Do Voluntary Corporate Restrictions on Insider Trading Eliminate
Informed Insider Trading? ☆
Inmoo Leea,*, Michael Lemmonb,1, Yan Lic,2, John M. Sequeirad,3
a
College of Business, Korea Advanced Institute of Science and Technology, 85 Hoegiro, Dongdaemun-gu,
Seoul, 130-722, Korea
b
Department of Finance, David Eccles School of Business, University of Utah, 1655 East Campus Center
Drive, Salt Lake City, UT 84112-9301, USA
c
World Bank Group Singapore Office, World Bank, 10 Marina Boulevard, 018983, Singapore
d
Monetary Authority of Singapore, 10 Shenton Way, MAS Building, 079117, Singapore
Abstract
We investigate whether voluntary corporate restrictions on insider trading effectively
prevent insiders from exploiting their private information. Our results show that insiders
of firms with seeming restrictions on insider trading continue to take advantage of positive
private information while being more cautious when exploiting negative private
information. The results suggest that insiders continue to exploit their informational
advantages in a way that minimizes their legal risk. We also find that the degree of
information asymmetry is significantly lower in firms with restriction policies and that
corporate governance significantly affects firms’ decisions to adopt these policies.
JEL classifications: G30, G34
Keywords: Corporate governance, Information asymmetry, Insider trading, Profitability
of insider trading, Voluntary corporate restrictions.
July 2014
Journal of Corporate Finance, forthcoming
☆
Earlier versions of this paper were circulated under the title “The effects of regulation on the volume,
timing, and profitability of insider trading.” Most of the work was completed while Yan Li was at Korea
University. The findings, interpretations, and conclusions expressed in this paper are entirely those of the
authors and they do not represent the views of the World Bank or the Monetary Authority of Singapore.
* Corresponding author: KAIST Business School, Korea Advanced Institute of Science and Technology, 85
Hoegiro, Dongdaemun-gu, Seoul, 130-722, Korea. Tel: +82 2 958 3441
E-mail addresses: [email protected] (I. Lee), [email protected] (M. Lemmon),
[email protected] (Y. Li), [email protected] (J.M. Sequeira).
1
Tel.: +1 801 581 7463
2
Tel: +65 81613202
3
Tel: +65 62299311
1
Introduction
Over the last twenty five years or so, there has been a series of changes in the
regulatory environment regarding insider trading, 4 which has made firms pay closer
attention to insider transactions. In particular, the Insider Trading and Securities Fraud
Enforcement Act (ITSFEA) passed by Congress in 1988, and the Stock Enforcement
Remedies and Penny Stock Reform Act (SERPSRA) passed in 1990 effectively rendered
firms and their top management responsible for their employees’ illegal trading.5
In response, firms have started to voluntarily adopt corporate policies that restrict
their insiders from freely trading their stock in order to reduce their own legal risk, with
many of these policies typically specifying certain time windows during which insiders are
allowed to trade their stock. Outside of the allowed periods, which are usually described
as blackout periods, insiders are basically restricted from trading without pre-clearance
from their firms. In most instances, firms choose short time windows after the quarterly
earnings announcement dates as allowed periods, with many requiring their insiders to
obtain pre-approvals even when they trade during the allowed time periods.6
4
Insiders are not allowed to trade based on material inside information according to Section 10(b) of the
Securities Exchange Act of 1934. Nevertheless, many studies have documented that insider trading contains
information regarding future stock returns (for example, Seyhun (1986, 1992), Lin and Howe (1990),
Meulbroek (1992), Rozeff and Zaman (1998), and Lakonishok and Lee (2001)). In early 2000, as a part of
its final rules surrounding Regulation Fair Disclosure (FD) that changed both the timing and content of
earnings information released by companies, the Securities and Exchange Commission reiterated the
importance of insiders not trading based on material non-public information (for example, Bailey, Li, Mao
and Zhong (2003)). In addition, the Sarbanes Oxley Act which took effect in August 2002, also increased
the scrutiny associated with insider trading by tightening the reporting requirements associated with insider
transactions.
5
Garfinkel (1997) finds that insider transactions around earnings announcements after the passage of
ITSFEA, become less “informed” with respect to the news in the announcement.
6
Both Jeng (1998) and Bettis, Coles, and Lemmon (2000) provide evidence that many companies adopt
internal policies restricting trading by insiders. Jagolinzer, Larcker and Taylor (2011) show that 80% of their
260 sample firms with restriction policies require all insiders’ trades to be pre-approved by the general
counsel even when those transactions are made during allowed time windows. The mean blackout period is
around 46 days prior to and one day after a quarterly earnings announcement.
2
In this paper, we examine how many firms have adopted voluntary restriction
polices over the years and whether insiders continue to earn significant abnormal profits
from their insider transactions even after the adoption of these restriction policies. While
previous studies such as Bettis, Coles, and Lemmon (2000) document that insiders of firms
with restriction policies earned significantly less abnormal profits, it remains unclear
whether insiders are still earning abnormal profits from their transactions after their firms’
adoption of restriction policies. This is a critical question that needs to be addressed since
it determines whether academics should continue to use insider trading as a source of
informed transactions in empirical studies, and whether professional investors would be
incentivized to revise their active investment strategies based on insider trading. Moreover,
the answer to the question would allow regulators and policy makers to evaluate the
effectiveness of regulations on insider trading and follow-up enforcements.
Unlike Bettis, Coles and Lemmon (2000) who employ a survey approach in
examining the profitability of insider trading for firms with self-imposed insider trading
restrictions, this study uses an approach that identifies firms with seeming restriction
policies based on insiders’ trading patterns, which is similar to the approach used in
Roulstone (2003), and allows us to use a more comprehensive dataset. While the survey
approach allows researchers to identify firms with restrictions without error, it tends to
limit the number of sample firms. In comparison, the approach used in this paper is subject
to a misclassification problem. To reduce the possibility of reaching a false conclusion
arising from misclassification, we check the robustness of the results using an alternative
approach that identifies firms with restriction policies by searching for company web sites
as used in Dai et al. (2013).
3
Betties, Coles and Lemmon (2000)’s main reason for examining abnormal profits
earned by insiders was to check whether market makers would face lower adverse selection
costs during blackout periods. Therefore, they only examined short-term (one-week)
abnormal profits earned from insider transactions. In this paper, however, we focus on
longer investment horizons, namely, three- and six-month horizons. This is because our
main question addresses whether insiders continue to earn abnormal profits from their
insider transactions even after the adoption of voluntary restriction policies. Additionally,
the short-swing rule, under which only those profits earned from round-trip insider
transactions over a period longer than 6 months are legitimate, leads us to focus on longer
horizons.7
In addition, we closely examine whether insiders react to these restrictions in
strategic ways in order to continue exploiting their private information through a selective
assessment of it. Earlier work by Cheng and Lo (2006), for example, finds that managers
strategically choose disclosure policies and time their trading only for purchase
transactions but not for sale transactions, possibly to avoid potential legal troubles from
taking advantage of negative private information.8 If the litigation risk is higher following
insider sales than insider purchases, insiders would be more cautious in exploiting negative
7
Section 16(b) of the Securities and Exchange Act of 1934 requires short-term profits earned by insiders of
a firm from round-trip transactions made within a 6-month period to be returned to the company.
8
They point out that insider sales attract more legal troubles than insider purchases since investors who suffer
from actual losses from a significant price decrease following insider sales are more likely to allege a
violation of the “disclose or abstain” rule than investors who suffer from opportunity losses from a significant
price increase following insider purchases are. The “disclose or abstain” rule is based on Cady, Roberts and
Co. (40 SEC 907 [1961]) and requires that anyone in possession of material non-public information should
either disclose it to the public before trading or abstain from trading. As explained earlier, Bettis, Coles and
Lemmon (2000) find that insiders of firms with blackout periods earn significantly lower abnormal returns
from their sale transactions than those of other firms, but the result does not hold for their purchase
transactions. This is also indicative of the possibility that insiders respond to corporate restrictions on insider
trading in an asymmetric way to continue to exploit their information advantage without violating restriction
rules and increasing their legal risk.
4
private information while continuing to exploit positive private information.9
Finally, we examine how restriction policies are related to the degree of information
asymmetry and corporate governance. Since restriction policies significantly limit timely
informed trading, it is possible that such policies would negatively affect the stock price
informativeness and increase the degree of information asymmetry. Conversely, these
policies may encourage other informed investors, such as hedge funds, to trade more, and
as a consequence, increase price informativeness. We also investigate whether corporate
governance significantly affects a firm’s decision to adopt restriction policies.10
Our analysis is based on a comprehensive database of U.S. insider transactions in
12,332 firms between 1986 and 2010. We find that more than 70% of the sample firms
have likely adopted voluntary trading restrictions by the end of our sample period. We
also find that firms with restrictions have faced less information asymmetry, as measured
by the analysts’ earnings forecast dispersion, idiosyncratic volatility and the probability of
information-based trading (PIN), suggesting that restriction polices do not seem to impose
negative externalities in terms of price informativeness.11 Moreover, we find that restricted
9
During June 1974 and June 2012, there were 318 articles in The New York Times (NYT) and 31 abstracts
in the Wall Street Journal Abstract (WSJ), which include the term, “illegal insider trading” and are available
in LexisNexis. A close examination of the articles and abstracts reveals 18 lawsuits and 54 prosecutions in
the NYT articles, and seven lawsuits and one prosecution in the WSJ abstracts. Out of the cases reported in
NYT, we were able to identify that 11 (five lawsuits and six prosecutions) cases were related to insider sales
while 20 (four lawsuits and 16 prosecutions) were related to insider purchases. For other cases, we could not
clearly identify the types of illegal insider transactions. We also noticed that most illegal purchases were
related to mergers and acquisitions (M&A). Excluding M&A related cases, there were 14 sales related cases
and only two purchases related cases, suggesting that for non-M&A related cases, the majority of such cases
are related to illegal insider sales. This supports the possibility that insiders perceive a higher litigation risk
from informed insider sales than informed insider purchases.
10
In a sense, well governed firms with less anti-takeover provisions are more likely to adopt restriction
policies as their managers are less likely to be entrenched, and may therefore be more willing to accept insider
trading restrictions. However, it is also plausible that managers of poorly governed firms with more antitakeover provisions could be less concerned about short-term market pressures and would tend to focus more
on long-term value creation instead, in which case, they are therefore more likely to adopt restriction policies.
11
Related to this, Bettis, Coles and Lemmon (2000) show that the blackout period is associated with a modest
reduction in the adverse selection component of bid-ask spreads but market makers widen the spread during
5
firms have, on average, more anti-takeover provisions as measured by the index used in
Gompers, Ishii and Metrick (2003).
Regarding our main question, we find that insiders continue to earn significant
abnormal returns from their aggregate purchase and sale transactions even after adopting
restriction policies. In addition, for overall sample, size and book-to-market equity ratio
(B/M) adjusted abnormal returns of restricted firms over the 6-month period following their
insiders’ transactions are not significantly different from those for unrestricted firms.
However, when we control for other factors known to affect the profitability of insider
trading, we find that abnormal returns from aggregate transactions are significantly less for
those transactions made by restricted firms’ insiders, indicating the possibility of restriction
policies significantly affecting the profitability of insider trading.
When we separately examine abnormal profits earned from purchases and those
from sales, we find interesting results. In general, insiders of restricted firms earn
significantly positive abnormal returns from their purchases, unlike from their sale
transactions. More importantly, for purchase transactions, we do not find any evidence that
insiders of restricted firms earn significantly less abnormal profits from their purchase
transactions than insiders of unrestricted firms do. For sale transactions, however, we find
some evidence that restricted firms’ insiders earn significantly less than unrestricted firms’
insiders do. The results are consistent with the possibility that insiders strategically respond
to restriction policies in a way that minimizes legal risk by continuing to exploit positive
private information while being more cautious in exploiting negative private information.
To check the robustness of our results, we use a portfolio approach, measuring the
allowed trading days. Overall, they conclude that restriction policies do not seem to significantly reduce the
liquidity of restricted firms’ shares.
6
abnormal returns based on the 4-factor model in Carhart (1997). We confirm that, in
general, the previous results hold even in this alternative setting. In addition, we find
qualitatively similar results when we use an alternative way of identifying restricted firms
based on the searches of company web sites.
The rest of the paper is as follows. In section 2, we present the data and
methodology together with the characteristics of firms with voluntary restrictions on
insider trading. In Section 3, we analyze the profitability of insider trading in a variety of
ways. Finally, we conclude the paper in Section 4.
2.
Voluntary Corporate Restrictions on Insider Trading
2.1.
Data
The insider trading data used in our study covers the period from January 1, 1986
to December 31, 2010, and is based on the comprehensive insider trading data cleaned and
distributed by Thomson Financial.
It covers all insider transactions reported to the
Securities and Exchange Commission (SEC) by insiders who are required to do so
according to Section 16(a) of the Securities and Exchange Act of 1934. Our sample
includes companies in the insider trading data that are available on both the CRSP and
Compustat databases. Insiders are classified into management, large shareholders, and
others who are required to report all trades to the SEC. “Management” refers to CEOs,
CFOs, chairmen of the board, directors, officers, presidents, and vice presidents. “Large
shareholders” are shareholders who own more than 10% of shares in a firm, but are not
management.
The insider transactions that we examine relate to open market or private purchases
7
and sales of common stock, as well as the acquisition of stock through the
exercise/conversion of options, warrants or convertible bonds and option-related sales.12
After applying various filters, we arrived at a sample of 4,981,383 insider transactions
representing 167,023 firm-years.
To control for the factors that are known to affect insider trading behavior, we
follow Lakonishok and Lee (2001) and classify our sample firms into three size and bookto-market (B/M) groups. We use the cutoff points provided by Kenneth R. French in his
web site to form size and B/M deciles.13 The size cutoff points are based on the June market
capitalization of NYSE listed firms, while B/M cutoff points are based on the market
capitalization at the end of December of the previous calendar year and book equity values
in the prior fiscal year end of NYSE listed firms with positive book equity values. The
bottom three size deciles are classified as small firms, the top three deciles as large firms,
and the remaining four deciles as medium firms. Similarly, we independently form three
B/M groups: low B/M firms, medium B/M firms and high B/M firms, based on the bottom
30%, middle 40% and top 30% of B/M cutoff points, respectively. After merging insider
trading data with CRSP and Compustat data, the sample is reduced to 3,766,313 insider
transactions representing 100,976 firm-year observations.
Finally, we collect earnings announcement data from the Compustat Industrial
Option-related sales are those open market or private sale transactions that occur within six months after
the exercises of options, with the number of shares sold being less than or equal to the number of shares
acquired through prior exercise of these options. For consistency, we also adopted the Thomson Financial
classification for all option-related sales that occur after 1993. As explained in Carpenter and Remmers
(2001), before May 1991, insiders were required to hold shares acquired through exercises of options for at
least six months to avoid the short-swing rule (Section 16(b) of the Securities and Exchange Act of 1934)
under which insiders are required to return any profits made from a round-trip transaction of less than six
months. Since May 1991, insiders have been free to immediately sell the shares acquired through option
exercises.
13
http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html
12
8
Quarterly file. For each insider transaction, we identify the two closest consecutive
quarterly earnings announcement dates that contain the particular transaction date. Due to
missing earnings announcement data, the final sample is further reduced to 3,371,399
insider transactions representing 88,266 firm-years.
2.2.
Summary Statistics of Insider Trades
Table 1 shows the summary statistics for each firm-size group and insider type. The
table reports the fraction of firms in each group with at least one insider trade (Fraction)
per year, the average number of trades per company per year (# of trades), and the average
ratio of the individual company’s total insider trading dollar volume during a year to the
market capitalization of the corresponding company at the beginning of the year (% Mkt
Cap). In addition, we report the average annual aggregate dollar trading volume (Total $)
for each insider type and each insider transaction type. All dollar values are reported in
constant 2005 dollars.
From the results, we see that managers (i.e., “management”) are the most active
insiders in terms of trading frequency, and are generally more active in larger firms. In all
firm sizes across all insider types, insider sales exceed purchases both in terms of number
of transactions and dollar volume. Based on 2005 dollar value, management in aggregate
bought an average of $1.4 billion worth of their stocks through open market or private
transactions and about $5.8 billion through the exercise/conversion of options, warrants, or
convertible bonds in each year during our sample period. They directly sold $23.8 billion
of their stocks and engaged in $8.8 billion of option-related sales. Overall, trading by
management accounts for approximately 31% (78%) of the total dollar volume of insider
9
purchases (option conversions) and 57% (85%) of the total dollar volume of insider sales
(option related sales). Large shareholders comprise the next most active group, while
trading by other insiders represents only a small fraction of insider trading activity.
For the rest of the paper, we only use management’s trading in our analysis because
voluntary corporate policies on insider trading are likely to affect the informativeness of
management only. Separately, our analyses using all insiders reported in Table 1 produce
results that are qualitatively similar. In addition, we focus on open market or private
purchase and sale transactions in the following analyses as in most previous studies.
2.3.
Insider Trading Patterns and Voluntary Corporate Restrictions on Insider Trading
In this section, we explore how changes in the regulatory environment and firms’
responses to these changes have affected the timing of insider trading. We focus on the
timing of insider trading activity relative to earnings announcements. As pointed out
earlier, Bettis, Coles, and Lemmon (2000) show that the majority of the firms that have
adopted explicit blackout periods define blackout periods in relation to earnings
announcements.14 They find that the most typical restriction policy specify an allowed
trading period beginning on the third trading day following an earnings announcement and
ending on the twelfth trading day after the announcement.
To assess the effects of regulatory changes on insider trading activity, we divide
our sample period into four corresponding sub-periods as follows:15 sub-period 1 (1986 to
14
Bettis, Coles and Lemmon (2000) state that in rare cases, blackout periods are defined relative to dividend
announcements, mergers, bankruptcy filings, board meetings, the end of the quarter, or other important
corporate events. Jeng (1998) reports that the median number of annual trading days allowed for her sample
firms with trading restrictions is 88 days (i.e., 22 days per quarter).
15
In 1984, U.S. Congress passed the Insider Trading Sanctions Act (ITSA), which gave the SEC the
discretion to seek civil penalties up to three times the profit made or loss avoided, as well as increased
10
1988) which mostly covers the period before the ITSFEA; sub-period 2 (1989 to 1990)
which mostly covers the period after the ITSFEA but before the SERPSRA of 1990; subperiod 3 (1991 to 2002) which mostly covers the period after the SERPSRA and before the
change in reporting time in Section 16(a) of the Securities Exchange Act of 1934; and subperiod 4 (2003 to 2010) which mostly covers the period after the change in the reporting
time regulation associated with the Sarbanes Oxley Act.
Next, for each firm, we divide the period between two consecutive earnings
announcements into three equal intervals, with each interval being about one-month in
length (approximately 21 trading days). To measure overall trading activity, we compute
the total number of insider transactions that occur in each interval. For the timing of
trading, we compute the fraction of insider transactions for the first interval, denoted PEC1,
calculated as the number of insider transactions recorded in the first interval divided by the
total number of transactions during the period between two consecutive earnings
announcement dates. Since PEC1 would be meaningless for a quarter with a small number
of insider transactions, we therefore compute PEC1 only for the quarters with at least three
insider transactions. Figure 1 provides details regarding the computation of PEC1.
Figure 2 shows the values of PEC1 to PEC3 over the period 1986-2010. The lower
criminal penalties from $10,000 to $100,000. Four years later, the Insider Trading and Securities Fraud
Enforcement Act (ITSFEA) was passed, which held companies and top management liable for illegal trading
of employees and increased the maximum criminal penalties tenfold to $1 million and doubled maximum
prison sentences to 10 years for improper insider trading. The Securities Enforcement Remedies and Penny
Stock Reform Act of 1990 (SERPSRA) empowered the SEC to directly impose civil penalties to securities
law violators, including illegal insider trading, and increased penalties and fines for delinquent filings of
insider transactions. The most recent amendment to Section 16(a) of the Securities and Exchange Act of 1934
which was introduced in August 2002 forced insiders to report their trading within the two business days
after trading. Before this regulation change, insiders were required to report to the SEC by the 10 th day after
the end of the trading month.
11
right graph in Figure 2, which displays the results for all sample firms, shows that PEC1
increases gradually over the sample period, while PEC3 exhibits a corresponding decrease
over the same period. In fact, differences between PEC1 and PEC3 widen significantly
until 2010. At the beginning of our data period in 1986, PEC1 and PEC3 are 34% and 24%,
respectively. However, towards 2010, PEC1 increases to 57% while PEC3, on the other
hand, falls to 13%.
Unreported t-statistics for the test of differences in PEC1 and PEC3 for each
calendar year show that the differences are significant at the 1% significance level, and that
the significance has increased over time. This result is consistent with an increasing number
of firms over the sample period imposing explicit corporate restrictions on insider trading,
which typically force insiders to trade in the first interval. Insiders seem to be dissuaded
from trading during the period immediately prior to an earnings announcement and this is
captured by the fall in PEC3 over the sample period. The other graphs in the figure show
the results for different firm size groups. In earlier periods, the values of PEC1 are
generally higher in larger firms as compared to smaller firms, while smaller firms have
higher values of PEC3. However, the differences across different size groups become
smaller in later periods.
To identify firms that are likely to have adopted voluntary corporate policies to
restrict insider trading, we follow Roulstone (2003) who examines the relation between
executive compensation and insider trading restrictions, and classify a firm as a restricted
firm beginning in the quarter when PEC1 is greater than or equal to 75%.16 We choose the
16
Jagolinzer, Larcker and Taylor (2011) show that about 24% of transactions done by insiders of their 260
sample firms with corporate policies on insider trading occurred during the restricted period. They attribute
this partly to Rule10b5-1, enacted in October 2000, which provides a defense against legal actions to insiders
who pre-plan their non-information based trades (Jagolinzer (2009)).
12
75% cutoff point based on the findings in Bettis, Coles and Lemmon (2000), who find that
insiders are three times more likely to trade in allowed trading windows than during
blackout periods, which would imply that PEC1 is likely to be higher in restricted firms.
Since a firm is not likely to drop its restriction policy after adoption, we also require that
PEC1 should be greater than or equal to 50% in subsequent quarters, where PEC1 is
calculated based on all management transactions in subsequent quarters.17 To check the
validity of this method, we apply it to the survey sample in Bettis, Coles and Lemmon
(2000), and find that in 71% of all cases in their sample, our method correctly classifies
firms into their corresponding groups. In addition, as explained in Section 3.3, we use an
alternative method of identifying restricted firms to ensure that our main results are not due
to misclassification.
Figure 3 shows the percentage of firms classified as restricted firms in each year
for each of the three size groups, as well as the average probabilities of information-based
trading (PIN) for each size group.18 The results indicate an upward trend in the percentage
of restricted firms for all size groups, with the percentage of large firms classified as
restricted firms being mostly higher. More generally, we find that the percentage of all
firms classified as restricted firms has increased dramatically from less than 7% in 1986 to
more than 80% for medium and large firms in 2010. For small firms, the percentage of
17
Once a firm is classified as a firm with restrictions from a certain quarter, it remains classified as such until
the end of the sample period. It is certainly possible that some insiders may voluntarily trade right after
earnings announcements even without restrictions imposed by their companies. As such, since the majority
of insider transactions occur right after earnings announcements for all firms that are classified as restricted
firms according to our method, at the minimum, the results presented in this paper may be interpreted as the
results for those firms with insider trading restrictions voluntarily imposed by either corporations or
individuals.
18
The annual PIN estimates for all NYSE/Amex common stocks during 1983 and 2001 were made available
by Soeren Hvidkjaer in “https://sites.google.com/site/hvidkjaer/data”, and Easley, Hvidkjaer and O’Hara
(2002) describe the procedures used to calculate the PIN measure.
13
restricted firms is closer to 75% in 2010. The results indicate that larger firms are more
likely to be classified as restricted.19
On the other hand, the average PINs are lower for large firms than for smaller firms.
Also, decreases in PIN tend to occur for large and medium firms as the ratios of firms with
restriction policies increase. However, there is no clear downward trend for the average
PIN of small firms, possibly due to higher PIN values of small technology stocks during
the internet bubble period in the late 1990s and early 2000s.
To investigate whether insider trading activities have changed following the
adoption of voluntary restriction policies, we report in Table 2, the averages of four
measures that capture the level of insider trading activities before and after firms are
classified as restricted firms; namely, the number of trades, dollar value of trades as a
percentage of market capitalization, total dollar value of trades, and PEC1. Except in subperiod 4, the number of trades for each type of insider transaction significantly increases
after the firm is classified as restricted. In sub-period 4, only purchase exhibits this pattern.
On average, there are 2.85 (14.52) purchases (sales) per year before being classified as
restricted firms and 4.45 (16.80) purchases (sales) thereafter. The related t-statistic of 14.06 (-1.64) suggests that the difference is highly (marginally) significant. Overall, with
the exception of sales in the last sub-period, there is no evidence to suggest that insiders
significantly decrease their trading activity following the adoption of restriction policies.
2.4 Characteristics of Firms with Voluntary Corporate Restrictions on Insider Trading
To understand the characteristics of firms with voluntary trading restrictions, we
19
Large firms may be more likely to adopt a policy restricting insider trading due to greater scrutiny from
outsiders and the larger potential costs arising from insiders’ illegal trading.
14
ran regressions where the dependent variable is a dummy variable to indicate firms with
trading restrictions in a quarter. An understanding of the characteristics of restricted firms,
compared to unrestricted firms, would shed new light on the debates on whether insider
trading should be regulated.20
Since voluntary corporate restrictions on insider trading limit the timing of insider
trading, it is likely that less information would be reflected on a timely basis into stock
prices and, therefore, the adoption of such policies would affect the degree of information
asymmetry of the firm. To examine this, we employed various measures known to capture
the degree of information asymmetry, which include analyst earnings forecast dispersion
measured by the standard errors of earnings forecasts that were made for the most recent
fiscal year end and are available from the IBES (StErrForecasts), idiosyncratic volatility
(IdioVol) measured from the daily 4-factor model (Carhart (1997)) using daily returns data
during a quarter of transactions, and the probability of information-based trading (PIN).
Another issue that we examine is how corporate governance structures affect
restriction policy decisions. The measure of corporate governance that we adopt is the
index developed by Gompers, Ishii, and Metrick (2003) (GIM). The GIM index is higher
for poorly governed firms with many antitakeover provisions and is available for about
1,500 large firms during 1990 and 2006 through the Wharton Research Data Services
(Corporate Governance Legacy by RiskMetrics).
20
Notably, the most prominent criticism against insider trading regulations relates to the view that insider
trading enhances the market efficiency (see Manne (1966) and Carlton and Fischel (1983)). Fishman and
Hagerty (1992), however, find that informed insider trading may discourage outside investors from collecting
information by limiting potential gains from such activities. Nevertheless, Fernandes and Ferreira (2009)
show that the enforcement of insider trading regulation helps improve the market efficiency in developed
markets, although the result does not hold for emerging markets, lending support to the argument that in
countries with good legal infrastructures, the enforcement of insider trading encourages outsiders to invest
more resources to collect information and therefore increasing stock price informativeness.
15
As additional control variables, we include the natural log of firm size (LogAsset),
the log of the firm’s book-to-market equity ratio (LogBM), the standard deviation of stock
returns (StdDev), and indicator variables for banks (SICs 6000-6999, BankD) and utilities
(SICs 4900-4939, UtilityD).21
We also create three indicator variables to assess whether the adoption of restriction
policy has changed over time as follows. Yeardum_1 is equal to one for the period 19892010, and is zero otherwise. Similarly, we create Yeardum_2 for the period 1991-2010 and
Yeardum_3 for the period 2003-2010. As the period 1986-1988 represents the base period,
the estimates on each indicator variable therefore captures the incremental effect of each
additional period. To account for possible firm fixed effects in the regression, we include
dummies for each sample firm in most of the regressions, including those in Table 3. In
addition, we use clustered standard errors at the firm level as suggested by Petersen (2009)
to deal with the possibility of firm fixed effects not being permanent.
Table 3 presents the regression results for eight models, where the dependent
variable is a dummy variable to indicate those firms with voluntary restriction policies
during a quarter and each column represents the results from the various models.22 Models
(1) through (4) report the regression results excluding the GIM index. In particular, the
21
As discussed earlier, larger firms are more likely to adopt restriction policies because they are exposed to
larger litigation risks given that more analysts and investors follow them closely. For B/M, it is unclear how
a valuation ratio is related to the likelihood of adopting restriction policies. Firms with higher valuations are
likely to attract more attention from investors and analysts, which will in turn increase their litigation risk
and consequently the likelihood of adopting restriction policies. That being said, firms with lower valuations
may also put in place preventive measures to guard against legal troubles, since they are more susceptible
and are accordingly more cautious. We also include return volatilities and dummy variables for regulated
industries since opportunities to exploit inside information are likely to be related to return volatilities, while
insiders of regulated firms are more likely to be cautious in light of greater regulatory scrutiny inherent in
such firms.
22
The maximum number of firm-quarters in Table 3 (461,023) is greater than four times the number of firmyears with available insider trading data (88,266) because, the number of firm-quarters used in this table
includes those firm-quarters without any insider transactions.
16
coefficients of LogAsset and StdDev are positive and significant at the 1% significance
level in Models (1) through (4), suggesting that larger firms and firms with greater stock
return volatilities are more likely to have adopted restriction policies. In addition, the
significantly positive coefficients of time dummies indicate that more firms seem to have
adopted restriction policies in recent years. Meanwhile, the coefficients of LogBM are
significantly positive in Models (1) and (2) but become insignificant as we add information
asymmetry measures based on analyst information. Similarly, the coefficients of BankD
are negative and significant in Models (1), (2) and (4), whereas the coefficients of UtilityD
are not significant in all columns.
Regarding the stock price informativeness, the results indicate that firms with
restriction policies tend to have significantly less information asymmetry measured in
various ways. In most cases, the coefficients of various information asymmetry measures
are significantly negative at the 1% significance level.23
Models (5) through (8) add the GIM index to check whether corporate governance
structures are significantly different between firms with and without restriction policies.
Due to the data availability, the numbers of firm-quarters significantly drop when we
include the GIM index. The coefficients of the GIM index are significantly positive in
Models (5) through (8), indicating that firms with restriction policies tend to have more
anti-takeover provisions.
Managers of these firms might be more willing to adopt
restriction policies because they have protections against short-term market pressures and,
therefore, tend to pursue long-term shareholder interests instead of short-term profit
making opportunities as shown in Fu and Liu (2008).
23
The only noticeable exception is the insignificantly negative coefficient of StErrForecasts in Model (4)
where the sample is restricted to those with available PIN estimates.
17
In unreported analyses, we add four additional explanatory variables in the
regression analysis to examine whether executives’ characteristics also affect the decision
to adopt restriction policies using a subset of sample firms with available data in the S&P’s
ExecuComp database that covers the S&P 1,500 companies.24 When the GIM index is not
included in the analyses, the significance of IdioVol holds, whereas the coefficients of
StErrForecasts become insignificant. Among managerial characteristics, we find that the
coefficients of average percentage ownership held by executives and average ages of
executives are significantly negative, indicating that firms with restriction policies are
those with less ownership held by executives and with younger executives. This might be
because older executives who own a large amount of shares prefer not to be restricted in
disposing their shares and are, therefore, less likely to agree with voluntary restrictions.
Neither the coefficient of the overconfidence measure nor the coefficient of the percentage
of male executives is significant.
When the GIM index is added together with variables capturing managerial
characteristics, the coefficient for average ages remains significantly negative but that of
average percentage of shared held by executives becomes insignificant. In addition, the
24
In a recent paper, Malmendier, Tate and Yan (2011) show that managerial traits, such as overconfidence,
significantly influence corporate financial policies. It is therefore possible that managerial characteristics
may also affect the likelihood of adopting restriction polices regarding insider trading. To test this, we
include four variables that capture managerial characteristics: average percentage ownership held by
executives (Avg%ShraresOwned), average ages (AvgAge), the percentage of male executives (%Male) and
an overconfidence measure (OverConfidence) of executives that are covered by the ExecuComp database.
The overconfidence measure that we use is based on Campbell et al. (2011) and is defined as the realizable
value per unexercised exercisable option over the average exercise price of unexercised exercisable options
held by executives. More specifically, it is defined as [Opt_Unex_Exer_Est_Val / Opt_Unex_Exer_Num] /
[PRCCF - (Opt_Unex_Exer_Est_Val / Opt_Unex_Exer_Num)] where Opt_Unex_Exer_Est_Val is the
estimated value of unexercised exercisable options, Opt_Unex_Exer_Num is the number of unexercised
exercisable options, and PRCCF is the close price at the fiscal year end. Essentially, this is a moneyness
measure of in-the-money options that are not yet exercised and held by executives. The higher the value of
this measure, the more optimistic executives of the firm are because they are likely to hold deep in-the-money
options only when they expect even greater profits from waiting.
18
coefficient estimates of GIM and IdioVol all become insignificant possibly due to smaller
sample sizes.
3.
Profitability of Insider Trading
The results in the prior section suggest that insiders’ trading patterns have changed
significantly over time even though corporate voluntary restriction policies do not seem to
discourage them from trading. In this section, we examine whether insiders still earn
abnormal profits from their transactions even after a firm adopts voluntary restriction
policies on insider trading. One possibility is that restricting transactions to the period
immediately following earnings announcements effectively limits the informational
advantage of insiders without limiting their ability to trade. Another possibility, however,
is that the regulations have effectively created a de-facto “safe harbor” which allows
insiders to trade on their superior information with little fear of legal sanctions, especially
when their private information is not related to imminent earnings announcements, or when
they possess positive private information.
3.1. Profitability of Insider Trading for Restricted vs. Unrestricted Firms
To estimate abnormal profits earned by insiders from their insider transactions, we
first classify each of the three intervals between earnings announcement dates as either a
net purchase or a net sales event. In each interval with at least one insider purchase or sale
transaction, we count the numbers of purchase transactions and sale transactions and
classify the interval as a net purchase (sale) event if the number of purchases is greater
(less) than the number of sales. We use the nearest trading date following the endpoint of
19
each interval classified as a net purchase or sale interval as an event date in our analysis,
assuming that transactions occur on that date.25
We compute abnormal returns using 5 by 5 size and B/M benchmark portfolio
returns from Kenneth R. French’s website as a way to measure the profitability of insider
trading. First, we calculate 3-month and 6-month buy-and-hold returns of both sample
firms and their corresponding size and B/M portfolios starting from the pseudo transaction
date. The differences in buy-and-hold returns of the sample firms and their size and B/M
benchmark portfolios are our measures of buy-and-hold abnormal returns (BHARs). To
represent abnormal returns earned by insiders in a consistent way, we multiply the
abnormal returns by -1 for net sales so that higher abnormal returns have the same meaning
for both purchases and sales. Rather than paying attention to short-term abnormal returns
as in Bettis, Coles and Lemmon (2000) who examine whether market makers face less
adverse selection costs during blackout periods, we focus on 3-month and 6-month
abnormal returns. As explained earlier, this is because we intend to investigate whether
insiders earn abnormal profits from their insider trading even after adopting restriction
policies and the short-swing rule requires insiders to return short-term profits made in
round-trip transactions within a 6-month period.
In Table 4, to examine whether corporate restrictions on insider trading have a
significant impact on the profitability of insider trading, we compare the profitability of
insider trades in restricted firms with the profitability in unrestricted firms. We report the
results separately for different size groups, since smaller firms are less likely to adopt
25
Similar results are obtained if we define the event date as the middle point of the interval. Since we do
not use the exact transaction date as the starting date for abnormal return calculations, we will underestimate
abnormal returns if prices reflect insider trading information right after transaction dates but before the end
of the interval, which was used as the pseudo transaction date.
20
restriction policies (as shown in the previous section) and are more likely to have a larger
degree of information asymmetry.
The first panel in Table 4 reports the results for abnormal returns aggregated across
both purchases and sales. Insiders of both restricted and unrestricted firms earn
significantly positive abnormal profits over the 3-month period for all size groups. The
average abnormal return is 1.47% (1.34%) for small unrestricted (restricted) firms while
the average is a much smaller 0.42% (0.32%), albeit significant, for large unrestricted
(restricted) firms. We do not find any significant difference in abnormal returns between
unrestricted and restricted firms. Over the 6-month period, insiders of small and medium
firms continue to earn significantly positive abnormal returns whereas insiders of large
firms earn insignificant abnormal profits. Generally, the average abnormal returns are
smaller for firms with restrictions than those without restrictions. However, the return
differences are not statistically significant, except for small firms at the 6-month horizon
where insiders of unrestricted firms earn significantly higher abnormal returns (2.4%)
compared to those of restricted firms (1.81%).
Abnormal returns over the second 3-month period following the first 3-month
period after transactions (Diff3m) are significantly positive only for small firms, indicating
that insiders of small firms seem to base their trading decisions on both near and longerterm prospects. We do not find such evidence for insiders of medium and large firms.
Moreover, in small firms the Diff3m for the restricted firms is marginally significantly
lower than unrestricted firms. Surprisingly, insiders of large firms with restriction policies
earn marginally significantly negative abnormal returns during the second 3-month period
following their transactions.
21
Next, we separately examine abnormal returns of purchase and sale transactions in
Panels B and C. For purchase transactions, we find significantly positive abnormal returns
across all categories. When abnormal returns are compared between restricted and
unrestricted firms, we generally find lower abnormal returns for unrestricted firms,
although the differences are statistically significant only in medium firms for the 6-month
and the second 3-month periods. In any event, we do not find any indication that insiders
of firms with restriction policies earn significantly less abnormal profits from their
purchase transactions as compared to insiders of unrestricted firms.
On the other hand, we do find insignificant abnormal returns following insider
sales. Contrary to expectations, we actually find negative abnormal returns, especially for
small restricted firms over the 6-month period, which suggests that stock returns of firms
with insider sales are, on average, significantly higher than stock returns of firms with
similar size and B/M firms. 26 Abnormal returns over the second 3-month period are
significantly negative across all size groups, except for large firms without restriction
policies. The results in Table 4 suggest that insiders of firms with restriction policies
continue to earn abnormal returns from their purchase transactions over the 6-month period
following transactions, while they do not seem to earn abnormal profits following their sale
transactions.
The asymmetry in the profitability of purchases and sales for firms with trading
restrictions suggests that the adoption of trading restrictions may create an opportunity for
insiders to trade on positive private information, while suppressing trading based on
negative private information in order to avoid legal scrutiny. Cheng and Lo (2006) also
26
As explained earlier, abnormal returns following insider sales are multiplied by -1, so that larger abnormal
returns are interpreted as more profitable insider transactions.
22
present evidence that managers time their disclosures of negative news prior to purchases,
but do not exhibit similar behavior prior to sales, possibly due to more legal threats
involved with insider sales.
Table 5 reports the regression results of the differences in insider profits across
restricted and unrestricted firms. The dependent variable in the regressions is the buy-andhold abnormal return relative to the relevant size and B/M benchmark. The independent
variables include an indicator variable, RES, for restricted firms and another indicator,
ResPrd, to indicate transactions made during the blackout period. We assume the blackout
period to be the remaining two periods of three equally divided intervals between two
earnings announcement dates after excluding the first interval that follows an earnings
announcement. In Table 5, among the variables used to capture information asymmetry,
only the idiosyncratic volatility variable is used as a variable to capture information
asymmetry in the main regression analyses. Other variables based on financial analyst
information are used as instrumental variables only in Panels C and D, along with the
restriction indicator variable which is endogenized in a two-stage regression approach. In
addition, we use LogAsset, LogBM, StdDev, BankD and UtilityD as in Table 3. Finally, to
check whether corporate governance affects the profitability of insider trading, we repeat
the analyses using a subset of the sample after adding the GIM index as an additional
explanatory variable in Panels B and D.
Firstly, Panel A shows the regression results after controlling for firm fixed effects
via firm dummy variables. As in Table 3, clustered standard errors are also used in
significance tests. For aggregate purchases and sales, the results indicate that insiders in
restricted firms earn significantly lower abnormal returns compared to those in unrestricted
23
firms over the 6-month period, although we do not find a similar result over the 3-month
period. The coefficient of RES for 6-month abnormal returns is a significant -0.0105,
suggesting that holding all else constant, insiders of restricted firms earn 1.1% lower
abnormal returns than insiders of unrestricted firms do over the 6-month interval following
their purchase or sale transactions. The coefficients of ResPrd and its interaction term with
RES (RES×ResPrd) are not significant in all regressions, indicating that the profitability of
insider trading does not depend on whether an insider trades during a blackout period or
not, after controlling for other factors. However, the aggregate results should be interpreted
with caution, as R2 is less than 1%. When purchases and sales are examined separately,
we find that the results are consistent with the univariate findings in Table 4 that insiders
in restricted firms make significantly more profitable purchase transactions during the 6month period, but make less profitable sale transactions, all else equal. As stated earlier,
abnormal returns for sales are multiplied by -1, so that larger abnormal returns indicate
more profitable insider transactions.
In Panel B, where the GIM index is included, we find that the coefficients of GIM
are significantly positive for purchases and are significantly negative for sales. Managers
who are protected against short-term market pressures through anti-takeover provisions
seem to be less interested in exploiting short-term negative private information but more
willing to exploit positive inside information. On the other hand, the coefficients of RES
are significantly negative only for aggregate purchase and sale transactions, and become
insignificant when regressions are run separately for purchases and sale transactions,
possibly because GIM and RES are highly correlated. We address the high correlation
between the two variables in Panel D by endogenizing RES.
24
To deal with the potential endogeneity of the decision to adopt trading restrictions,
we employ two-stage least squares estimation in Panels C and D. As instrument variables
in the first stage regression, we use the three year dummy variables used in Table 3 and
StErrForecasts, which were shown to be significant in explaining the characteristics of
firms with restriction in Table 3. We also include the number of financial analysts as an
additional instrumental variable and other exogenous variables in the first-stage
regression.27
After controlling for potential endogeneity, the coefficients of RES continue to be
negative and significant for aggregate purchase and sale transactions, indicating that selfrestrictions do reduce the overall profitability of insider trading. When purchases and sales
are examined separately, we find that the coefficients are significantly negative only for
sales, indicating that the above results are primarily driven by sale transactions. In
comparison, the coefficients for purchase transactions are all significantly positive. The
results suggest that restrictions appear to discourage insiders from exploiting negative
information only. In Panel D where we add the GIM index, we find similar results for
aggregate transactions even though the coefficients of RES become insignificantly negative
for purchase transactions. On the other hand, for sale transactions, the coefficient of RES
for 6-month abnormal returns remains significantly negative even with the GIM index
27
Appropriate instrument variables should not be correlated with the dependent variable of the original
regression, namely the profitability of insider trading, to satisfy the exclusion criterion. Since the number of
analysts and the standard errors of earnings forecasts could be correlated with the information environment
of a firm, it is possible that they are associated with the profitability of insider trading, violating the exclusion
restriction. The results should therefore be interpreted with caution due to this potential issue. Regarding the
earnings forecast dispersion, Cen, Wei and Zhang (2007) argue that the negative relation between dispersion
in earnings forecast and the cross-section of stock returns is mainly driven by the average earnings forecast
used as the denominator in typical dispersion measures, not by the standard deviation of forecasts. Here, our
dispersion measure is the standard errors of forecasts that control for differences in the number of analysts
and therefore, it is less likely to be correlated with stock returns.
25
albeit the coefficient for 3-month abnormal returns becomes insignificant. These results
are consistent with insiders being more concerned about legal troubles related to sales
preceding negative stock-price performance, but are less concerned about purchasing
shares in advance of stock-price increases.28
3.2. Do Insiders Still Earn Abnormal Profits Even After Adopting Restriction Policies?
In this section, we examine the profitability of insider trading using the Carhart
(1997)’s 4-factor model in Table 6. This allows us to directly address one of our main
questions on whether insiders continue to earn abnormal profits even after the adoption of
voluntary restriction policies on insider trading. To calculate the abnormal returns earned
by insiders after controlling for the market, size, B/M and momentum factors, in each
calendar month, we count the number of insider purchase and sale transactions of each firm
and classify each firm as either a net buying or net selling firm. We then form monthly
portfolios comprising all net purchasing (selling) firms during the past three calendar
months. This allows us to examine the performance of insider trading firms approximately
over 3 months after transactions. We calculate both equal- and value-weighted returns of
portfolios, where the weights for the value-weighted portfolio are based on the market
capitalization at the end of the previous month.
In Panel A, the results for equally-weighted portfolio returns of restricted firms
show that the intercepts (alphas) are significantly positive for purchase transactions,
suggesting that insiders continue to earn significantly positive abnormal returns of 1.17%
28
Bettis, Coles and Lemmon (2000) document that purchases made by insiders of restricted firms are less
informative only when they are traded during blackout periods while insider sales of restricted firms are less
informative regardless of trading time, indicating some differences in the impact of restriction policies on the
profitability of sales compared to that of purchases.
26
per month from their purchase transactions even after a firm adopts a restriction policy. We
find no evidence that insiders of restricted firms earn less than those of unrestricted firms
from their purchase transactions. However, for sale transactions, the alphas are not
significant in both restricted and unrestricted firms, indicating that insiders, on average, did
not earn abnormal returns from their sales after controlling for market, size, B/M and
momentum effects. As pointed out in earlier studies (for example, Lakonishok and Lee
(2001)), insiders may sell their shares due to many reasons that are not information-based
(for example, liquidity reasons), while they will most likely buy additional shares only if
they have positive private information. Therefore, we expect purchases to be more
informative than sales and the results seem to be consistent with this expectation.
We find similar results in Panel B when we examine value-weighted portfolio
returns. The only exception is for purchase transactions made by insiders of unrestricted
firms where the alpha becomes insignificant, suggesting that insiders of large unrestricted
firms did not earn significant abnormal returns from their purchase transactions.
In Panels C and D, we add three indicator variables, Sub2, Sub3 and Sub4, to
represent three sub-periods. This differs from three year dummies used in previous
regressions, as each indicator variable takes the value of one only for months during each
sub-period. As explained earlier, we divided our sample period into four sub-periods to
reflect the changes in regulations. By adding three indicators, we can test whether insiders
made significant abnormal returns in each sub-period.29 For the equally-weighted portfolio
returns reported in Panel C, the intercept of 0.6927 for the purchase portfolios of all firms
29
For this test, we report p-values from the Wald test in Panels C and D. The p-values reported in the
Alpha+Sub2 (Alpha+Sub3 or Alpha+Sub4) row are for the test of whether abnormal returns during the
second (third or fourth) sub-period (i.e., the sum of intercept and the coefficient estimate of Sub2 (Sub3 or
Sub4)) are significantly different from zero.
27
is statistically significant at the 5% level, indicating that insiders made a significant
abnormal return of 0.7% per month from their purchase transactions during the first subperiod ending in 1988. Among the indicator variables for sub-periods, we find that only
the coefficient of Sub3 is significant for purchase portfolios of all firms, suggesting that
abnormal returns were 0.93% higher per month during the third sub-period compared to
abnormal returns during the first sub-period. However, reported p-values from a Wald test
suggest that regardless of the presence of restriction policies, insiders earned significantly
positive abnormal returns from their purchase transactions in all sub-periods. The only
exception applies to purchase transactions made by insiders of unrestricted firms during
the last sub-period.
For equally-weighted returns of sale portfolios, we find that insiders of all firms
made significant abnormal returns from their sale transactions, although the abnormal
returns become insignificant once we separately examine the results for restricted and
unrestricted firms. The coefficients of the indicator variables for the second and third subperiods are significantly negative, suggesting that during the second and third sub-period,
insiders earned significantly lower abnormal returns from their sale transactions.
Moreover, the Wald test results also suggest that insiders of all firms earned significant
abnormal returns from their sale transactions only in the first sub-period, 1986-1988. For
restricted firms, we find similar results for the sub-period indicators, but the Wald test
results show that insiders of restricted firms made marginally significant negative abnormal
returns during the second and third sub-periods from their sale transactions, whereas those
of unrestricted firms did not earn any significant abnormal returns from their sale
transactions in all sub-periods.
28
The results for the value-weighted portfolio return in Panel D shows that insiders
did not seem to earn abnormal returns even from their purchase transactions during the last
sub-period starting from 2003. This suggests that insiders of large firms were not able to
exploit even the positive private information during the last sub-period. However, given
similar results for unrestricted firms during the last sub-period, it would therefore be
premature to conclude that voluntary restriction policies were effective in eliminating
informed insider trading, at least for those insiders in large firms.
In sum, the results in Table 6 show that at least for insiders of smaller firms,
voluntary corporate policies to restrict insider trading do not seem to be effective in
preventing insiders from exploiting positive inside information. Thus it may be worthwhile
for investment managers to continue paying attention to insider transactions, especially
purchase transactions made by insiders of small firms, in order to identify better trading
opportunities. The conclusion in Lakonishok and Lee (2001) that publicly available insider
trading information is useful to outside investors for purchase transactions made by insiders
of small firms seems to be still valid even now when most firms adopt voluntary restrictions
on insider trading.
3.3. Alternative identification of restricted firms.
Even though the method we use to classify restricted firms does a reasonable job
and correctly classifies restricted firms in 71% of the cases among those used in the survey
sample of Bettis, Coles and Lemmon (2000), it may still be possible that the main results
are due to the misclassification of our sample firms. This raises a concern about the validity
of our conclusions. It is certainly possible that many insiders who are aware of the legal
29
risks arising from the exploitation of private information close to a release of that
information, may voluntarily trade right after the release of important information, such as
earnings, even without restriction policies. In fact, Ke, Huddart and Petroni (2003) suggest
that insiders take advantage of extreme earnings news long before their releases to avoid
legal risk. To address this concern, we use the hand-collected data in Dai et al. (2013) to
reexamine our main results.30 They collect the data on insider trading policies by manually
searching firms’ websites as in Jagolinzer, Larcker, and Taylor (2011). However, one
disadvantage of using this approach is that we cannot identify when a restricted firm has
adopted its policy. As in Dai et al. (2013), we are forced to classify all firm-years of a firm
as restricted if there is evidence of a restriction policy on its website. Given these issues
surrounding the hand-collected data from Dai et al. (2013), we use their data only for
robustness checks.
In Table 7, we recalculate the abnormal returns reported in Tables 4 and 6 based on
this alternative classification method. For brevity, we only report the results for the 3month holding period in Panel A and the intercepts from the Carhart 4-factor model in
Panel B. Consistent with the results reported in Tables 4 and 6, the use of an alternative
classification method does not change the main results. They continue to show that insiders
of restricted firms earn significantly positive abnormal profits from their purchase
transactions but not from their sale transactions, alleviating our concern that our main
results might possibly be due to the misclassification of restricted firms.
3.4. Insider Trading Profits and Future Earnings News
30
We thank the authors of Dai, Fu, Kang and Lee (2013) for sharing the data.
30
The fact that insider trading remains profitable even though insiders have shifted
their trades to periods following earnings announcements indicates that insiders do not
trade on current earnings news but possibly do so on earnings news that is revealed in
subsequent earnings announcements. Relatedly, Ke, Huddart and Petroni (2003) show that
insiders tend to take advantage of long-term, rather than short-term, extreme earnings news.
Or alternatively, it may be the case that insider trading profits are driven by news that is
unrelated to earnings.
Table 8 explores this issue further by reporting the fraction of the 6-month abnormal
returns that can be attributed to news contained in the two subsequent earnings
announcements made within the 6-month interval that follows the pseudo transaction date.
The results are presented separately for different size groups and for restricted and
unrestricted firms, with the second row of each panel showing the 6-month abnormal
returns (BHARs), the third row showing the sum of the 3-day BHARs around the
subsequent two earnings announcements that occur during the 6-month return
measurement period, and the final row showing the 6-month BHARs net of the earnings
announcement returns. Panel A finds that insiders in small firms earn an average 6-month
BHAR of 2.12% across both purchases and sales over the 6-month period following an
insider trade. Of this amount, 0.40% is related to future earnings news. For medium sized
firms, the aggregate insider profits over the 6-month period are smaller but still
significantly positive, whereas in large firms aggregate insider profits are not any more
significant even though it is positive. On the other hand, the abnormal returns around future
earnings announcements following insider transactions are, on average, negative for
medium and large firms. Similar patterns are observed across both restricted and
31
unrestricted firms. Overall, these results indicate that insider trading profits come
predominantly from news that is not revealed in earnings announcements.
Separating purchases and sales in Panels B and C shows that earnings news does
contribute to abnormal returns following purchase transactions across all size groups. In
contrast, subsequent earnings news actually lowers the abnormal profits following sale
transactions, and this effect is larger in firms with trading restrictions, supporting the
argument that insiders try to avoid taking advantage of negative private information,
especially regarding earnings.
In sum, the results in Table 8 suggest that even after firms adopt policies to restrict
their transactions, insiders continue to make purchase transactions to take advantage of
subsequent earnings news, as well as private information regarding other aspects, while
not basing their sale transactions on subsequent earnings announcements. This result likely
reflects insiders’ desires to exploit their private information while minimizing their legal
risks, as previously highlighted in Ke, Huddart and Petroni (2003) and Cheng and Lo
(2006).
4.
Conclusions
The regulatory environment surrounding insider trading has substantially changed
over time, exposing firms to more legal risk arising from insider trading. Firms have
responded to these changes by voluntarily adopting corporate policies that restrict insider
trading to reduce their own legal risk. We first analyze the characteristics of the firms that
are likely to have adopted voluntary insider trading restrictions, and find that restricted
firms tend to have less information asymmetry. Additionally, we find that firms with more
32
anti-takeover provisions are more likely to adopt restriction policies, possibly because antitakeover provisions play a positive role in helping managers to focus more on long-term
values rather than their negative role of entrenching managers. Moreover, larger firms are
more likely to have adopted voluntary restriction polices, and more firms seem to have
adopted restriction policies in recent years.
Next, we examine whether self-restrictions have effectively eliminated informed
insider trading. The results suggest that insiders exploit negative inside information less
after firms adopt self-restrictions, whereas they continue exploiting positive private
information even after adopting corporate policies on insider trading. This asymmetric
effect of self-restrictions on the profitability of insider trading is consistent with insiders
being more cognizant of potential legal problems arising from the exploitation of negative
private information, especially the private information related to nearby earnings
announcements.31
Alternatively, it is possible that the results are driven by the fact that we do not
control for another critical corporate policy on insider trading, namely, a pre-approval
requirement by corporate counsels. Using 260 sample firms with corporate policies on
insider trading, Jagolinzer, Larcker and Taylor (2011) show that insider sales made during
blackout periods are less informative than those made during allowed trading windows
only when firms have policies requiring insiders to obtain pre-approvals from general
counsels for all their trades, notwithstanding the results being inapplicable to insider
purchases. More importantly, their results indicate that policies on the general counsel pre-
31
Similar to the conclusion made in this paper, Jagolinzer and Roulstone (2009) conclude that insider trading
behavior is significantly affected by firms’ litigation risk, but that increased trading regulation enactments do
not seem to have significant impact on the profitability of insider trading, albeit they find some evidence that
the effects of insider trading regulation are stronger at firms with greater litigation risk.
33
approval can significantly affect the profitability of insider trading. However, given that
Jagolinzer, Larcker and Taylor (2011) also find an asymmetric impact of the general
counsel rule on the profitability of sales versus purchases, the findings regarding the
strategic responses of insiders are likely to be valid even after controlling for this additional
policy.
In sum, corporate restrictions on insider trading do not seem to be effective in
preventing insiders from exploiting their private information, especially positive private
information. In addition, insiders strategically adapt their trading strategies to continue to
take advantage of their informational advantage even after firms adopt restriction policies
on insider trading. Thus, insider trading information still seems to be useful to investors
and academics as a source of information-based trading. There is also scope for policy
makers and regulators to continue paying attention to insider transactions to monitor
whether investors are being exploited by insiders even after firms adopt voluntary
restriction policies.
Acknowledgements
The authors would like to thank seminar participants at the 2009 Western Finance
Association Meetings, Korea Advanced Institute of Science and Technology, Korea
University, National University of Singapore, and Yonsei University for useful comments.
We are also grateful to an anonymous referee and the editor for valuable comments. Inmoo
Lee and John Sequeira acknowledge the financial support from National University of
Singapore through Academic Research Grants.
34
References
Bailey, W., Li, H., Mao, C.X, Zhong, R., 2003. Regulation Fair Disclosure and earnings
information: market, analyst, and corporate responses. J. Financ. 58, 2487-2514.
Bettis, J.C., Coles, J.L., Lemmon, M.L., 2000. Corporate policies restricting trading by
insiders. J. Financ. Econ. 57, 191–220.
Campbell, T.C., Gallmeyer, M., Johnson, S.A., Rutherford, J., Stanley, B.W., 2011. CEO
optimism and forced turnover. J. Financ. Econ. 101, 695-712.
Carlton, D.W., Fischel, D.R., 1983. The regulation of insider trading. Stanford Law Rev.
35, 857-895.
Carhart, M., 1997. On persistence in mutual fund performance. J. Financ. 52, 57-82.
Carpenter, J.N., Remmers, B., 2001. Executive stock option exercises and inside
information. J. Bus. 74, 513-534.
Cen, L., Wei, K.C.J., Zhang, J., 2007. Dispersion in analysts’ earnings forecasts and the
cross-section of stock returns: What is the driving factor? Working paper, Hong
Kong University of Science and Technology.
Cheng, Q., Lo, K., 2006. Insider trading and voluntary disclosures. J. Account. Res. 44,
815-848.
Dai, L., Fu, R., Kang, J.K., Lee, I., 2013. Corporate governance and insider trading.
Working paper, Korea Advanced Institute of Science and Technology.
Easley, D., Hvidkjaer, S., O’Hara, M., 2002. Is information risk a determinant of asset
returns? J. Financ. 57, 2185-2221.
Fama, E.F., French, K., 1993. Common risk factors in the returns on bonds and stocks. J.
35
Financ. Econ., 33, 3-53.
Fernandes, N., Ferreira, M.A., 2009. Insider trading law and stock price informativeness.
Rev. Financ. Stud. 22, 1846-1887.
Fishman, M.J., Hagerty, K.M., 1992. Insider trading and the efficiency of stock prices.
RAND J. Econ. 23, 106-122.
Fu, J., Liu, M.H., 2008. Anti-takeover provisions and corporate disclosure. Working Paper,
University of Kentucky.
Garfinkel, J.A., 1997. New evidence on the effects of federal regulations on insider trading:
the Insider Trading and Securities Fraud Enforcement Act (ITSFEA). J. Corp.
Financ. 3, 89–111.
Gompers, P.A., Ishii, J.L., Metrick, A., 2003. Corporate governance and equity prices, Q.
J. Econ. 118, 107-156.
Jagolinzer, A.D., 2009. SEC Rule 10b5-1 and insiders’ strategic trade. Manag. Sci. 55, 24239
Jagolinzer, A.D., Roulstone, D.T., 2009. The effects of insider trading regulation on trade
timing, litigation risk, and profitability. Working paper, Stanford University.
Jagolinzer, A.D., Larcker, D.F., Taylor, D.J., 2011. Corporate governance and the
information contents of insider trades. Working paper, Stanford University.
Jeng, L.A., 1998. Insider trading and the window of opportunity. Working paper, Boston
University.
Ke, B., Huddart, S., Petroni, K., 2003. What insiders know about future earnings and how
they use it: Evidence from insider trades. J. Account. Econ. 35, 315–46.
Lakonishok, J., Lee, I., 2001. Are insider trades informative? Rev. Financ. Stud. 14, 79–
36
111.
Lin, J., Howe, J.S., 1990. Insider trading in the OTC market. J. Financ. 45, 1273-1284.
Malmendier, U., Tate, G., Yan, J., 2011. Overconfidence and early-life experiences: The
effect of managerial traits on corporate financial policies. J. Financ. 66, 1687-1733.
Manne, H., 1966. Insider trading and the stock market. Free Press, New York, NY.
Meulbroek, L., 1992. An Empirical Analysis of Illegal Insider Trading. J. Financ. 47, 16611700.
Petersen, M. A., 2009. Estimating standard errors in finance panel data sets: Comparing
approaches. Rev. Financ. Stud. 22, 435-480.
Roulstone, D.T., 2003. The relation between insider-trading restrictions and executive
compensation. J. Account. Res. 41, 525–551.
Rozeff, M. S., Zaman, M.A., 1998. Overreaction and insider trading: evidence from growth
and value portfolios. J. Financ. 53, 701-716.
Seyhun, H.N., 1986. Insiders’ profits, costs of trading, and market efficiency. J. Financ.
Econ. 16, 189-212.
Seyhun, H.N., 1992. The effectiveness of the insider trading sanctions. J. Law Econ. 35,
149–182.
37
Table 1
Summary statistics: Annual insider trading activities. This table presents summary statistics of insider trading data over the period from 1986 to 2010 based on the
insider trading data cleaned and distributed by Thomson Financial. “Management” refers to CEOs, CFOs, chairmen of the board, directors, officers, presidents,
and vice presidents. “Large shareholders” refer to shareholders who own more than 10% of shares, but are not management. “Purch.” and “Sales” refer to open
market or private purchases and sales of common stocks. “Option” refers to the exercise/conversion of options, warrants, or convertible bonds. “ORS” are optionrelated sales, which refer to sales that occur within 6 months after the exercises of options with the number of shares sold less than or equal to the number of shares
acquired through prior exercise of options or that are indicated as option-related sales by Thomson Financial. “Fraction” is the average fraction of firms in our
sample with at least one insider trade per year. “# of trades” is the average number of trades per company per year. “% Mkt Cap” is the average ratio of the
individual company’s total insider trading dollar volume during a year to the market capitalization of the corresponding company at the beginning of the year.
“Total $” is the average aggregate total insider transaction dollar volume in 2005 million dollars per year. Firms are divided into three size groups using the cutoffs
based on the market capitalization of NYSE listed firms at the end of June of each year. Bottom 30% and top 30% are classified as small and large firms, respectively.
Management
Large shareholders
Others
Total
Purch. Sales
ORS Option
Purch. Sales
ORS Option
Purch. Sales
ORS Option
Purch. Sales ORS Option
Panel A: small firms
0.58
0.60
0.21
0.47
0.08
0.09
0.00
0.02
0.10
0.14
0.03
0.11
0.63
0.66
0.22
0.50
Fraction
4.23 10.38
3.10
3.62
9.74
9.19
0.18
0.42
1.74
5.76
0.74
1.08
4.44
9.39
2.31
2.76
# of trades
421
3,124
841
612
700
2,094
66
168
125
692
73
76
1,245 5,909 980
856
Total $(m)
1.76% 3.38% 0.12% 0.48%
0.30% 0.75% 0.06% 0.12%
0.60% 1.54% 0.19% 0.28%
%Mkt Cap 0.26% 0.85% 0.16% 0.19%
Panel B: medium firms
0.53
0.79
0.42
0.71
0.04
0.07
0.00
0.01
0.05
0.20
0.07
0.18
0.55
0.82
0.43
0.73
Fraction
3.07 25.14 10.04
9.08
13.62 19.03
0.92
1.77
0.86
10.31
2.12
2.14
3.43 21.80 7.84
7.09
# of trades
445
6,730 2,285 1,431
993
5,205
415
243
137
1,487
170
148
1,575 13,422 2,869 1,823
Total $(m)
1.00% 4.00% 0.24% 0.20%
0.05% 0.48% 0.05% 0.04%
0.17% 1.21% 0.21% 0.15%
%Mkt Cap 0.06% 0.64% 0.17% 0.12%
Panel C: large firms
0.59
0.86
0.52
0.83
0.02
0.04
0.00
0.01
0.04
0.23
0.10
0.25
0.60
0.87
0.52
0.84
Fraction
3.02 39.91 16.90 16.75
10.60 42.56
0.65
0.72
0.33
15.38
4.44
3.35
2.74 34.71 13.56 12.85
# of trades
504 13,970 5,702 3,751
1,056 6,201
139
723
85
2,426
670
294
1,645 22,598 6,511 4,768
Total $(m)
0.69% 2.28% 0.05% 0.14%
0.01% 0.15% 0.03% 0.02%
0.07% 0.48% 0.12% 0.09%
%Mkt Cap 0.02% 0.44% 0.16% 0.13%
Panel D: total firms
0.59
0.69
0.31
0.59
0.06
0.08
0.00
0.02
0.08
0.17
0.05
0.15
0.63
0.73
0.32
0.61
Fraction
4.01 19.27
7.14
7.28
10.73 13.18
0.35
0.69
1.33
8.52
1.68
1.78
4.17 16.53 5.38
5.50
# of trades
2,749 13,500 620
1,134
347
4,605
912
519
4,466 41,929 10,360 7,447
Total $(m) 1,370 23,824 8,828 5,795
1.58% 3.51% 0.14% 0.41%
0.19% 0.60% 0.05% 0.08%
0.44% 1.40% 0.20% 0.24%
%Mkt Cap 0.18% 0.78% 0.18% 0.17%
38
Table 2
Annual insider trading activities before and after restrictions. This table compares insider transactions of
firms with insider trading restrictions before and after adopting corporate restrictions. Only those insider
transactions made by management are included here. Purchases and Sales refer to open market or private
purchases and sales of common stocks. “# of trades” refers to the average number of trades per firm per year.
“% Mkt Cap” is the average ratio of the individual firm’s total insider trading dollar volume during a year to
the market capitalization of the corresponding firm at the beginning of the year. “Total $” refers to the average
annual total insider transaction dollar volume per firm in 2005 million dollars. PEC1 is the proportion of
management transactions that occur during the first sub-interval in three equal sub-intervals between two
consecutive quarterly earnings announcements. A company is classified as a restricted firm from the quarter
when PEC1 is greater than or equal to 0.75 and PEC1 being greater than 0.5 in subsequent quarters. PEC1
is calculated only for the quarters with the minimum of three management transactions. However, for the
PEC1 row in the table, PEC1 is calculated based on a single insider transaction type without the minimum
number of insider transaction requirement. The column “t-stat” tests the significance of the change in the
measure after the adoption of corporate restrictions.
Purchases
Sales
Sample period
Before
After
t-stat
Before
After
t-stat
# of trades
2.03
2.81
(-4.65)
3.41
5.34
(-5.88)
Sub-period 1
% Mkt Cap
0.08%
0.11%
(-1.92)
0.31%
0.34%
(-0.82)
0.16
0.25
(-1.92)
1.06
1.95
(-3.25)
1986  year  1988 Total $ (m)
PEC1
0.39
0.57 (-10.52)
0.37
0.59 (-15.33)
# of trades
2.72
3.61
(-3.89)
3.47
5.70
(-6.04)
Sub-period 2
% Mkt Cap
0.12%
0.12%
(0.04)
0.33%
0.48%
(-2.48)
0.26
0.33
(-0.78)
1.12
2.05
(-2.76)
1989  year  1990 Total $ (m)
PEC1
0.41
0.56
(-9.01)
0.42
0.58 (-10.79)
# of trades
2.96
4.01
(-8.06)
6.44
10.84
(-9.88)
Sub-period 3
% Mkt Cap
0.20%
0.20%
(-0.05)
0.86%
0.94%
(-1.53)
0.34
0.37
(-0.75)
3.83
7.45
(-4.33)
1991  year  2002 Total $ (m)
PEC1
0.44
0.61 (-30.20)
0.45
0.64 (-40.54)
# of trades
4.03
4.84
(-2.86)
59.69
28.52
(4.85)
Sub-period 4
% Mkt Cap
0.20%
0.16%
(0.74)
1.00%
0.73%
(2.92)
Total $ (m)
0.37
0.39
(-0.27)
8.93
7.50
(0.86)
2003  year
PEC1
0.56
0.73 (-17.70)
0.43
0.66 (-34.55)
# of trades
2.85
4.45 (-14.06)
14.52
16.80
(-1.64)
% Mkt Cap
0.17%
0.18%
(-0.64)
0.77%
0.82%
(-1.05)
All
Total $ (m)
0.31
0.37
(-1.45)
4.10
6.14
(-3.34)
PEC1
0.45
0.65 (-48.58)
0.43
0.64 (-61.74)
39
Table 3
Characteristics of firms with voluntary corporate restrictions on insider trading. This table presents the results of regression analyses where the dependent variable
is a dummy variable to indicate firms with voluntary corporate restrictions on insider trading in each quarter during our sample period 1986-2010. To identify
firms with restrictions, in each quarter, using the insider trading data cleaned and distributed by Thomson Financial, we calculate PEC1, which is the proportion of
management transactions that occur during the first sub-interval in three equal sub-intervals between two consecutive quarterly earnings announcements. A
company is classified as a restricted firm from the quarter when PEC1 is greater than or equal to 0.75 and PEC1 being greater than 0.5 in subsequent quarters. GIM
refers to the Gomper, Ishii and Metrick (2003)’s corporate governance index, PIN refers to the probability of information-based trading, StErrForecasts is the
standard error of analyst earnings forecasts, and IdioVol is the annualized idiosyncratic volatilities calculated from the Carhart’s (1997) 4-factor model. LogAsset,
LogBM, and StdDev refer to the log of total assets value, the log of (1 + B/M) where B/M refers to the book-to-market equity ratio and is set to zero for negative
values, and the annualized standard deviation of daily stock returns during a quarter, respectively. UtilityD and BankD are dummy variables indicating utility (SIC
codes 4900-4939) and bank or thrift (SIC codes 6000-6999) industries. Yeardum_1 equals to 1 if transaction date is from 1989 to the end of the sample period, else
it equals to 0. Yeardum_2 equals to 1 if transaction date is from 1991 to the end of the sample period, else it equals to 0. Yeardum_3 equals to 1 if transaction date
is from 2003 to the end of the sample period, else it equals to 0. To control for firm fixed effects, firm dummy variables are included in the regression. Numbers
in parentheses are clustered standard errors at the firm level. ***, **, and * refer to the significant results at the 1%, 5% and 10% significant levels, respectively.
Variables
GIM
Model (1)
Model (2)
Model (3)
Model (4)
-0.0008***
(0.000)
-0.1802***
(0.017)
0.0778***
(0.005)
0.0010
(0.009)
0.2091***
(0.015)
-0.0344
(0.023)
-0.0440
(0.049)
0.1941***
(0.007)
-0.5910***
(0.062)
0.0026
(0.005)
-0.1424***
(0.034)
0.0789***
(0.010)
-0.0203
(0.017)
0.2317***
(0.028)
-0.0606*
(0.033)
0.0111
(0.067)
0.2034***
(0.010)
PIN
StErrForecasts
IdioVol
LogAsset
LogBM
StdDev
BankD
UtilityD
Yeardum_1
0.0852***
(0.003)
0.0325***
(0.006)
0.0265***
(0.003)
-0.0471***
(0.016)
-0.0573
(0.043)
0.1146***
(0.004)
-0.1302***
(0.015)
0.0713***
(0.004)
0.0126**
(0.006)
0.1403***
(0.014)
-0.0353**
(0.016)
-0.0472
(0.043)
0.1236***
(0.005)
Model (5)
0.0303***
(0.004)
0.0539***
(0.008)
-0.0290**
(0.013)
0.0831***
(0.009)
-0.0105
(0.025)
-0.0567
(0.049)
Model (6)
0.0299***
(0.004)
Model (7)
0.0279***
(0.004)
-0.1140***
(0.028)
0.0512***
(0.008)
-0.0255**
(0.013)
0.1796***
(0.027)
-0.0103
(0.024)
-0.0570
(0.049)
-0.0005***
(0.000)
-0.1020***
(0.031)
0.0587***
(0.009)
-0.0375***
(0.014)
0.1734***
(0.030)
-0.0006
(0.028)
-0.0564
(0.050)
Model (8)
0.0302***
(0.006)
-0.7435***
(0.086)
-0.0276
(0.022)
-0.0113
(0.049)
0.0573***
(0.013)
-0.0146
(0.020)
0.0989**
(0.045)
-0.0090
(0.030)
-0.0358
(0.051)
40
Yeardum_2
Yeardum_3
Constant
Firm fixed effects
# of firm-quarters
R-squared
# of firms
0.1827***
(0.005)
0.2348***
(0.005)
-0.4260***
(0.019)
Yes
461,023
0.230
12,332
0.1890***
(0.006)
0.2075***
(0.005)
-0.3211***
(0.021)
Yes
428,231
0.224
12,162
0.1983***
(0.008)
0.1632***
(0.006)
-0.4129***
(0.031)
Yes
253,841
0.239
9,060
0.1754***
(0.010)
-0.3521***
(0.077)
Yes
78,483
0.244
2,946
0.1457***
(0.009)
0.0938***
(0.007)
-0.2217***
(0.067)
Yes
103,572
0.105
3,202
0.1473***
(0.009)
0.0912***
(0.007)
-0.1977***
(0.068)
Yes
103,210
0.104
3,197
0.1406***
(0.009)
0.0851***
(0.007)
-0.2234***
(0.072)
Yes
93,317
0.104
3,063
0.1130***
(0.010)
-0.1597
(0.113)
Yes
42,045
0.087
1,612
41
Table 4
Abnormal returns earned by insiders from net insider purchases and sales. This table presents the buy-and-hold
abnormal returns (BHARs) from net purchases and net sales by management of firms with insider trading
restrictions (restricted firms) and without such restrictions (unrestricted firms). A company is classified as a
restricted firm from the quarter when at least 75 % of insider trading made by management occurs in the first subinterval out of three equal sub-intervals between two consecutive quarterly earnings announcements (i.e., PEC1
≥0.75) and at least 50% in subsequent quarters (i.e., PEC1≥0.50 where PEC1 is calculated based on all
management transactions in subsequent quarters). PEC1 is calculated only for the quarters with the minimum of
3 management transactions. Bottom 30% and top 30% are classified as small and large firms, respectively based
on the market capitalization of NYSE listed firms at the end of June of each year. BHARs are calculated using
the size and B/M benchmark portfolio returns. After dividing the period between two consecutive earnings
announcements into three equal intervals, for each interval, we count the total number of insider purchases and
sales for each firm. We take the nearest trading date following the end point of each interval as the pseudo
transaction date for the interval. The pseudo transaction date is classified as a net purchase (sale) date if the
number of total purchase transactions is greater (less) than the number of total sale transactions for the interval.
BHARs are calculated starting from the pseudo transaction date for each interval. We multiply BHARs by -1 for
net sale date. # of obs refers to the number of intervals used to calculate the average BHARs. Under each BHAR,
the figure in the parentheses is the corresponding t-statistic. The column “t-stat” shows the computed t-statistic
testing the null hypothesis that the average BHARs for unrestricted firms are not significantly different from the
average BHARs for restricted firms. 3m refers to the 3-month holding period, 6m refers to the 6-month holding
period and Diff3m refers to the 3-month holding period starting from the day after the end of the first 3-month
holding period following the pseudo transaction date.
Small
Unres
Res
t-stat
Panel A: aggregate purchases and sales
#
77,907 67,889
3m
6m
Diff3m
1.47%
(9.40)
2.40%
(6.90)
0.63%
(4.17)
Panel B: purchases
#
37,696
3.18%
3m
(12.13)
5.62%
6m
(8.83)
1.68%
Diff3m
(6.44)
Panel C: sales
#
3m
6m
Diff3m
40,211
-0.12%
(-0.65)
-0.63%
(-1.78)
-0.34%
(-1.91)
1.34%
(9.58)
1.81%
(7.07)
0.31%
(2.18)
28,023
3.58%
(14.31)
5.99%
(12.57)
1.76%
(6.83)
39,866
-0.23%
(-1.39)
-1.12%
(-3.64)
-0.71%
(-4.28)
(0.78)
(2.00)
(1.92)
(-1.32)
(-0.70)
(-0.29)
(0.55)
(1.61)
(1.90)
Unres
Medium
Res
31,781
0.65%
(4.25)
0.57%
(2.05)
-0.10%
(-0.61)
47,365
0.54%
(4.49)
0.66%
(3.02)
0.04%
(0.36)
10,073
1.87%
11,344
2.11%
(8.25)
2.36%
(6.09)
0.49%
(1.94)
(8.07)
3.53%
(6.73)
1.35%
(5.13)
21,708
0.09%
(0.43)
-0.27%
(-0.71)
-0.37%
(-1.81)
36,021
0.05%
(0.33)
-0.25%
(-0.98)
-0.37%
(-2.71)
t-stat
(0.73)
(-0.40)
(-0.89)
(-0.80)
(-2.42)
(-2.79)
(0.23)
(-0.06)
(0.02)
Unres
Large
Res
16,866
0.42%
(2.50)
0.45%
(1.46)
0.19%
(1.17)
33,385
0.32%
(2.61)
0.12%
(0.56)
-0.21%
(-1.90)
4,599
0.83%
6,629
1.25%
(3.53)
1.11%
(2.88)
0.37%
(1.51)
(4.93)
1.59%
(3.96)
0.36%
(1.52)
12,267
0.27%
(1.21)
0.20%
(0.48)
0.12%
(0.58)
26,756
0.09%
(0.65)
-0.25%
(-0.95)
-0.36%
(-2.65)
t-stat
(0.67)
(1.46)
(2.65)
(-1.33)
(-1.05)
(0.03)
(1.02)
(1.68)
(2.73)
42
Table 5
Impact of insider trading restrictions on abnormal profits earned by insiders. This table presents regression analyses of the
impact of restriction policies on abnormal profits earned by management from their insider trading. Panels A and B show the
results from the OLS regressions and Panels C and D show the results from the 2SLS regressions where the dummy variable
to indicate firms with restriction policies is endogenized. In the 2SLS regressions, three time dummies (Yeardum_1,
Yeardum_2 and Yeardum_3), number of financial analysts following and standard error of analyst earnings forecasts are used
as instruments in the first stage regressions. Yeardum_1 equals to 1 if pseudo transaction date is from 1989 to the end of the
sample period, else it equals to 0. Yeardum_2 equals to 1 if pseudo transaction date is from 1991 to the end of the sample
period, else it equals to 0. Yeardum_3 equals to 1 if pseudo transaction date is from 2003 to the end of the sample period, else
it equals to 0. Buy-and-hold abnormal returns (BHARs) are the dependent variable. BHARs are calculated using the size and
B/M benchmark portfolio returns. We divide the period between two consecutive earnings announcements into three equal
intervals. For each interval we count the total number of insider purchases and sales for each firm. We take the nearest trading
date following the end point of each interval as the transaction date for the interval. The pseudo transaction date is classified
as a net purchase (sale) date if the number of total purchase transactions is greater (less) than the number of total sale
transactions for the interval. BHARs are calculated starting from the pseudo transaction date for each interval. We multiply
BHARs by -1 for net sale date. RES is equal to 1 for restricted firms and 0, otherwise. In the 2SLS regression, RES is the
endogenized variable. ResPrd is a dummy to indicate transactions made during the blackout period, GIM refers to the Gomper,
Ishii and Metrick (2003)’s corporate governance index, and IdioVol is the annualized idiosyncratic volatilities calculated from
the Carhart’s (1997) 4-factor model. LogAsset, LogBM, and StdDev, refer to the log of total assets value, the log of (1 + B/M)
where B/M refers to the book-to-market equity ratio and is set to zero for negative values, and the standard deviation of stock
returns, respectively. UtilityD and BankD are dummy variables indicating utility (SIC codes 4900-4939) and bank or thrift
(SIC codes 6000-6999) industries. To control for firm fixed effects, firm dummy variables are included in the regression.
Numbers in parentheses are clustered standard errors at the firm level in Panels A and B while they are simple standard errors
in Panels C and D. ***, **, and * refer to the significant results at the 1%, 5% and 10% significant levels, respectively.
Observations refers to the number of pseudo transaction dates used in each regression.
Panel A: results from firm fixed effects OLS
Aggregate Purch. and Sales
Variables
3m
6m
RES
-0.0036
-0.0105***
(0.002)
(0.004)
ResPrd
0.0010
0.0019
(0.002)
(0.002)
RES×ResPrd
-0.0016
-0.0021
(0.002)
(0.003)
IdioVol
-0.0403** -0.1063***
(0.017)
(0.027)
LogAsset
0.0109***
0.0169***
(0.002)
(0.003)
LogBM
0.0264***
0.0343***
(0.006)
(0.011)
StdDev
0.0787***
0.1589***
(0.014)
(0.024)
BankD
0.0088
0.0247*
(0.007)
(0.014)
UtilityD
-0.0135
-0.0408**
(0.012)
(0.019)
Constant
-0.0928*** -0.1413***
(0.011)
(0.020)
Firm fixed effects
Yes
Yes
Observations
265,603
265,603
R-squared
0.0029
0.0028
# of securities
10,388
10,388
Purchases
3m
6m
0.0064
0.0176**
(0.005)
(0.008)
-0.0018
-0.0023
(0.003)
(0.004)
-0.0010
-0.0084
(0.004)
(0.006)
-0.0746*
-0.1286**
(0.040)
(0.064)
-0.0579***
-0.1237***
(0.003)
(0.007)
0.1251***
0.2072***
(0.011)
(0.019)
0.1256***
0.2129***
(0.032)
(0.058)
0.0204
0.0692
(0.017)
(0.045)
0.0036
0.0077
(0.038)
(0.070)
0.2812***
0.6252***
(0.023)
(0.050)
Yes
Yes
94,885
94,885
0.0173
0.0239
9,182
9,182
Panel B: results from firm fixed effects OLS with the GIM index
RES
-0.0099*** -0.0171**
-0.0027
(0.004)
(0.007)
(0.008)
ResPrd
-0.0007
0.0013
-0.0018
-0.0128
(0.013)
0.0014
Sales
3m
6m
-0.0056* -0.0175***
(0.003)
(0.005)
0.0009
0.0010
(0.001)
(0.002)
-0.0017
0.0003
(0.002)
(0.003)
-0.0830*** -0.2123***
(0.016)
(0.025)
0.0444*** 0.0863***
(0.002)
(0.003)
-0.1221*** -0.2421***
(0.007)
(0.013)
0.0816*** 0.1860***
(0.014)
(0.022)
0.0022
0.0044
(0.010)
(0.016)
-0.0295
-0.0658*
(0.022)
(0.036)
-0.2499*** -0.4780***
(0.013)
(0.023)
Yes
Yes
170,718
170,718
0.0175
0.0315
9,107
9,107
-0.0059
(0.005)
-0.0017
-0.0107
(0.009)
-0.0031
43
(0.002)
0.0008
(0.002)
GIM
-0.0010
(0.001)
IdioVol
-0.0022
(0.029)
LogAsset
0.0173***
(0.002)
LogBM
-0.0028
(0.010)
StdDev
0.0401*
(0.024)
BankD
0.0092
(0.011)
UtilityD
-0.0021
(0.015)
Constant
-0.1304***
(0.019)
Firm fixed effects
Yes
Observations
90,757
R-squared
0.0022
# of securities
2,952
(0.003)
0.0020
(0.003)
-0.0022
(0.002)
-0.0099
(0.048)
0.0325***
(0.005)
-0.0131
(0.018)
0.0710*
(0.040)
0.0120
(0.018)
-0.0284
(0.021)
-0.2427***
(0.037)
Yes
90,757
0.0032
2,952
(0.004)
-0.0019
(0.005)
0.0053***
(0.002)
0.0471
(0.065)
-0.0423***
(0.005)
0.1220***
(0.017)
0.0335
(0.056)
-0.0096
(0.024)
0.0465
(0.041)
0.2185***
(0.040)
Yes
23,924
0.0198
2,531
(0.006)
0.0013
(0.008)
0.0087**
(0.004)
0.1519
(0.107)
-0.0848***
(0.009)
0.2273***
(0.029)
0.0028
(0.092)
0.0010
(0.031)
0.0374
(0.059)
0.4580***
(0.069)
Yes
23,924
0.0330
2,531
(0.002)
0.0014
(0.002)
-0.0043***
(0.001)
-0.0816***
(0.031)
0.0368***
(0.003)
-0.1398***
(0.013)
0.0627**
(0.026)
0.0050
(0.014)
-0.0188
(0.021)
-0.1992***
(0.023)
Yes
66,833
0.0158
2,819
(0.003)
0.0044
(0.003)
-0.0081***
(0.003)
-0.1815***
(0.053)
0.0713***
(0.006)
-0.2848***
(0.025)
0.1306***
(0.045)
-0.0016
(0.024)
-0.0478
(0.034)
-0.3824***
(0.045)
Yes
66,833
0.0298
2,819
Panel C: results from 2SLS
RES
-0.0063***
(0.002)
ResPrd
0.0001
(0.002)
RES×ResPrd
-0.0002
(0.002)
IdioVol
-0.0394***
(0.010)
LogAsset
0.0172***
(0.001)
LogBM
0.0142***
(0.003)
StdDev
0.0786***
(0.008)
BankD
0.0144**
(0.007)
UtilityD
-0.0099
(0.014)
Constant
-0.1375***
(0.007)
Firm fixed effects
Yes
Observations
201,656
# of securities
7,918
-0.0131***
(0.003)
-0.0004
(0.002)
0.0003
(0.003)
-0.1187***
(0.015)
0.0320***
(0.002)
0.0015
(0.005)
0.1741***
(0.012)
0.0363***
(0.011)
-0.0373*
(0.022)
-0.2476***
(0.011)
Yes
201,656
7,918
0.0086*
(0.005)
0.0000
(0.003)
-0.0014
(0.004)
-0.0228
(0.020)
-0.0487***
(0.002)
0.1194***
(0.006)
0.1259***
(0.017)
0.0488***
(0.016)
0.0258
(0.029)
0.2372***
(0.018)
Yes
59,795
6,757
0.0225***
(0.007)
-0.0041
(0.005)
-0.0068
(0.007)
-0.0487
(0.032)
-0.0991***
(0.004)
0.2008***
(0.010)
0.2328***
(0.027)
0.1222***
(0.026)
0.0389
(0.046)
0.4973***
(0.028)
Yes
59,795
6,757
-0.0097***
(0.003)
-0.0020
(0.002)
0.0016
(0.002)
-0.0964***
(0.011)
0.0420***
(0.001)
-0.1109***
(0.004)
0.0755***
(0.009)
-0.0000
(0.008)
-0.0251
(0.017)
-0.2461***
(0.008)
Yes
141,861
7,053
-0.0204***
(0.004)
-0.0021
(0.003)
0.0031
(0.003)
-0.2337***
(0.016)
0.0813***
(0.002)
-0.2378***
(0.006)
0.1781***
(0.013)
0.0057
(0.011)
-0.0584**
(0.024)
-0.4693***
(0.011)
Yes
141,861
7,053
-0.0007
(0.008)
-0.0020
(0.005)
-0.0026
(0.006)
-0.0078
(0.011)
0.0010
(0.007)
-0.0009
(0.009)
-0.0050
(0.004)
-0.0018
(0.003)
0.0011
(0.003)
-0.0125**
(0.006)
-0.0034
(0.004)
0.0038
(0.004)
RES×ResPrd
Panel D: results from 2SLS with the GIM Index
RES
-0.0087** -0.0176***
(0.004)
(0.005)
ResPrd
-0.0011
0.0001
(0.002)
(0.003)
RES×ResPrd
0.0006
0.0018
(0.003)
(0.004)
44
GIM
-0.0010
(0.001)
IdioVol
-0.0099
(0.017)
LogAsset
0.0180***
(0.002)
LogBM
-0.0121**
(0.005)
StdDev
0.0423***
(0.015)
BankD
0.0068
(0.008)
UtilityD
-0.0033
(0.014)
Constant
-0.1335***
(0.013)
Firm fixed effects
Yes
Observations
85,940
# of securities
2,844
-0.0025**
(0.001)
-0.0237
(0.025)
0.0351***
(0.002)
-0.0373***
(0.008)
0.0750***
(0.022)
0.0086
(0.012)
-0.0303
(0.020)
-0.2510***
(0.019)
Yes
85,940
2,844
0.0069***
(0.002)
0.0613
(0.040)
-0.0426***
(0.004)
0.1246***
(0.010)
0.0235
(0.036)
-0.0069
(0.022)
0.0468
(0.029)
0.2105***
(0.034)
Yes
21,732
2,407
0.0110***
(0.003)
0.1621***
(0.058)
-0.0847***
(0.006)
0.2455***
(0.014)
-0.0100
(0.052)
-0.0014
(0.032)
0.0371
(0.042)
0.4432***
(0.050)
Yes
21,732
2,407
-0.0041***
(0.001)
-0.0761***
(0.019)
0.0370***
(0.002)
-0.1439***
(0.007)
0.0585***
(0.016)
0.0035
(0.009)
-0.0198
(0.016)
-0.2037***
(0.014)
Yes
64,208
2,710
-0.0079***
(0.001)
-0.1588***
(0.027)
0.0724***
(0.002)
-0.3026***
(0.009)
0.1139***
(0.023)
-0.0006
(0.013)
-0.0488**
(0.023)
-0.3925***
(0.020)
Yes
64,208
2,710
45
Table 6
Abnormal profits earned by insiders: Measured based on the 4-factor model. This table presents the alphas from
the extended Fama-French 3-factor model. A company is classified as a restricted firm from the quarter when at
least 75 % of insider trading made by management occurs in the first sub-interval out of three equal sub-intervals
between two consecutive quarterly earnings announcements (i.e., PEC1 ≥ 0.75) and at least 50% in subsequent
quarters (i.e., PEC1 ≥ 0.50 where PEC1 is calculated based on all management transactions in subsequent
quarters). PEC1 is calculated only for the quarters with the minimum of 3 insider transactions made by
management. For each calendar month during February 1986 and March 2011, we identified all the net purchases
/net sales firms in each month during the last three months (not including the current month) and form the monthly
portfolio comprising firms with net purchases (sales) in any months during the past three months. To identify the
net purchases/net sales firms for each calendar month, we count the total number of management purchases and
sales for each firm in each month during the last three months. Net purchases (sales) firms are those where the
number of total purchases transaction is greater (less) than the number of total sale transactions in any months
during the last three months. The dependent variable is the monthly equal- or value-weighted return of the
portfolio, using the market values at the end of the previous month as the weight for value-weighted return
calculations. Independent variables are 4 factors (market risk premium, HML, SMB, and momentum factors)
from the extended Fama-French 3-factor model (Carhart (1997) and Fama and French (1993)). In Panels C and
D, three dummy variables to indicate three sub-periods are included. Sub2, Sub3 and Sub4 refer to the sub-period
2 between 1989 and 1990, the sub-period 3 between 1991 and 2002, and the last sub-period starting from 2003,
respectively. # of obs refers to the number of calendar months used in the regression. Numbers in parentheses
are robust standard errors. ***, **, and * refer to the significant results at the 1%, 5% and 10% significance levels,
respectively. Alpha+ Sub2 (Sub3 or Sub4) rows report p-values from the Wald test to check whether the sum of
the intercept and the coefficient of Sub2 (Sub3 or Sub4) is different from zero.
Panel A: equally-weighted portfolio
All Firms
Variables
Purchases
Sales
1.1100***
-0.0039
Alpha
(0.107)
(0.053)
0.9489***
-1.0348***
Market
(0.024)
(0.012)
0.6759***
-0.6661***
SMB
(0.033)
(0.017)
0.2793***
-0.0706***
HML
(0.037)
(0.018)
-0.3095***
0.0324***
WML
(0.022)
(0.011)
Observations
302
302
R-squared
0.9059
0.9756
Adj. R-sqrd
0.9046
0.9753
Res Firms
Purchases
Sales
1.1664***
-0.0646
(0.107)
(0.056)
0.9805***
-1.0484***
(0.024)
(0.013)
0.6729***
-0.6357***
(0.033)
(0.017)
0.2795***
-0.0702***
(0.037)
(0.019)
-0.3116***
0.0227**
(0.022)
(0.011)
302
302
0.9097
0.9735
0.9085
0.9731
Unres Firms
Purchases
Sales
0.9435***
0.1595
(0.135)
(0.099)
0.8482*** -0.9846***
(0.031)
(0.022)
0.6554*** -0.7692***
(0.042)
(0.031)
0.2413***
-0.0394
(0.046)
(0.034)
-0.2899*** 0.0771***
(0.028)
(0.020)
302
302
0.8347
0.9228
0.8324
0.9217
Panel B: value-weighted portfolio
0.2893***
0.0288
Alpha
(0.079)
(0.033)
0.9836***
-1.0156***
Market
(0.018)
(0.008)
-0.0909***
0.0835***
SMB
(0.025)
(0.010)
0.1793***
0.0881***
HML
(0.027)
(0.011)
-0.1145*** -0.0566***
WML
(0.016)
(0.007)
Observations
302
302
R-squared
0.9221
0.9864
Adj. R-sqrd
0.9211
0.9862
0.3332***
(0.083)
0.9770***
(0.019)
-0.0798***
(0.026)
0.1610***
(0.029)
-0.1202***
(0.017)
302
0.9139
0.9127
0.1261
(0.163)
1.0169***
(0.037)
-0.1561***
(0.051)
0.3088***
(0.056)
-0.0781**
(0.033)
302
0.7388
0.7352
0.0254
(0.034)
-1.0061***
(0.008)
0.0865***
(0.011)
0.0988***
(0.012)
-0.0602***
(0.007)
302
0.9852
0.9850
0.1045
(0.102)
-1.0689***
(0.023)
0.0265
(0.032)
-0.0230
(0.035)
-0.0183
(0.021)
302
0.8934
0.8920
46
Panel C: equally-weighted portfolio with sub-period dummies
0.6927**
0.3023**
0.6441**
Alpha
(0.293)
(0.152)
(0.293)
0.9443***
-1.0347***
0.9762***
Market
(0.023)
(0.012)
(0.023)
0.6801***
-0.6668***
0.6775***
SMB
(0.032)
(0.017)
(0.032)
0.2722***
-0.0716***
0.2736***
HML
(0.036)
(0.018)
(0.035)
-0.3249***
0.0350***
-0.3273***
WML
(0.021)
(0.011)
(0.021)
0.6189
-0.5001**
0.8196*
Sub2
(0.462)
(0.239)
(0.461)
0.9303***
-0.3756**
1.0321***
Sub3
(0.327)
(0.169)
(0.327)
-0.1896
-0.2704
-0.0671
Sub4
(0.341)
(0.177)
(0.341)
Alpha+ Sub2
0.000
0.287
0.000
Alpha+ Sub3
0.000
0.344
0.000
Alpha+ Sub4
0.004
0.725
0.001
Observations
302
302
302
R-squared
0.9137
0.9761
0.9174
Adj. R-sqrd
0.9117
0.9755
0.9154
0.2035
(0.158)
-1.0480***
(0.013)
-0.6382***
(0.017)
-0.0715***
(0.019)
0.0267**
(0.011)
-0.5387**
(0.249)
-0.3648**
(0.176)
-0.1625
(0.184)
0.083
0.046
0.665
302
0.9741
0.9735
0.8543**
(0.375)
0.8426***
(0.030)
0.6595***
(0.041)
0.2308***
(0.045)
-0.3055***
(0.027)
0.0514
(0.591)
0.6389
(0.418)
-0.6236
(0.436)
0.045
0.000
0.305
302
0.8446
0.8409
0.4060
(0.283)
-0.9858***
(0.022)
-0.7662***
(0.031)
-0.0409
(0.034)
0.0747***
(0.020)
-0.2510
(0.445)
-0.1781
(0.315)
-0.4254
(0.329)
0.654
0.114
0.909
302
0.9233
0.9215
Panel D: value-weighted portfolio with sub-period dummies
0.1061
0.0559
0.1658
Alpha
(0.223)
(0.094)
(0.236)
0.9815***
-1.0163***
0.9752***
Market
(0.018)
(0.008)
(0.019)
-0.0881***
0.0825***
-0.0758***
SMB
(0.025)
(0.010)
(0.026)
0.1767***
0.0855***
0.1599***
HML
(0.027)
(0.011)
(0.029)
-0.1220*** -0.0578***
-0.1273***
WML
(0.016)
(0.007)
(0.017)
0.3320
-0.1564
0.4006
Sub2
(0.351)
(0.149)
(0.371)
0.4182*
0.0383
0.3710
Sub3
(0.248)
(0.105)
(0.263)
-0.1113
-0.0942
-0.1103
Sub4
(0.259)
(0.110)
(0.274)
Alpha+ Sub2
0.108
0.384
0.050
Alpha+ Sub3
0.000
0.051
0.000
Alpha+ Sub4
0.969
0.497
0.694
Observations
302
302
302
R-squared
0.9248
0.9866
0.9162
Adj. R-sqrd
0.9230
0.9863
0.9142
0.1611*
(0.097)
-1.0069***
(0.008)
0.0850***
(0.011)
0.0954***
(0.012)
-0.0606***
(0.007)
-0.3523**
(0.153)
-0.0847
(0.108)
-0.1998*
(0.113)
0.108
0.123
0.506
302
0.9856
0.9852
-0.3232
(0.465)
1.0142***
(0.037)
-0.1613***
(0.051)
0.2996***
(0.056)
-0.0856**
(0.034)
0.1557
(0.732)
0.7973
(0.518)
0.2016
(0.540)
0.768
0.046
0.662
302
0.7425
0.7364
-0.0546
(0.290)
-1.0696***
(0.023)
0.0318
(0.032)
-0.0205
(0.035)
-0.0232
(0.021)
0.5478
(0.457)
0.2543
(0.323)
-0.0114
(0.337)
0.164
0.177
0.704
302
0.8945
0.8919
47
Table 7
Abnormal returns earned by insiders based on an alternative method to identify restricted firms. In this table, a firm is classified
as a restricted firm if there is an evidence of voluntary insider trading policy being present in its website. In Panel A, the
average buy-and-hold abnormal returns (BHARs) from net purchases and net sales by management of firms with insider
trading restrictions (restricted firms) and without such restrictions (unrestricted firms) are reported. Bottom 30% and top 30%
are classified as small and large firms, respectively based on the market capitalization of NYSE listed firms at the end of June
of each year. BHARs are calculated using the size and B/M benchmark portfolio returns. We divide the period between two
consecutive earnings announcements into three equal intervals. For each interval we count the total number of insider
purchases and sales for each firm. We take the nearest trading date following the end point of each interval as the pseudo
transaction date for the interval. The pseudo transaction date is classified as a net purchases (sales) date if the number of total
purchase transactions is greater (less) than the number of total sale transactions for the interval. BHARs are calculated starting
from the pseudo transaction date for each interval. We multiply BHARs by -1 for net sales date. # of obs refers to the number
of intervals used to calculate the average BHARs. Under each BHAR, the figure in the parentheses is the corresponding tstatistic. The column “t-stat” shows the computed t-statistic testing the null hypothesis that the average BHARs for unrestricted
firms are not significantly different from the average BHARs for restricted firms. 3m refers to the 3-month holding period,
6m refers to the 6-month holding period and Diff3m refers to the 3-month holding period starting from the day after the end
of the first 3-month holding period following the pseudo transaction date. In Panel B, the alphas from the extended FamaFrench 3-factor model are reported. For each calendar month during February 1986 and March 2011, we identified all the net
purchases/net sales firms in each month during the last three months (not including the current month) and form the monthly
portfolio comprising firms with net purchases (sales) in any months during the past three months. To identify the net
purchases/net sales firms for each calendar month, we count the total number of management purchases and sales of each firm
in each month during the last three months. Net purchases (sales) firms are those where the number of total purchase
transactions is greater (less) than the number of total sale transactions in any months during the last three months. The
dependent variable is the monthly equal- or value-weighted return of the portfolio, where the market values at the end of the
previous month are used for the weights in value-weighted return calculations. Independent variables are 4 factors (market
risk premium, HML, SMB, and momentum factors) from the extended Fama-French 3-factor model (Carhart (1997) and Fama
and French (1993)). Numbers in parentheses are robust standard errors. ***, **, and * refer to the significant results at the
1%, 5% and 10% significance levels, respectively.
Panel A: abnormal BHARs
Tran.
type
Unres
#
12,254
Agg.
1.23%
3m
Purch.
(3.41)
& Sales
0.30%
Diff 3m
(0.92)
#
4,329
6.09%
3m
Purch.
(9.59)
3.13%
Diff 3m
(5.07)
#
7,925
-1.43%
3m
Sales
(-3.31)
-1.25%
Diff 3m
(-3.27)
Small
Res
28,031
1.13%
(4.82)
0.28%
(1.22)
9,095
7.45%
(16.05)
5.54%
(11.31)
18,936
-1.90%
(-7.34)
-2.24%
(-9.24)
t-stat
(0.28)
(0.04)
(-2.03)
(-3.67)
(1.22)
(2.71)
Unres
7,421
0.81%
(2.44)
0.12%
(0.40)
1,454
3.39%
(4.48)
0.55%
(0.85)
5,967
0.19%
(0.52)
0.01%
(0.04)
Medium
Res
26,210
0.19%
(1.13)
-0.28%
(-1.71)
4,921
3.61%
(9.87)
2.60%
(6.55)
21,289
-0.60%
(-3.16)
-0.95%
(-5.32)
t-stat
(2.26)
(1.45)
(-0.29)
(-2.91)
(2.73)
(3.23)
Panel B: abnormal returns (intercept) from the Carhart 4-factor model
All firms
Res firms
Purchases
Sales
Purchases
Sales
1.1100***
-0.0039
1.4838***
-0.3628***
Equally-weighted
(0.107)
(0.053)
(0.124)
(0.082)
0.2893***
0.0288
0.3744***
-0.1281*
Value-weighted
(0.079)
(0.033)
(0.136)
(0.074)
Unres
5,015
0.48%
(1.87)
-0.23%
(-1.03)
880
2.23%
(3.48)
0.28%
(0.49)
4,135
0.10%
(0.37)
-0.34%
(-1.28)
Large
Res
19,570
0.03%
(0.17)
-0.40%
(-2.47)
3,279
1.27%
(3.69)
0.43%
(1.39)
16,291
-0.22%
(-1.09)
-0.57%
(-2.99)
t-stat
(1.77)
(0.65)
(1.48)
(-0.23)
(1.18)
(0.80)
Unres firms
Purchases
Sales
1.7025*** -0.1321
(0.179)
(0.098)
0.6422*** 0.2202
(0.215)
(0.143)
48
Table 8
Insider trading profits and future earnings news. This table reports the fraction of the 6-month buy-and-hold abnormal returns (BHARs) to insider transactions
made by management, which can be attributed to news contained in subsequent earnings. A company is classified as a restricted firm (Res) from the quarter when
at least 75 % of insider trading made by management occurs in the first sub-interval out of three equal sub-intervals between two consecutive quarterly earnings
announcements (i.e., PEC1≥0.75) and at least 50% in subsequent quarters (i.e., PEC1≥0.50 where PEC1 is calculated based on all management transactions in
subsequent quarters). PEC1 is calculated only for the quarters with the minimum of 3 management transactions. Bottom 30% and top 30% are classified as small
and large firms, respectively based on the market capitalization of NYSE listed firms at the end of June of each year. BHARs are calculated using the size and B/M
benchmark portfolio returns. We divide the period between two consecutive earnings announcements into three equal intervals. For each interval we count the total
number of insider purchases and sales for each firm. We take the nearest transaction date following the end point of each interval as the pseudo transaction date for
the interval. The pseudo transaction date is classified as a net purchases (sales) date if the number of total purchase transactions is greater (less) than the number of
total sale transactions for the interval. BHARs are calculated starting from the pseudo transaction date for each interval. We multiply BHARs by -1 for net sales
date. For each pseudo transaction date, we calculate the 3-day BHARs for any future earnings announcements that are made within the 6 month window starting
from the pseudo transaction date. # of obs refers to the number of net insider transaction dates used to calculate the average BHARs. Under each BHAR, the figure
in the parentheses is the corresponding t-statistic. The column “t-stat” shows the computed t-statistic testing the null hypothesis that the average BHARs for
unrestricted firms are not significantly different from the average BHARs for restricted firms.
Unres
Panel A: aggregate sales and purchases
79,293
# of pseudo transaction dates
6-month BHAR
3-day BHAR around future
announcement dates
6-month BHAR net of 3-day
BHAR
Panel B: purchases
# of pseudo transaction dates
6-month BHAR
3-day BHAR around future
announcement dates
6-month BHAR net of 3-day
BHAR
Res
Small
Allfirm
2.40%
(6.97)
0.49%
(6.95)
1.91%
(5.72)
69,181
1.79%
(7.04)
0.30%
(3.58)
1.50%
(6.23)
148,474
2.12%
(9.64)
0.40%
(7.42)
1.72%
(8.11)
38,359
5.62%
(8.92)
1.18%
(11.08)
4.45%
(7.22)
28,457
5.95%
(12.57)
1.37%
(10.14)
4.58%
(10.02)
66,816
5.76%
(13.83)
1.26%
(14.93)
4.50%
(11.08)
t-stat
(2.10)
(2.61)
(1.49)
(-0.64)
(-1.64)
(-0.27)
Unres
Medium
Res
Allfirm
33,133
0.46%
(1.59)
-0.13%
(-1.29)
0.58%
(2.12)
48,970
0.61%
(2.82)
-0.35%
(-4.41)
0.96%
(4.79)
82,103
0.55%
(3.12)
-0.26%
(-4.17)
0.81%
(4.88)
10,465
2.35%
(6.08)
0.49%
(3.77)
1.86%
(4.88)
11,758
3.55%
(6.69)
1.01%
(6.31)
2.54%
(5.15)
22,223
2.98%
(8.85)
0.77%
(7.20)
2.22%
(6.97)
Large
t-stat
(-0.70)
(2.86)
(-1.76)
(-2.56)
(-3.72)
(-1.51)
Unres
Res
Allfirm
18,014
0.44%
(1.43)
-0.08%
(-0.76)
0.52%
(1.91)
35,359
0.14%
(0.66)
-0.19%
(-2.44)
0.32%
(1.67)
53,373
0.24%
(1.39)
-0.15%
(-2.41)
0.39%
(2.47)
4,875
1.10%
(2.71)
0.32%
(2.13)
0.78%
(1.97)
6,971
1.76%
(4.37)
0.57%
(3.45)
1.18%
(3.15)
11,846
1.48%
(5.11)
0.47%
(4.03)
1.02%
(3.71)
t-stat
(1.40)
(1.41)
(0.96)
(-1.49)
(-1.72)
(-0.96)
Panel C: sales
49
#
6-month BHAR
3-day BHAR around future
announcement dates
6-month BHAR net of 3-day
BHAR
40,934
-0.62%
(-1.76)
-0.16%
(-1.53)
-0.47%
(-1.39)
40724
-1.12%
(-3.65)
-0.46%
(-4.03)
-0.66%
(-2.33)
81,658
-0.87%
(-3.69)
-0.31%
(-4.02)
-0.56%
(-2.55)
(1.63)
(3.18)
(0.68)
22,668
-0.42%
(-1.07)
-0.41%
(-2.97)
0.00%
(-0.01)
37,212
-0.32%
(-1.25)
-0.78%
(-7.72)
0.46%
(1.95)
59,880
-0.36%
(-1.62)
-0.64%
(-7.74)
0.28%
(1.37)
(-0.38)
(3.92)
(-1.90)
13,139
0.20%
(0.48)
-0.23%
(-1.60)
0.43%
(1.16)
28,388
-0.26%
(-1.05)
-0.38%
(-3.97)
0.11%
(0.48)
41,527
-0.12%
(-0.54)
-0.33%
(-4.15)
0.21%
(1.06)
(1.81)
(1.64)
(1.30)
50
Earnings Ann at t-1
Earnings Ann at t
Period1
N1
Period2
N2
Period3
N3
N1, N2 and N3 are number of insider transactions during period 1, period 2 and period 3, respectively.
,
and
, where N1 + N2 + N3 ≥ 3
Fig. 1. Computation of PEC1, PEC2 and PEC3. This figure shows how we compute the
percentage of insider transactions made by management within each sub-interval. For each
insider trading date, two subsequent earnings announcement dates around this transaction
date are identified. We then divide the interval into three equal-sized sub-intervals. The
percentage of insider transactions for the first sub-interval denoted PEC1 is then calculated
by dividing the number of management transactions recorded in the first sub-interval by
the total number of transactions within the interval between two subsequent earning
announcement dates. PEC2 and PEC3 are calculated in a similar manner. PEC1, PEC2
and PEC3 are calculated only for the quarters with the minimum of 3 management
transactions during the interval.
51
Fig. 2. Time plots of PEC1, PEC2 and PEC3. This figure shows the time plots of PEC1,
PEC2 and PEC3 for small, medium, and large size groups, as well as for total sample firms.
PEC1 is the proportion of insider transactions made by management, which occur within
the first sub-interval, while PEC2 and PEC3 refer to the corresponding proportions in the
second and third sub-intervals, respectively, where the interval between two subsequent
earnings announcement dates surrounding an insider transaction is divided into three equallength sub-intervals. PEC1, PEC2 and PEC3 are calculated only for the quarters with the
minimum of three management transactions. Firms are divided into three size groups using
the cutoffs based on the market capitalization of NYSE listed firms at the end of June of
each year. Bottom 30% and top 30% are classified as small and large firms, respectively.
52
Small
Medium
Large
PIN-Small
PIN-Medium
PIN-Large
2010
2008
2006
2004
2002
2000
1998
1996
1994
1992
1990
1988
1986
90.00%
80.00%
70.00%
60.00%
50.00%
40.00%
30.00%
20.00%
10.00%
0.00%
Fig. 3. Ratio of firms with insider trading restrictions over the period 1986 to 2010 and
average PIN over the period 1986 to 2001. This figure shows the ratio of firms with insider
trading restrictions over the period from 1986 to 2010 and the average probability of
information-based trading (PIN) over the period from 1986 to 2001. A company is
classified as a restricted firm from the quarter when at least 75 % of insider trading made
by management occurs in the first sub-interval (i.e., PEC1 ≥ 0.75) out of three equal subintervals between two consecutive quarterly earnings announcements and at least 50% in
subsequent quarters (i.e., PEC1 ≥ 0.50 where PEC1 is calculated based on all management
transactions in subsequent quarters). The annual PIN estimates for all NYSE/Amex
common stocks during 1983 and 2001 were made available by Soeren Hvidkjaer in
“https://sites.google.com/site/hvidkjaer/data”, and Easley, Hvidkjaer and O’Hara (2002)
describe the procedures used to calculate the PIN measure. Firms are divided into three
size groups using the cutoffs based on the market capitalization of NYSE listed firms at the
end of June of each year. Bottom 30% and top 30% are classified as small and large firms,
respectively.
53