted index s prior annual return (VWPR)} and SAPR {APR standardized by variances: SEO PR/Var(SEO PR) VWPR/Var(VWPR)}, are used to measure the relative performance of SEO firms and the market. Second, the exogenous need to sell is a function of insiders unexpected consumption needs. For example, unexpected medical expenses would increase the exogenous need to sell. Since it is difficult to believe that an unexpected consumption need will occur to all insiders in a company, the total number of insider transactions is used as a proxy variable for

10 68 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Table 4 Instrumental variables used in the inequality test Factor Instrumental Variables 1) Diversification APR [SEO PR b VWPR c ] SAPR [SEO PR/Var(SEO PR) d 2) Emergencies NIT Total # of VWPR/Var(VWPR)] Expected a Relation Value of Instrument Positive Z* i 1ifZ i Z otherwise Positive Z* i 1ifZ i Z otherwise Negative Z* i 1ifZ i Z otherwise Negative Z* 1 if there is no insider transactions 3) Control ANTI Any takeover action e i announcement 0 otherwise Table shows the list of instrumental variables used in the inequality test described in Section 3. Each instrumental variable proxies for a big exogenous consumption shock. a Represents expected correlation between instrumental variables and exogenous trading needs to sell. b SEO PR is the raw buy-and-hold return of an SEO firm for 252 trading days ending on the day before issue date, or the number of days between the first listed date and the day before issue date, whichever is smaller. c VWPR is the PR of the CRSP value-weighted index. d The variance of SEO PR (and VWPR) is calculated by using the daily returns over the same holding period as the one used to calculate SEO PR. e If there exists no anti-takeover measure announcement during the one year period before and after SEO, the instrument is set to 1 and 0 otherwise. Only the anti-takeover measure announcements in Jarrell and Poulsen (1987: by 600 firms in the period ) are used. emergency needs. The larger the number of insider transactions, the less likely there is an emergency. Third, the exogenous need to trade will depend on any kinds of threats to the safety of insiders positions at the company. For instance, if there is any hostile takeover threat that insiders want to defend against, insiders will increase their holdings. Therefore, the firms that have taken any anti-takeover measure during the 1-year period before and after SEOs are treated as a group of companies with insiders exogenous need to buy their shares (i.e., they might purchase their shares not because of favorable insider information but because they want to defend against the hostile takeover attempt). As a first step to incorporate this idea, 600 companies listed in the appendix of Jarrell and Poulsen (1987) are used as the population of firms with any anti-takeover measure. 6 Notice that this variable is the only instrumental variable that is not publicly available at the time of issue. For an instrumental variable (Z i ) which has a positive (negative) prior relation with the exogenous need to sell, the value is transformed by setting Z* i 1 if the value of the instrumental variable is larger (smaller) than the average, and Z* i 0 otherwise. This is to use only those firms with insiders who are more likely to have great needs to sell their shares. Using these instrumental variables, the inequality tests are conducted. Detailed description of test procedures appears in Appendix. The results are reported in Table 5. The null hypothesis of a positive abnormal return of

11 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Table 5 Inequality test results Mean of AR a (in %) W-statistics p-value APR SAPR NIT ANTI Total NYSE/Amex Nasdaq This table reports the results of the inequality test described in Section 3. The hypothesis of the test is; H 0 : E(SEO BHR Average Matching Firms BHR) V Z 0; and H A : E(SEO BHR Average Matching Firms BHR) V Z R, where SEO BHR is the 3-year holding period return of SEO firms with prior pure insider purchases (PP). The matching firms are 5 firms that have the closest prior annual returns among the firms in the same size and BM quintiles. W-statistic is defined in the paper and is distributed as a weighted sum of chi-squared distribution. The weights are estimated by using the Monte Carlo simulation method with 10,000 draws. PP firms are the ones with only insider purchases and no insider sales for the previous 6-month period including issue dates. a The mean of abnormal return (SEO s BHR the average BHR of the matching firms) after multiplying instrument values for each observation. APR [SEO prior annual return CRSP value-weighted index s prior annual return], SAPR [SEO prior return/var(seo prior return) CRSP value-weighted index s prior return/var(crsp value-weighted index s return)], NIT is the number of total insider trading, and ANTI is anti takeover measure announcement. issuing firms with prior insider purchases is rejected at a 10% significance level in every case. Notice that there is no case in which the average transformed (by multiplying instrumental variables) abnormal return is positive. These results strongly suggest that insider purchases are not a credible signal of the issuing firms good future stock price performance even when we focus on those firms the insiders of which would have not purchased due to some exogenous needs to trade. We cannot pick better quality issuing firms just by looking at the number of prior insider purchases. 4. Summary and conclusion Previous empirical studies in the insider trading literature have not controlled for insiders exogenous trading needs because it is difficult to measure those exogenous needs. For example, Lee (1997) documents a poor relation between pre-issue insider trading and the post-issue long-run performance of primary seasoned equity issuing firms, and points out that the poor relation might be due to no control of exogenous trading needs in his tests. An inequality test with instrumental variables (ITIV) method is introduced as a way to control for exogenous consumption shocks in empirical studies on insider trading. This ITIV method has been used in statistical tests involving unobservable factors. For example, Boudoukh, Richardson, and Smith (1993) use the ITIV method in their test of the positivity of the ex ante risk premium to take into account the unobservability of expected returns. This ITIV method is used to examine whether or not there is any close relation between pre-issue insider trading and the post-issue long-run performance of primary seasoned equity issuing firms after controlling for exogenous consumption shocks. It turns out that we cannot

12 70 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) observe a close relation even after indirectly controlling for exogenous consumption shocks. This supports the Lee s (1997) contention that the poor relation is likely to be due to increased free cash flow problems after primary seasoned equity offerings rather than due to exogenous consumption shocks. In addition, the decrease in insider sales right after primary SEOs also supports this since primary SEO firms are substantially overvalued right after the primary SEOs. In summary, this paper emphasizes the importance of exogenous consumption shocks in insider trading, and stresses the needs to control for exogenous needs to trade in empirical studies on insider trading. It would be interesting to see whether or not the indirect control of exogenous trading needs change the empirical results documented in the insider trading literature. Notes 1. Throughout the paper, the value-weighted NYSE/Amex index return is used for the NYSE/Amex-listed firms and the value-weighted Nasdaq index return is used for the Nasdaq-listed firms. 2. Related to the differences between NYSE/Amex SEO firms and Nasdaq SEO firms, Loderer, Sheehan and Kadlec (1991) report that there is virtually no evidence of short-run underpricing for NYSE SEOs, mixed evidence for Amex SEOs, and strong evidence for Nasdaq SEOs. 3. If there is no insider trading, the ratio of purchase transaction is set equal to Both Karpoff and Lee (1991) and Kahle (1995) also show that the number of insider sales significantly increases before SEOs. 5. It is possible that the lock-up provision (usually 90 days or 180 days) has caused the results since the insiders of issuing firms are not allowed to sell their shares without the prior written consent of underwriters within this period. To see whether this is the main cause, several different 6-month periods (including the one starting 3 months after issuing) were examined and the results were similar to the one reported in Table These are the companies that adopted antitakeover amendments in the period Acknowledgments I would like to thank Narasimhan Jegadeesh, Tim Loughran, Neil Pearson, George Pennacchi, and Jay Ritter for many helpful comments. I am also grateful to seminar participants at the University of Waterloo, University of Illinois, Purdue University, University of Arizona, University of New Orleans, Case Western Reserve University and Michigan State University for useful discussions. Appendix A. Statistical test procedures using the inequality test with instrumental variables methodology First, the sample mean, ÂR Z is estimated. Each element of the vector ÂR Z is estimated as follows.

13 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) ÂR Z 1 T R i T j R m j z 1,...N j 1 where R j is the 3-year buy-and-hold return of buyer j, R m j is the average 3-year holding period return of the matching firms, Z ji is the ith instrumental variable for buyer j, and T is the total number of SEOs. The covariance matrix ( ˆ ) is estimated by using White s (1980) heteroskedasticity-consistent estimator. Then, the following quadratic programming problem is solved and ÂR R Z, a restricted (to be positive) estimator of AR Z, is estimated. Min ÂR AR Z Z AR Z ˆ 1 ÂR Z AR Z s.t. AR Z 0 Finally, the test statistic, W, is defined as W T ÂR R Z ÂR Z ˆ 1 ÂR R Z ÂR Z Wolak (1989) shows that the asymptotic distribution of W-statistic satisfies N sup Pr W c AR z 0 k 0 Pr k 2 c w N, N k, ˆ T, where c R is the critical value for a given size; Pr (W c) is the probability of W greater than a critical value c given AR Z and ˆ /T; Pr[ 2 k c] is the probability of chi-squared variable with k degrees of freedom being greater than c; and w(n, N k, ˆ /T) is the probability that ÂR R Z has exactly N k positive elements. A Monte Carlo simulation method is used to estimate the weight, w, and is described in the following [see also Wolak (1989) and Boudoukh, Richardson and Smith (1993)]. 1. Generate 10,000 random N 1 vectors (Xˆ s) from the multivariate normal distribution with zero mean and covariance matrix ( ˆ /T). 2. For each random vector, Xˆ, the following minimization problem is solved to get Xˆ R Min X Xˆ X ˆ /T 1 Xˆ X s.t. X 0 3. The weights are estimated as the fraction of 10,000 Xˆ Rs which have exactly N k elements exceeding 0. Note: IMSL (FORTRAN) subroutines are used to generate the random vectors and to solve the minimization problems. References Boudoukh, J., Richardson, M., & Smith, T. (1993). Is the ex ante risk premium always positive? A new approach to testing conditional asset pricing models. Journal of Financial Economics, 34,

14 72 I. Lee / The Quarterly Review of Economics and Finance 42 (2002) Elliot, J., Morse, D., & Richardson, G. (1984). The association between insider trading and information announcements. Rand Journal of Economics, 15, Finnerty, J. E. (1976). Insiders and market efficiency. Journal of Finance, 31, Givoly, D., & Palmon, D. (1985). Insider trading and the exploitation of inside information: some empirical evidence. Journal of Business, 58, Jaffe, J. F. (1974). Special information and insider trading. Journal of Business, 47, Jarrell, G. A., & Poulsen, A. B. (1987). Shark repellents and stock prices: the effects of antitakeover amendments since Journal of Financial Economics, 19, John, K., & Lang, L. H. P. (1991). Insider trading around dividend initiation: theory and evidence. Journal of Finance, 46, Kahle, K. M. (1995). Insider trading and new securities issues. Working paper, Ohio State University. Karpoff, J. M., & Lee, D. (1991). Insider trading before new issue announcements. Financial Management, 20, Lee, I. (1997). Do firms knowingly sell overvalued equity? Journal of Finance, 52, Lin, J., & Howe, J. (1990). Insider trading in the OTC market. Journal of Finance, 45, Loderer, C. F., Sheehan, D., & Kadlec, G. (1991). The pricing of equity offerings. Journal of Financial Economics, 29, Loughran, T., & Ritter, J. R. (1995). The new issues puzzle. Journal of Finance, 50, Rozeff M. S., & Zaman, M. A. (1988). Market efficiency and insider trading: new evidence. Journal of Business, 61, Seyhun, H. N. (1986). Insiders profits, costs of trading, and market efficiency. Journal of Financial Economics, 16, Spiess, D. K., & Affleck-Graves, J. (1995). The long-run performance following seasoned equity offerings. Journal of Financial Economics, 38, White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, Wolak, F. A. (1989). Testing inequality constraints in linear econometric models. Journal of Econometrics, 41,