Professional Documents
Culture Documents
Dai (lili.dai@anu.edu.au) is from the ANU College of Business and Economics, Australian National University,
Canberra, ACT 0200, Australia. Parwada (j.parwada@unsw.edu.au) and Zhang (bohui.zhang@unsw.edu.au) are from the
School of Banking and Finance, UNSW Australia, Sydney, NSW 2052, Australia. We thank Douglas Skinner (the editor),
an anonymous referee, Rene Adams, Sudipta Basu, Oleg Chuprinin, David Feldman, Neal Galpin, Ian Gow, Mark
Greenblatt, Wayne Guay, Rebecca Hann, Jonathan Karpoff, Lawrence Kryzanowski, Mark Lang, Christian Leuz,
Ningzhong Li, Ronald Masulis, Greg Miller, Peter Pham, Richard Roll, Jianfeng Shen, Ahmed Tahoun, Dan Taylor, Irem
Tuna, Terry Walter, Ivo Welch, Sarah Zechman, and Qiaoqiao Zhu. We thank seminar participants at the Australian
National University, UCLA, the University of Miami, UNSW Australia, and the University of Washington. We also thank
participants at the 2014 Journal of Accounting Research Conference, Chicago; the 2014 European Financial Management
Annual Meetings, Rome; the 2014 World Finance Conference, Venice; the 2013 Northern Finance Association
Conference, Quebec City; the 2013 FIRN Conference, Hunter Valley; the 2013 Behavioural Finance and Capital Markets
Conference, Adelaide; and the 2013 RSFAS Research Summer Camp, Batemans Bay, for providing many helpful
comments and suggestions.
ABSTRACT
We investigate whether the media plays a role in corporate governance by disseminating news. Using
a comprehensive dataset of corporate and insider news coverage for the 2001-2012 period, we show
that the media reduces insiders future trading profits by disseminating news on prior insiders trades
available from regulatory filings. We find support for three economic mechanisms underlying the
disciplining effect of news dissemination: the reduction of information asymmetry, concerns
regarding litigation risk, and the impact on insiders personal wealth and reputation. Our findings
provide new insights into the real effect of news dissemination.
1. Introduction
That the media plays a role in corporate governance is well known.1 What is less clear is how
the governance effect of the media works. Existing evidence supports the notion that the media
disciplines managers by creating content that exposes governance problems (Miller [2006], Dyck,
Volchkova, and Zingales [2008]). We use evidence from a large sample of insider trading filings to
investigate whether the medias news dissemination role directly affects governance.
The SEC requires insiders to report their trading activities on Form 4 filings, which are
typically disseminated through the media. 2 This setting provides us with a useful opportunity to
examine the effect of the medias dissemination role on corporate governance, and specifically in
restricting insiders trading profits. Since news dissemination increases the breadth of coverage and
the attention of investors through repetition (Fang and Peress [2009], Bushee et al. [2010],
Blankespoor, Miller, and White [2014], Peress [2014]), we conjecture that the media reduces the
profitability of insiders future transactions by disseminating regulatory releases of prior insider
trading activities. We call this view, which forms our main hypothesis, disciplining via dissemination.
Our main hypothesis is based on three mechanisms. First, by disseminating news on prior
insider trades, insiders information advantage is reduced and prices adjust more quickly to the news
(Bushee et al. [2010], Tetlock [2010]), directly reducing the profitability of future insiders trades. We
refer to this attenuation effect of the media on insiders profits as the information asymmetry channel.
Second, recent studies show that litigation can restrict and punish insiders opportunistic behavior,
especially their selling activities (Cheng, Huang, and Li [2013], Billings and Cedergren [2014]).
Therefore, because of concerns regarding litigation risk, insiders in firms in the media spotlight avoid
opportunistic trading strategies and thereby earn reduced profits. We refer to this effect of news
See, for example, Dyck and Zingales [2002], Miller [2006], Core, Guay, and Larcker [2008], Dyck, Volchkova, and
Zingales [2008], Joe, Louis, and Robinson [2009], Dyck, Morse, and Zingales [2010], Kuhnen and Niessen [2012], and
Liu and McConnell [2013].
2
Regarding the typical financial press coverage of insider trades, see, for example, Lock-Up Expires for LinkedIn
Shares, and Insiders Cash In, New York Times, 22 November 2011. Moreover, numerous websites are dedicated to
insiders trading activities gleaned from SEC filings, for example, www.secform4.com and www.insiderslab.com. Stories
featuring data from such websites frequently appear in prominent newspapers and business magazines.
coverage as the litigation risk channel. Third, since the dissemination of insider trading news can
adversely affect executives personal wealth and reputation (Dyck, Volchkova, and Zingales [2008]),
we expect that the disciplining effect of news is more pronounced when executives have a greater
amount of personal capital tied to firms. We refer to this mechanism as the capital-at-risk channel.
We examine our hypotheses using more than 1.375 million trades by U.S. corporate insiders
from Thomson Reuters.3 Corporate news coverage data from RavenPack provide us with the number
of Dow Jones news releases that are associated with the insiders firms. Following Jagolinzer, Larcker,
and Taylor [2011], we compute insiders profits as the alpha earned during the 180-day window after
an insiders buy or sell transaction. We investigate whether insiders consistently earn future abnormal
profits when they face news coverage on their prior trades. Our examination is similar to recent
studies of corporate activities conditional on prior media coverage (e.g., Core, Guay, and Larcker
[2008], Kuhnen and Niessen [2012] on executive compensation; Braggion and Giannetti [2013] on
limited voting shares). Consistent with our disciplining via dissemination hypothesis, we find a
negative association between insiders future trading profits and news coverage of regulatory releases
of insiders prior trading activities.
Next, we find evidence suggesting that our disciplining via dissemination hypothesis operates
through three economic channels, namely, information asymmetry, litigation risk, and capital-at-risk.
First, we show that news coverage is more effective in attenuating insiders profits in firms with
higher analyst forecast dispersion and in firms that are not audited by Big N auditors. Second, we
adopt Kim and Skinners [2012] litigation risk measure and find that the effect of news coverage is
more pronounced in firms that face higher litigation risk. Finally, we find that the relation between
insiders trading profits and news coverage is magnified when insiders personal capital, proxied by
Following the literature, we examine legal insider trading that is disclosed to the SEC. We discuss the regulation of
insider trades in detail in the Internet Appendix. The profitability of insider trades is one of the most tangible signs of poor
corporate governance (Aboody and Lev [2000], Frankel and Li [2004], Piotroski and Roulstone [2005], Huddart, Ke, and
Shi [2007], Ravina and Sapienza [2010], Jagolinzer, Larcker, and Taylor [2011], Cohen, Malloy, and Pomorski [2012],
Dai et al. [2014], Cziraki, De Goeij, and Renneboog [2014]).
2
executives equity-based compensation and the strength of firms corporate social responsibility (Gao,
Lisic, and Zhang [2014]), is more closely tied to firms.
In additional tests, we provide evidence in support of our main findings. First, the negative
effect of news coverage on insiders trading profits operates by reducing trading profits instead of
increasing trading losses. Second, to rule out the possibility that the media engages in information
creation rather than mere news dissemination, we show that initial news coverage alone does not have
a significant influence on insiders trading profits. Third, news coverage significantly reduces the
incidence of abnormal insider trading activities and transactions executed around earnings
announcements. Finally, using both instrumental variable (IV) and natural experiment approaches,
our endogeneity analysis suggests that news dissemination has a causal disciplining effect on insiders
trading profits.
Our paper is among the first to link news coverage with insider trading. Frankel and Li [2004]
examine the effect of financial statement informativeness, analyst following, and company news on
insiders profits and frequency. The authors find a nonsignificant relation between a firms general
news coverage (e.g., new product announcements), which is used as an alternative variable to proxy
for a firms information environment, and its insiders profits. Instead of focusing on firms general
news coverage, our study examines the dissemination effect of insider trading-related news and shows
that such news coverage does play a governance role in restraining insiders trading profits.
More broadly, our paper contributes to the literature on the corporate governance role of the
media. The literature recognizes the role of the news media in bringing governance issues to the
attention of shareholders (Tetlock [2007], Dyck, Volchkova, and Zingales [2008]), board members
(Joe, Louis, and Robinson, [2009]), corporate policymakers (Dyck, Morse, and Zingales [2010]), and
the general public (Miller [2006], Dyck, Volchkova, and Zingales [2008]). Examples of documented
ways in which the media exerts a governance role include reversing governance violations (Dyck,
Volchkova, and Zingales [2008]), pressuring managers to act in ways that are socially acceptable
(Dyck and Zingales [2002]), and monitoring executive compensation (Core, Guay, and Larcker
[2008], Kuhnen and Niessen [2012]). By isolating the news dissemination role of the media from its
information creation role, we show that news dissemination by itself can discipline managers.
In the remainder of this article, Section 2 develops our hypotheses. In Section 3, we describe
our data and methodology. We present the main findings in Section 4 and the results of additional
tests in Section 5. Section 6 concludes the paper.
2. Hypothesis Development
The media rebroadcasts or disseminates regulatory information to the market more broadly
compared with the regular regulatory filing process. Previous studies highlight the information
dissemination effect of media coverage with regard to, for instance, SEC insider trading filings
(Chang and Suk [1998]), general regulatory filings (10-K and 10-Q) (Li, Ramesh, and Shen [2011]),
corporate governance violations (Dyck, Volchkova, and Zingales [2008]), and accounting fraud
(Miller [2006]). Moreover, the dissemination of regulatory information by the media can enhance the
breadth of coverage and can attract the attention of investors through repetition (Fang and Peress
[2009], Bushee et al. [2010], Blankespoor, Miller, and White [2014], Peress [2014]). Therefore, we
propose the disciplining via dissemination view as our main hypothesis:
H1: Media coverage of insider trades reduces the profitability of insiders future
transactions.
The counterfactual to our main hypothesis is that news dissemination by the media does not
affect insiders trading profits. If the media is only effective through news creation and analysis, then
news coverage that simply repeats regulatory releases of insider trading activities without adding new
information should not improve firms information environment or actively influence insiders trading
activities. Therefore, either insiders information advantage relative to the public is not affected, or
insiders do not view the media as a potential threat to their careers or personal wealth. We consider
this view to be the null hypothesis.
4
We further identify three economic mechanisms underlying the disciplining effect of news
dissemination (H1), which are framed as additional hypotheses. First, stock prices adjust more
quickly to firms fundamentals through the dissemination of news coverage on prior insider trades,
and thus, the profitability of future trades is reduced for firms with lower information transparency.
Therefore, the hypothesis regarding this information asymmetry channel is formally stated as follows:
H2a: The disciplining effect of media coverage is stronger in firms with greater information
asymmetry.
Our second and third mechanisms identify credible circumstances under which informed
insiders should be concerned about past and future news coverage of their trades and thus alter their
future actions. With regard to litigation risk, Billings and Cedergren [2014] find evidence that
litigation risk increases with insiders opportunistic selling. According to some accounts, 40-60% of
all securities class actions involve allegations of insider trading (Erickson [2010]). Because of
concerns regarding litigation risk, insiders significantly reduce their selling activities after securities
class actions (Cheng, Huang, and Li [2013]) or avoid profitable transactions immediately prior to
earnings announcements (Huddart, Ke, and Shi [2007]). Based on these findings, we hypothesize that
after the financial press disseminates news of insiders trading activities, insiders fear of being sued
affects the extent to which they seek to profit from subsequent trades by relying on non-public
material information.4 We expect this effect to be particularly evident in firms with high litigation risk.
We propose our hypothesis on the litigation risk channel as follows:
H2b: The disciplining effect of media coverage is stronger in firms with higher litigation
risk.
For the third mechanism, we argue that the dissemination of insider trading news can
adversely affect insiders personal wealth and reputational capital. First, the intensive public exposure
of opportunistic trading by insiders can raise investors concerns regarding a firms internal
Ex ante, opportunistic insider trades do not necessarily trigger SEC enforcement. However, securities class actions
can be filed or can be augmented even if the insider trades are legal. The probability of enforcement and the resulting
punishment, as perceived by insiders, can be influenced by the media (Dyck, Volchkova, and Zingales [2008]).
5
governance and growth prospects and can hence reduce insiders personal wealth that is tied to firm
value. Moreover, the ability to exercise discretionary trading in the future is valuable to insiders
because it allows them to sell for consumption (Roulstone [2003]) and rebalance their personal
portfolio for hedging purposes (Ofek and Yermack [2000]). Hence, we expect insiders with large
equity holdings to be concerned about further restrictions on their ability to trade profitably, when
their prior opportunistic trades capture media attention.
Second, prior studies suggest that reputation plays a role in disciplining managers and
directors (Dyck and Zingales [2002], Dyck, Volchkova, and Zingales [2008]). Liu and McConnell
[2013] further argue that the media, through effects from both news dissemination and news creation,
affects managerial reputational capital. Therefore, managers should be concerned about the possibility
that extensive news coverage of insider trading leads to the destruction of their reputational capital.
This reasoning is similar in spirit to that of Gao, Lisic, and Zhang [2014], who attribute their finding
of a negative association between corporate social responsibility and insiders trading profits to
circumstances when executives personal interests are more closely aligned with the interests of the
firm. Taken together, these arguments suggest that the medias disciplining effect is stronger when
executives personal wealth and reputational capital are more closely tied to their firm. We propose
our hypothesis regarding this capital-at-risk channel as follows:
H2c: The disciplining effect of media coverage is stronger in firms in which executives have
greater concerns about the impact of media coverage on their personal capital.
These channels are not mutually exclusive. For instance, the last two channels may be jointly
related to a naming and shaming effect on insiders. Finding evidence in support of these channels
would strengthen the plausibility of our baseline results.
3. Research Design
3.1. DATA
Our insider trading sample is based on the Thomson Financial Insiders Data Feed sourced
from Form 4 filings over the period from January 2001 to June 2012, and it includes corporate
insiders transactions involving stocks that are listed on the NYSE, AMEX, or NASDAQ.5 The data
contain trade information on directors, officers, and large stockholders with holdings greater than 10%
of a firms stock, all of which are subject to SEC disclosure requirements.
We exclude transactions by large shareholders and retain only those attributed to officers and
directors. We focus on valid purchases and sales of common stocks (share codes 10 and 11) in the
Center for Research in Security Prices (CRSP) database.6 We further exclude (1) transactions with
fewer than 100 shares or those with trading prices of less than $2, (2) transactions with traded prices
outside the range between the daily low and high prices reported in the CRSP database, (3)
transactions with the number of shares exceeding the total number of outstanding shares in the CRSP
database, (4) transactions with the number of shares traded exceeding the total daily trading volume in
the CRSP database, and (5) firms in the financial or utilities industries (firms with SIC codes between
6000 and 6999 or between 4900 and 4999). Finally, we exclude trades of stocks that have never been
covered by RavenPack. These restrictions result in a final sample of more than 1.375 million insider
transactions.
Following Jagolinzer, Larcker, and Taylor [2011], we define the profitability of insider trading
as the annualized abnormal return from the Carhart [1997] four-factor model estimated over the 180
calendar days after the transaction date (Alpha) based on the CRSP database.7 From the insider trade
Our insider trading sample ends in June 2012 because our stock return data in the CRSP database are available up to
December 2012, whereas we estimate the trade profits for six months after the transaction.
6
A valid transaction is one without a Cleanse Indicator code of A or S in the Thomson Financial Insiders Data
Feed. The Cleanse Indicator denotes Thomson Reuters' level of confidence concerning the accuracy of a particular record.
We disregard observations that are coded A (indicating that an attempt was made at cleansing but that the data still
appear unreasonable or inconsistent) and S (where the security does not meet the collection requirements).
7
The 180-day window is particularly appropriate for accommodating the six-month legal prohibition on short-swing
profits: the SEC requires company insiders to return any profits made from the purchase and sale of company stock if
both transactions occur within a six-month period. For sales, Alpha is multiplied by -1.
7
data, we also construct other variables, including trade size (Trade Size), with each transaction size in
dollars, and trade frequency (Trade Frequency), which is the number of insider transactions of a firm
in the 360-day period preceding each insider trade.
We obtain data on the news coverage of NYSE, AMEX, or NASQAQ stocks during the period
from January 2000 to June 2012 from RavenPack, a leading global news database that includes Dow
Jones Newswire alerts on stories that are published about companies.8 News articles are categorized
by using RavenPacks proprietary text and part-of-speech tagging or labeling. The news category
regarding insider trading contains three subcategories: regulatory releases of insiders transactions
(98.15%), insiders share gifts (1.84%), and insiders lawsuits (0.01%). Given that we explore the
news dissemination effect of the media, we drop news events concerning insiders share gifts and
lawsuits and concentrate on regulatory releases of insiders transactions.
To measure the informational content of a news story, RavenPack generates company
relevance and novelty scores ranging from zero to 100. The company relevance scores allow us to
extract and compute aggregate counts of news articles that are related to specific firms in the database.
We only include news articles with relevance scores equal to 100, which indicates that a firm is
quoted as the main subject of a news release. In an additional analysis, we utilize the novelty scores to
identify the originality of the news releases. Further, and importantly for our empirical design, based
on the news category information, we count news articles that are specific to regulatory releases of
prior insider trading activities (News CoverageIT) separately from a firms general news coverage
(News CoverageNon-IT) over the 360 calendar days, scaled by 100.
Finally, we obtain accounting information from Compustat, analyst forecast data from I/B/E/S,
and audit data from Audit Analytics. Managerial compensation information is taken from BoardEx,
and measures of adherence to corporate social responsibility are estimated based on KLD Research
RavenPack has been increasingly used in the literature (e.g., Kolasinski, Reed, and Ringgenberg [2013], Dang,
Moshirian, and Zhang [2014], Shroff, Verdi, and Yu [2014]).
8
and Analytics. Our final sample includes over 4,600 stocks. As the data requirements differ across our
tests, the sample size varies depending on data availability.
relation between News CoverageIT and firm performance (Return and ROA) indicates that the media
concentrates on insiders in well-performing firms, where insiders trading behavior is likely to be of
greater interest to the market. By contrast, News CoverageNon-IT is negatively associated with firm
performance, which is consistent with the literature documenting that poor performing firms attract
media attention on issues such as excessive executive compensation (Core, Guay, and Larcker [2008]).
(1)
Our main variable of interest is News CoverageIT, denoting the number of news releases
related to prior insider trading activities. 11 We expect the coefficient on News CoverageIT to be
significantly negative (H1).
Following Lakonishok and Lee [2001], we include Size and MB to control for size and bookto-market effects, respectively (Fama and French [1997]). In addition, following Brochet [2010], we
control for momentum (Return), stock return volatility (STD), R&D, Trade Size, and Trade Frequency.
Return controls for insiders potentially contrarian behavior. STD and R&D are included because
insider trades are likely to be more informative in firms with higher information risk and higher R&D
intensity (Aboody and Lev [2000]). Trade Size controls for the possible link between the importance
10
See, for example, Core, Guay and Larcker [2008] and Kuhnen and Niessen [2012] on executive compensation and
Braggion and Giannetti [2013] on limited voting shares.
11
In this main test, we also control for News CoverageNon-IT, denoting the number of new articles that are not related to
insider trading. Since the coefficient on this variable is not statistically significant, in subsequent analyses, we do not
control for this variable. However, our results in subsequent analyses are robust to the inclusion of News CoverageNon-IT.
11
of private information and the number of shares traded. Trade Frequency controls for either preemptions of the information content of insider trades or the reinforcement of prior signals from such
trades. We also add industry and year fixed effects to control for cross-sectional and time-series
dependence, and we cluster standard errors at the firm level.12
Furthermore, we examine the economic channels through which news dissemination can have
a disciplining effect on insiders (H2) by using the following model:
(2)
Each Channel Factor is a proxy that is variously related to information asymmetry (analyst
forecast dispersion and a Big N auditor indicator), litigation risk (Kim and Skinners [2012] litigation
risk measure), or managers personal wealth and reputation tied to the firm (managers equity-based
compensation and the strength of a firms adherence to corporate social responsibility). The
coefficient on the interaction term between News Coverage IT and each Channel Factor is expected to
be significantly negative (except for the Big N auditor indicator, which is an inverse measure of
information asymmetry), implying that the medias governance effect operates through these three
economic mechanisms (H2).
4. Results
4.1. BASELINE FINDINGS
The evidence presented in Table 4, based on regression equation (1), shows that previous
news coverage regarding firm insiders trading activities generally reduces insiders subsequent
trading profits. In Model (1), the coefficient on News CoverageIT is significantly negative (t-stat = 3.89). Thus, future insiders profits are largely attenuated when the press reports prior transactions of
12
We control for industry fixed effects because Solomon and Soltes [2012] show that industry-specific factors explain
a significant proportion of the variation in firm media coverage.
12
insiders, which is consistent with H1, our disciplining via dissemination hypothesis. Economically,
the magnitude of this effect is also sizeable. A one-standard-deviation increase in news coverage
related to insider trading is associated with a 3.4% decrease in the abnormal stock returns of insider
trades. However, in Model (2), the coefficient on News CoverageNon-IT is positive but nonsignificant
(t-stat = 0.68). This coefficient remains nonsignificant in Model (3), where we include both News
CoverageIT and News CoverageNon-IT, suggesting that a firms general news coverage does not
necessarily restrict insider trading activities (Frankel and Li [2004]).
In Models (4) to (7), we subject our baseline findings to various robustness tests. Model (4)
addresses the impact of overlapping and confounding events around insider trades. Specifically, we
estimate our model based on a subsample of insider trades with no earnings announcements in a tenday window [-10, 10] before and after the transaction dates of insider trades. In Model (5), we use the
change in news coverage from year t-1 to year t as the focal explanatory variable, and we control for
firm fixed effects.13 In Models (6) and (7), we aim to trace the media effect to insiders who can be
specifically named in the news. We regress Alpha of insider i in firm j on news coverage related to a
specific insiders transactions (News CoverageSIT) and on news coverage that is not related to a
specific insiders transactions (News CoverageNon-SIT). News CoverageSIT is the number of firm js
news releases related to insider trading in a window [0, 1] around the SEC filing date of insider i,
while News CoverageNon-SIT is equal to the total number of firm js news articles minus news coverage
related to a specific insiders transactions. Both the variables are scaled by 100. The results indicate
that news coverage of prior trades made by a specific insider effectively restricts his/her trading
profits in the future.
We find that our results are also robust to: (1) using alternative measures of insiders' trading
profits, clustering methods, and sample selections; (2) controlling for stale news coverage estimated in
13
Prior studies on the media (e.g., Core, Guay, and Larcker [2008], Kothari, Li, and Short [2009], Bushee et al. [2010],
Liu and McConnell [2013]) do not control for firm fixed effects, possibly owing to the sticky effect of news coverage. We
replace the level of news coverage with the change in news coverage to increase the time-series variation of the news
coverage measure and then control for firm fixed effects. If news coverage data are not available in year t-1, we use the
average news coverage among all the firms in year t as the benchmark to calculate the change in news coverage.
13
year t-2 and t-3; (3) focusing on news coverage with high value impact on the market, and news
articles published within 180 days before an insider transaction; and (4) adding other governance
factors, including institutional ownership, analyst coverage, Big N auditor, board independence, and
CEO-Chairman duality into our analysis.14
14
The results of robustness tests are reported in the Internet Appendix Tables IA1-IA4.
Behn, Choi, and Kang [2008] demonstrate that firms that are audited by the Big N audit firms report more credible
and accurate financial statements.
15
14
support the conjecture that the reduction of information asymmetry is an underlying mechanism
through which news coverage affects insiders trading profits.
4.2.2. The litigation risk channel. With respect to this channel, we hypothesize that after the
financial press disseminates trading news on company executives, insiders fear of being sued or
being charged by the SEC if they profit from any non-public material information affects the extent to
which they seek to profit from subsequent trades. Therefore, news coverage is more likely to affect
insiders in firms with higher potential for litigation (H2b).
We adopt Kim and Skinners [2012] litigation risk (LR) measure, which is calculated prior to
the insider trading dates. The Kim and Skinner [2012] measure is particularly appropriate for our
analysis because it measures the predictability of securities class actions against firms based on their
corporate attributes. Securities class actions often challenge insider trading by individuals associated
with such public companies. We report the results in Panel B of Table 5. Consistent with H2b, in
Model (1), we show that the media is more effective in reducing the profits of insiders in firms that
face higher litigation risk. This finding implies that because of their fear of litigation, insiders pursuit
of profits from trading is restricted when the public and media pay more attention to likely
opportunistic insiders.
As an alternative test for the litigation risk channel, we investigate whether the news coverage
effect is influenced by the direction of trades and the type of insiders. Since courts look for evidence
that insiders engaged in selling rather than buying to establish that the defendants acted with scienter,
or intent, in allegations of insider trading and securities fraud,16 we expect the media effect to be
stronger for insider sales. In addition, we conjecture that the medias disciplining effect is stronger on
officers than on the board of directors, as officers have more to lose from litigation. For Models (2) to
16
See, for example, Sale [2002], Cheng and Lo [2006], Rogers [2008], Brochet [2010], Rogers, Van Buskirk, and
Zechman [2011], Billings and Cedergren [2014], and Brochet and Srinivasan [2014].
15
(7), the results confirm that the attenuating effect of news on insiders profits pertains to insider sales
and officers trades.17
As a by-product, our results suggest that the media, through the litigation risk channel, is at
least partially responsible for the common empirical finding that, on average, insiders profit from
purchases (which we find are apparently immune to the news coverage effect) but not from sales (see,
e.g., Aboody and Lev [2000], Huddart and Ke [2007], Jagolinzer, Larcker, and Taylor [2011],
Davidson, Dey, and Smith [2013]).
4.2.3. The capital-at-risk channel. We hypothesize that the association between news coverage
and insiders profits is stronger when insiders are more concerned about the future value of their
equity holdings, as well as their reputational capital that is tied to the firm (H2c).
To test this hypothesis, we use variables for executives equity compensation, CEO and
Executives Equity Compensation, measured by the log value of the amount of total equity
remuneration, as proxies for insiders concerns regarding their personal wealth (Roulstone [2003],
Ofek and Yermack [2000]). Our corporate social responsibility variables (Corporate Social
Responsibility) represent insiders concerns regarding their reputational capital (Gao, Lisic, and
Zhang [2014]).
Consistent with the personal wealth argument in H2c, the results reported in Panel C of Table
5 indicate that equity-based incentives increase the impact of news coverage on insiders trading
profits. In Models (1) and (2), the coefficients of the interaction terms between news coverage and
managerial equity compensation are negative and significant. In addition, consistent with the
managerial reputation argument in H2c, Models (3) and (4) indicate that the disciplining effect of the
media is stronger when managers are more concerned about their reputational capital.
17
In another test for the litigation channel presented in Internet Appendix Table IA5, we run logistic models for firmyear panel data by regressing SEC investigation and/or lawsuit indicators in year t on the interaction terms between
average firm-level Alpha and News CoverageIT in year t-1. The results indicate that firms with more profitable insider
trades and greater news coverage regarding these transactions are more likely to be investigated by SEC and to be sued
owing to legal, insider trading, and/or fraud issues. These indicator variables are constructed based on relevant regulatory
and litigation news releases from RavenPack.
16
5. Additional Tests
To corroborate our main results, we perform additional analyses on the governance effect of
news coverage. Specifically, we (1) examine whether news coverage increases insiders trading losses
instead of reducing their trading profits, (2) perform stress tests on the disciplining effect of news
dissemination, (3) check whether news coverage also affects insiders trading intensity, and (4)
attempt to establish a causal relation between news dissemination and insiders trading profits.
18
See Williams [2012] for a detailed explanation of the marginal effects at the means (MEM) analysis, which captures
the predicted value of the dependent variable in response to a particular explanatory variable, when the effects of all other
explanatory variables in the model are held constant at the sample means.
17
[2006]). To further ascertain whether the disciplining effect of news coverage operates through
information dissemination, we first regress Alpha on Adjusted News CoverageIT, which is determined
by News CoverageIT minus the number of Form 4 filings (Filing Frequency). We present the results in
Table 7. We find that both Adjusted News CoverageIT and Filing Frequency reduce the profitability of
insider trades in Models (1) and (2). However, in Model (3), when both variables are included in the
same regression, only Adjusted News CoverageIT retains statistical significance, implying that the
news disciplining effect operates through the dissemination of regulatory filings.
Further, since RavenPack provides additional information for news originality in novelty
scores, we can observe whether a story is a breaking news item (First News Coverage) or a repeated
news article (Repeated News CoverageIT). From Models (4) to (6), we find that only Repeated News
CoverageIT has an effect in attenuating insiders profits.19
Based on the findings in this subsection, we conclude that news coverage reduces insiders
trading profits through the medias news dissemination role rather than through its information
creation role, adding value by broadening the dissemination of insider trading regulatory filings
through replication.
19
In Model (6), the positive sign on First News CoverageIT results from multicollinearity with Repeated News
CoverageIT (the correlation coefficient is 0.88). In untabulated results, we regress Repeated News CoverageIT on First
News CoverageIT to acquire a measure of Residual Repeated CoverageIT. In a model including both First News CoverageIT
and Residual Repeated CoverageIT, we find that the former has a nonsignificant sign and that the coefficient on the latter is
significantly negative.
18
CoverageIT in year t-1. The coefficients on News CoverageIT are negative, implying that media
coverage deters insiders trading activities.
Second, in Models (3) and (4), we focus on the effect of media coverage on the timing of
insiders trading relative to earnings announcements (Huddart, Ke, and Shi [2007], Billings and
Cedergren [2014]). In this way, we can investigate how media coverage affects the timing of the
trades around the release of stock price sensitive information. We construct two firm-level proxies for
the timing of insiders trades, Trade VolumeRatio and Trade SizeRatio. Trade VolumeRatio (Trade SizeRatio)
denotes insiders transaction volume in shares (transaction size in dollars) during the eleven-day
window [-5, 5] around quarterly (annual) earnings announcements, scaled by total transaction volume
(transaction size) in a year. The coefficients on News CoverageIT are negative in both models,
indicating that the execution of insider trades is attenuated around earnings announcements.
These findings together with our main results suggest that the media not only reduces the
profitability of completed trades but also attenuates the execution of transactions.
20
For example, Miller [2006] finds that the media is more likely to report on accounting fraud in firms with a larger
public following. In Table 3, we also find that larger firms have more coverage on insider trading news.
19
Moreover, a long distance between news outlets and a firm can reduce the medias attention and
interest in following the firm and the activities of the firms insiders. However, no economic intuition
directly links this instrument to insiders profit potential, which, to an extent, supports the exclusion
restriction on the instrument.
We apply three variations of distance as IVs for robustness: (1) geographic proximity rank
(ProximityRank) is the descending decile rank of the minimum distance between the headquarters of a
firm and Dow Jones' offices; (2) geographic proximity dummy (ProximityDummy) is a dummy variable
that equals one if the minimum distance between the headquarters and Dow Jones' offices is within
the 40th percentile of minimum distance values for all firms in the sample and zero otherwise; and (3)
geographic proximity category (ProximityCategory) is a categorical variable that takes the value of two
(one) if the headquarters is located in the same city (state) as one of Dow Jones' offices and zero
otherwise.
We present the results in Panel A of Table 9. In the first stage, all three of the measures for
geographic proximity are positively associated with news coverage of insider trading. In the second
stage, the predicted news coverage (Predicted News CoverageIT) is a significant attenuator of insider
profits. Following Larcker and Rusticus [2010], we reject the null hypothesis of weak instrument with
Kleibergen-Paap Wald F statistics equal to 6.40, 10.07, and 5.52 for the three respective IVs (Stock
and Yogo [2005]).21 In untabulated tests, we find that distance to a media source has virtually no
impact on insiders profits for firms with low news coverage (News CoverageIT = 0 or News
CoverageIT 0.01 at the 10th percentile), which further mitigates concerns regarding the violation of
the exclusion restriction on our IVs. That is, proximity cannot directly affect trade profits without the
existence of news coverage.22
21
We find p-values equal to 0.011, 0.002, and 0.019, and the partial R2 equal to 1.25%, 2.32%, and 1.48% for the
three IVs, respectively, which are significantly different from zero.
22
We repeat the analysis by using the educated population in the state where firm is located as an alternative IV based
on data from the U.S. Census Bureau. Our findings remain unaffected. The results even hold when we include both the
proximity and the education IVs in the first stage.
20
5.4.2. Natural Experiment. Dow Jones formerly created a product focused on covering news
about insider trading filings in January 2004. Since Dow Jones is one of the dominant players in the
financial news business, its formal initiation of coverage on Form 4 filings should represent an
exogenously positive shift in the news coverage of insider trading.
In untabulated analysis, we indeed find an increase in the cross-sectional variation of news
coverage on insider trading after Dow Joness formal initiation of such coverage: the standard
deviation of News CoverageIT increases from 0.01 (in 2003) to 0.34 (in 2004). Therefore, we expect
to find a stronger media effect in the post-initiation period. This natural experiment provides us an
opportunity to identify a causal mechanism between News CoverageIT and Alpha and further address
any endogeneity concerns.
We report supportive evidence of this conjecture in Panel B of Table 9. In Models (1) to (6),
we regress insider trading profit on our IVs in the subperiods before and after Dow Joness formal
initiation of news coverage on insider trading. The results show that before the exogenous shock
(2001-2003), the impact of geographic distance on trading profits is nonsignificant and that it
becomes significantly negative in the post-initiation period (2004-2007) for the second and third IVs.
These results support the validity of our natural experiment, showing that the indirect impact of
proximity on trading profits is mediated through news coverage after the exogenous shock. In Models
(7) and (8), we perform our analysis over two subperiods. Prior to the exogenous shift in coverage,
the relation between News CoverageIT and Alpha is nonsignificant; however, it becomes significant
and negative after the initiation of news coverage. Our findings further hold when we drop the year
2004 from the post-initiation period.
If Dow Jones endogenously covers well-governed firms after the initiation of news coverage
on insider trading, we should not obtain significant coefficients for the predicted news coverage when
we use geographic proximity in our IV analysis. Therefore, taking the results for the IV analysis and
21
the natural experiment together, we find evidence supporting a causal effect of news coverage on
insiders trading profits.
6. Conclusion
The governance role of the media has received considerable attention from academic
researchers. However, given that the media has a dual role in reporting corporate business events,
undertaking original investigations and/or rebroadcasting existing news, whether one or both of these
functions play a disciplining role in protecting investors interests remains unclear in the literature.
In this paper, we use the insider trading context as a laboratory to explore the channels through
which news coverage facilitates corporate governance outcomes by regurgitating regulatory
information. We find that the media restricts insiders future trading profits by disseminating
regulatory releases of prior insider trading activities. Further evidence indicates that news coverage
has a disciplining effect on insider trading via three economic mechanisms: news dissemination
reduces information asymmetry between insiders and investors; insiders with litigation risk concerns
react to the media spotlight; and insiders with personal wealth and reputational capital that are tied to
their firm are susceptible to the medias disciplining effect.
Following prior studies such as Bushee et al. [2010], we believe that our work provides
avenues for further research on the real effects of news dissemination. In particular, one promising
direction for future research is to further investigate the economic implications of this effect on other
important corporate decisions that are made by insiders and managers, as well as the associated
outcomes. Such work would contribute to the question of the fundamental benefits of the medias role
in financial markets.
22
APPENDIX
Variable Definitions
Variable
Acronym
Definition
Data Source
Alpha
Annualized abnormal daily return based on the four-factor model in a window [1,180] following the transaction.
Trade Size
Thomson Reuters
Trade frequency
Trade Frequency
Log of number of insider trades in a window [-360, -1] prior to the transaction.
Thomson Reuters
Filing frequency
Filing Frequency
Number of form filings to SEC in a window [-360, -1] prior to the transaction.
Thomson Reuters
Trade VolumeAbn
Log transformation of abnormal transaction volume of shares in a year adjusted by transaction volume in last year.
Thomson Reuters
Trade SizeAbn
Log transformation of abnormal transaction size in dollars in a year adjusted by transaction size in last year.
Thomson Reuters
Trade VolumeRatio
Insiders' transaction volume during eleven-day earnings announcement window [-5, 5] scaled by total volume in a year.
Thomson Reuters
Trade SizeRatio
Insiders' transaction size during eleven-day earnings announcement window [-5, 5] scaled by total transaction size in a year.
Thomson Reuters
News CoverageIT
Number of news releases related to insider trading in a window [-360, -1] prior to the transaction scaled by 100.
RavenPack
News CoverageNon-IT
Number of news releases not related to insider trading in a window [-360, -1] prior to the transaction scaled by 100.
RavenPack
News CoverageSIT
Number of news releases related to a specific insider's transactions based on a window [0,1] around the SEC filing date
RavenPack
B. Media Variables
transactions
News coverage not related to a specific insiders
transactions
Number of news releases not related to a specific insider's transactions based on a window [0,1] around the SEC filing date
RavenPack
Number of news releases minus number of form filings in a window [-360, -1] prior to the transaction scaled by 100.
RavenPack
Number of repeated news releases related to insider trading in a window [-360, -1] prior to the transaction scaled by 100.
RavenPack
Number of the first news releases related to insider trading in a window [-360, -1] prior to the transaction scaled by 100.
RavenPack
23
APPENDIX
Variable Definitions
Variable
Acronym
Definition
Data Source
Size
CRSP
Market-to-book ratio
MB
Return
CRSP
STD
Standard deviation of daily stock returns in a window [-360, -1] prior to the transaction.
CRSP
R&D
Dummy variable that equals one if there are positive R&D expenses.
Compustat
Return-on-asset ratio
ROA
Compustat
S&P 500
Dummy variable that equals one if a firm is included in the S&P 500 index.
Compustat
ProximityRank
Descending decile rank of the minimum distance between the firms headquarter and Dow Jones' offices.
ProximityDummy
Dummy variable that equals one if the minimum distance between the headquarter of a firm and Dow Jones' offices
D. Other Variables
is lower than the 40th percentile of minimum distance values for firms in the sample.
Geographic proximity category
ProximityCategory
Categorical variable that equals two (one) if the headquarter of a firm is located in the same city (state) as one of Dow
Big N auditors
Big N
Dummy variable that equals one if the firm is audited by a Big N auditor.
Audit Analytics
IBES
Litigation risk
LR
BoardEx
BoardEx
Dummy variable that equals one if the number of CSR strengths is greater than that of CSR concerns; CSR is based
on six areas: community, diversity, employee relations, environment, human rights, and product.
Analytics
Dummy variable that equals one if the number of CSR strengths is greater than that of CSR concerns; CSR is based
on seven areas: corporate governance, community, diversity, employee relations, environment, human rights, and product.
Analytics
24
REFERENCES
Aboody, D., and B. Lev. Information Asymmetry, R&D and Insider Gains. Journal of Finance 55
(2000): 27472766.
Behn, B.; J. H. Choi; and T. Kang. Audit Quality and Properties of Analyst Earnings Forecasts. The
Accounting Review 83 (2008): 327359.
Billings, M. B., and M. C. Cedergren. Strategic Silence, Insider Selling and Litigation Risk.
Working paper, New York University, 2014.
Blankespoor, E.; G.S. Miller; and H.D. White. The Role of Dissemination in Market Liquidity:
Evidence From Firms Use of Twitter. The Accounting Review 89 (2014): 79112.
Braggion, F., and M. Giannetti. Public Debate and Stock Prices: Evidence from the Voting Premium.
Working paper, Tilburg University, 2013.
Brochet, F. Information Content of Insider Trades Before and After the Sarbanes-Oxley Act. The
Accounting Review 85 (2010): 419446.
Brochet, F., and S. Srinivasan. Accountability of Independent Directors: Evidence from Firms
Subject to Securities Litigation. Journal of Financial Economics 111 (2014): 430449.
Bushee, B. J.; J. E. Core; W. R. Guay; and S. J. W. Hamm. The Role of the Business Press as an
Information Intermediary. Journal of Accounting Research 48 (2010): 119.
Carhart, M. On Persistency In Mutual Fund Performance. Journal of Finance 52 (1997): 5782.
Chang, S., and D. Suk. Stock Prices and Secondary Dissemination of Information: The Wall Street
Journals Insider Spotlight Column. Financial Review 33 (1998): 115128.
Cheng, C. S. A.; H. H. Huang; and Y. Li. Does Shareholder Litigation Deter Insider Trading?
Working paper, Hong Kong Polytechnic University, 2013.
Cheng, Q., and K. Lo. Insider Trading and Voluntary Disclosure. Journal of Accounting
Research 44 (2006): 815848.
Cohen, L.; C. J. Malloy; and L. Pomorski. Decoding Inside Information. Journal of Finance 67
(2012): 10091043.
Core, J. E.; W. Guay; and D. F. Larcker. The Power of the Pen and Executive Compensation.
Journal of Financial Economics 88 (2008): 125.
Cziraki, P.; P. De Goeij; and L. Renneboog. Corporate Governance Rules and Insider Trading
Profits. Review of Finance 18 (2014): 67108.
Dai, L.; R. Fu; J-K. Kang; and I. Lee. Corporate Governance and Insider Trading. Working paper,
Australian National University, 2014.
Davidson, R; A. Dey; and A. J. Smith. Executives Legal Records, Lavish Lifestyles and Insider
Trading Activities. Working paper, Georgetown University, 2013.
Dang, L. T.; F. Moshirian; and B. Zhang. Commonality in News Around the World. Journal of
Financial Economics (2014), forthcoming.
25
Dyck, A.; A. Morse; and L. Zingales. Who Blows the Whistle on Corporate Fraud? Journal of
Finance 65 (2010): 22132253.
Dyck, A.; N. Volchkova; and L. Zingales. The Corporate Governance Role of the Media: Evidence
from Russia. Journal of Finance 63 (2008): 10931135.
Dyck, A., and L. Zingales. The Corporate Governance Role of the Media. Working paper, NBER,
2002.
Erickson, J. Corporate Governance in the Courtroom: An Empirical Analysis. William and Mary
Law Review 51 (2010): 17491831.
Fama, E., and K. French. Industry Costs of Equity. Journal of Financial Economics 43 (1997):
153193.
Fang, L., and J. Peress. Media Coverage and the Cross-Section of Stock Returns. Journal of
Finance 64 (2009): 20232052.
Frankel, R., and X. Li. Characteristics of a Firms Information Environment and the Information
Asymmetry Between Insiders and Outsiders. Journal of Accounting and Economics 37(2004):
229259.
Gao, F; L. L. Lisic; and I. X. Zhang. Commitment to Social Good and Insider Trading. Journal of
Accounting and Economics (2014): Forthcoming.
Gurun, U. G., and A. W. Butler. Don't Believe the Hype: Local Media Slant, Local Advertising, and
Firm Value. Journal of Finance 67 (2012): 561598.
Huddart, S., and B. Ke. Information Asymmetry and Cross-sectional Variation in Insider Trading.
Contemporary Accounting Research 24 (2007): 195232.
Huddart, S.; B. Ke; and C. Shi. Jeopardy, Non-Public Information, and Insider Trading Around SEC
10-K and 10-Q Filings. Journal of Accounting and Economics 43(2007): 336.
Jagolinzer, A.; D. Larcker; and D. Taylor. Corporate Governance and the Information Content of
Insider Trades. Journal of Accounting Research 49 (2011): 12491274.
Joe, J. R.; H. Louis; and D. Robinson, Managers and Investors Responses to Media Exposure of
Board Ineffectiveness. Journal of Financial and Quantitative Analysis 44 (2009): 579605.
Kim, I., and D. J., Skinner. Measuring Securities Litigation Risk. Journal of Accounting and
Economics 53 (2012): 290310.
Kolasinski, A. C.; A. V. Reed; and M. C. Ringgenberg. A Multiple Lender Approach to
Understanding Supply and Search in the Equity Lending Market. Journal of Finance 68
(2013): 559595.
Kothari, S. P.; X. Li; and J. E. Short. The Effect of Disclosures by Management, Analysts, and
Business Press on Cost of Capital, Return Volatility, and Analyst Forecasts: A Study Using
Content Analysis. The Accounting Review 84 (2009): 16391670.
Kuhnen, C. M., and A. Niessen. Public Opinion and Executive Compensation. Management
Science 58 (2012): 12491272.
26
Lakonishok, J., and I. Lee. Are Insider Trades Informative? Review of Financial Studies 14 (2001):
79111.
Larcker, D. F., and T. O. Rusticus. On the Use of Instrumental Variables in Accounting Research.
Journal of Accounting and Economics 49 (2010): 186205.
Li, E. X.; K. Ramesh; and M. Shen. The Role of Newswires in Screening and Disseminating ValueRelevant Information in Periodic SEC Reports. The Accounting Review 86 (2011): 669701.
Liu, B., and J. J. McConnell. The Role of the Media in Corporate Governance: Do the Media
Influence Managers Capital Allocation Decisions? Journal of Financial Economics 110
(2013): 117.
Miller, G. S. The Press as a Watchdog for Accounting Fraud. Journal of Accounting Research 44
(2006): 10011033.
Ofek, E., and D. Yermack. Taking Stock: Equity-Based Compensation and the Evolution of
Managerial Ownership. Journal of Finance 55 (2000): 13671384.
Piotroski, J. D., and D. T. Roulstone. Do Insider Trades Reflect Both Contrarian Beliefs and
Superior Knowledge About Future Cash Flow Realizations? Journal of Accounting and
Economics 39 (2005): 5581.
Peress, J. The Media and the Diffusion of Information in Financial Markets: Evidence from
Newspaper Strikes. Journal of Finance (2014): Forthcoming.
Ravina, E., and P. Sapienza. What Do Independent Directors Know? Evidence from Their Trading.
Review of Financial Studies 23 (2010): 9621003.
Rogers, J. L. Disclosure Quality and Management Trading Incentives. Journal of Accounting
Research 46 (2008): 12651296
Rogers, J. L.; A. Van Buskirk; and S. L. C. Zechman. Disclosure Tone and Shareholder Litigation.
The Accounting Review 86 (2011): 2155-2183.
Roulstone, D. T. The Relation Between Insider-Trading Restrictions and Executive Compensation.
Journal of Accounting Research 41 (2003): 525551.
Sale, H. Judging Heuristics. University of California at Davis Law Review 35 (2002): 903963.
Shroff, N.; R. S. Verdi; and G. Yu. Information Environment and the Investment Decisions of
Multinational Corporations. The Accounting Review 89 (2014): 759790.
Solomon, D., and E. Soltes. Managerial Control of Business Press Coverage. Working paper,
University of Southern California, 2012.
Stock, J. H., and M. Yogo. Testing for Weak Instruments in Linear IV Regression. In Identification
and Inference for Econometric Models: A Festschrift in Honor of Thomas Rothenberg, edited
by J.H. Stock, and D.W.K. Andrews. Cambridge University Press, Cambridge, 2005.
Tetlock, P. C. Giving Content to Investor Sentiment: The Role of the Media in the Stock Market.
Journal of Finance 62 (2007): 11391168.
Tetlock, P. C. Does Public Financial News Resolve Asymmetric Information? Review of Financial
Studies 23 (2010): 35203557.
27
Williams, R. Using the Margins Command to Estimate and Interpret Adjusted Predictions and
Marginal Effects. The Stata Journal 12 (2012): 308331.
Zhang, X. Information uncertainty and stock returns. Journal of Finance 61 (2006): 105135.
28
TABLE 1
Summary Statistics
Variable
Alpha
News CoverageIT
News CoverageNon-IT
Size
MB
Return
STD
R&D
Trade Size
Trade Frequency
NObs
1,376,567
1,376,567
1,376,567
1,376,567
1,376,567
1,376,567
1,376,567
1,376,567
1,376,567
1,376,567
0.544
2.031
1.425
1.846
5.505
0.640
0.013
0.494
1.806
2.054
29
-0.641
0.010
0.270
5.260
1.230
-0.299
0.015
0.000
8.035
2.398
Q1
Median
Q3
P90
-0.303
0.120
0.410
6.246
1.948
-0.097
0.019
0.000
8.845
3.526
-0.010
0.350
0.640
7.295
3.288
0.117
0.026
1.000
10.046
4.890
0.291
0.830
1.110
8.585
5.962
0.441
0.034
1.000
11.422
6.252
0.667
2.080
2.310
10.053
10.869
0.949
0.045
1.000
12.715
7.707
TABLE 1 Continued
Variable
Panel B: Correlation Coefficients (Spearman for the upper-right part, highlighted; Pearson for the bottom-left part)
News
News
Alpha
CoverageIT
CoverageNon-IT
Size
MB
Return
STD
R&D
Trade Size
Alpha
News CoverageIT
News CoverageNon-IT
Size
MB
Return
STD
R&D
Trade Size
Trade Frequency
-0.095
-0.066
-0.090
-0.008
0.051
0.064
0.005
-0.017
-0.071
-0.092
0.327
0.364
0.366
0.177
-0.072
0.173
0.040
0.569
-0.099
0.407
0.696
0.074
-0.052
-0.249
0.076
0.275
0.257
-0.095
0.395
0.777
0.179
0.017
-0.440
0.009
0.402
0.376
-0.018
0.272
0.181
0.299
0.343
0.091
0.266
0.046
0.285
0.017
0.112
0.005
0.107
0.399
0.255
0.113
0.028
0.141
0.050
-0.136
-0.316
-0.476
0.078
0.142
0.173
-0.166
-0.120
0.006
0.126
0.025
-0.016
0.365
0.102
0.204
-0.025
0.175
-0.016
0.027
0.270
0.384
0.120
0.071
-0.203
-0.031
-0.130
Trade
Frequency
-0.082
0.626
0.324
0.352
0.347
0.176
-0.068
0.170
-0.147
-
This table presents the summary statistics and Spearman (Pearson) correlation coefficients of the main variables used in this study based on the main sample of insider trades.
The variables are alpha (Alpha), news coverage related to insider trading (News CoverageIT), news coverage not related to insider trading (News CoverageNon-IT), firm size
(Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research & development (R&D), trade size (Trade Size), and trade frequency
(Trade Frequency). The construction of these variables is detailed in the Appendix. Panel A reports the number of observations (NObs), mean, median, standard deviation
(STD), and the decile (90% and 10%) and quartile (75% and 25%) distribution of the variables. Panel B reports the correlation coefficients among the variables above, where
the upper-right part (bottom-left part) of the table refers to the Spearman (Pearson) correlation matrix. The sample period is between January 2001 and June 2012 based on
insider trade dates.
30
TABLE 2
Time-variation in News Coverage
Panel A: News Coverage by Year and Firm Size
Year
Small
News CoverageIT
Medium
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
0.127
0.049
0.001
0.251
0.630
0.246
0.544
0.514
0.269
0.347
0.442
0.372
0.240
0.081
0.001
0.510
1.814
0.832
1.105
1.183
0.657
0.765
0.631
0.626
Large
0.437
0.082
0.018
0.349
3.425
3.087
1.473
1.887
0.662
1.388
0.945
0.841
Small
News CoverageNon-IT
Medium
Large
0.235
0.308
0.345
0.398
0.427
0.468
0.477
0.486
0.404
0.397
0.349
0.392
0.365
0.472
0.493
0.558
0.851
0.795
0.712
0.725
0.631
0.717
0.572
0.616
1.144
1.121
1.386
1.759
2.454
2.672
2.278
2.460
1.989
2.650
1.813
1.875
News CoverageIT
News CoverageNon-IT
-5
0.243
0.126
0.117
0.271
-4
0.209
0.112
0.097
0.205
-3
0.189
0.103
0.086
0.206
-2
0.222
0.124
0.098
0.218
-1
0.278
0.149
0.129
0.253
0
1.172
0.748
0.425
0.270
1
0.435
0.200
0.235
0.214
2
0.301
0.165
0.137
0.153
3
0.233
0.132
0.101
0.139
4
0.215
0.126
0.090
0.139
5
0.225
0.133
0.092
0.163
This table presents time-variation in news coverage by year or around SEC filings. Panel A shows the distribution of news
coverage variables by year and firm size. The sample firms are classified into three groups by market capitalization: small,
medium, and large. Panel B shows time-series variation of news coverage around SEC filings. The window is specified
five days before and after SEC filings. The news coverage variables include news coverage not related to insider trading
(News CoverageNon-IT), news coverage related to insider trading (News CoverageIT), first news coverage related to insider
trading (First News CoverageIT), repeated news coverage related to insider trading (Repeated News CoverageIT), and filing
frequency (Filing Frequency). Values on the date of SEC filings are highlighted in bold. The sample period is between
January 2001 and June 2012 based on insider trade dates.
31
TABLE 3
Determinants of News Coverage
Variable
Trade Frequency
Size
MB
Return
STD
R&D
ROA
S&P 500
Fixed Effects
Obs
News CoverageIT
Model
(1)
News CoverageNon-IT
Model
(2)
0.050
(43.01)
0.027
(22.40)
0.000
(0.65)
0.009
(4.70)
0.007
(0.09)
0.024
(4.04)
0.022
(2.84)
0.004
(0.47)
IY
40,130
0.003
(0.99)
0.177
(31.96)
-0.011
(-6.18)
-0.039
(-8.81)
4.495
(15.35)
0.037
(1.96)
-0.106
(-4.06)
0.963
(22.77)
IY
40,130
R2Adj
40.2%
57.2%
This table presents a regression of news coverage related to insider trading (News CoverageIT) in Model (1) and news
coverage not related to insider trading (News CoverageNon-IT) in Model (2) on firm-level characteristics as well as
unreported industry- and year-fixed effects (IY). The prior year firm-level characteristics include trade frequency (Trade
Frequency), firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD),
research & development (R&D), return-on-asset ratio (ROA), and S&P 500 index membership (S&P 500). The
construction of these variables is detailed in the Appendix. The t-statistics shown in parentheses are based on standard
errors adjusted for heteroskedasticity and firm-level clustering. Obs denotes the number of firm-year observations, and
R2Adj is adjusted R2. The sample period is from 2000 to 2011 based on news release dates (one year lagged to the main
sample of insider trades).
32
TABLE 4
News Coverage and Insiders Trading Profits
Variable
News CoverageIT
News CoverageNon-IT
Model
(1)
Model
(2)
Model
(3)
0.004
(0.68)
-0.017
(-4.10)
0.007
(1.24)
-0.017
(-3.89)
News CoverageSIT
Ex EA
Model
(4)
Firm Fixed
Model
(5)
-0.015
(-3.84)
-0.019
(-2.50)
Model
(6)
-0.023
(-2.12)
Model
(7)
-0.023
(-1.94)
News CoverageNon-SIT
-0.004
(-0.69)
-0.017
0.253
Size
-0.017
-0.022
-0.021
-0.019
-0.017
(-2.62)
(10.51)
(-2.87)
(-3.28)
(-3.29)
(-3.21)
(-2.61)
0.001
0.003
MB
0.001
0.000
0.001
0.001
0.001
(0.41)
(0.86)
(0.52)
(0.00)
(0.61)
(0.45)
(0.44)
0.056
0.011
Return
0.049
0.047
0.050
0.046
0.046
(1.98)
(0.61)
(1.92)
(1.82)
(1.99)
(1.77)
(1.78)
0.298
2.290
STD
0.433
0.322
0.349
0.388
0.442
(0.29)
(2.66)
(0.47)
(0.34)
(0.37)
(0.42)
(0.48)
0.020
-0.153
R&D
0.010
0.005
0.009
0.007
0.009
(0.93)
(-0.56)
(0.53)
(0.24)
(0.47)
(0.38)
(0.45)
0.001
-0.002
Trade Size
0.001
0.001
0.001
0.000
0.000
(0.22)
(-1.40)
(0.42)
(0.23)
(0.44)
(0.12)
(0.17)
0.002
0.017
Trade Frequency
-0.000
-0.007
0.000
-0.002
-0.001
(0.20)
(2.92)
(-0.01)
(-0.87)
(0.04)
(-0.25)
(-0.19)
IY
FY
Fixed Effects
IY
IY
IY
IY
IY
978,435
1,376,567
Obs
1,376,567 1,376,567 1,376,567
1,376,567 1,376,567
3.6%
27.3%
R2Adj
3.1%
2.9%
3.1%
3.0%
3.0%
This table presents a regression of insiders trading profits measured by alpha (Alpha) on news coverage related to insider
trading (News CoverageIT), news coverage not related to insider trading (News CoverageNon-IT), as well as firm-level
control variables and unreported industry- and year-fixed effects (IY). The firm-level control variables include firm size
(Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research & development
(R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction of these variables is detailed in
the Appendix. Model (4) is based on a subsample of insiders trades with no earnings announcements in a window ten
days [-10, 10] before and after insiders trades. Model (5) uses the change in news coverage from year t-1 to year t as the
regressor (if news coverage dataset is not available in year t-1, we use the average news coverage among all the firms in
year t instead) and controls for firm- and year-fixed effects (FY). Models (6) and (7) present a panel regression of Alpha
on news coverage related to a specific insiders transactions (News CoverageSIT) and news coverage not related to a
specific insiders transactions (News CoverageNon-SIT). Key results are highlighted in bold. The t-statistics shown in
parentheses are based on standard errors adjusted for heteroskedasticity and firm-level clustering. Obs denotes the number
of transaction observations, and R2Adj is adjusted R2. The sample period is between January 2001 and June 2012 based on
insider trade dates.
33
TABLE 5
Economic Mechanisms
Variable
News CoverageITIA
News CoverageIT
IA
Size
MB
Return
STD
R&D
Trade Size
Trade Frequency
Fixed Effects
Obs
R2Adj
34
Big N
Model
(2)
-7.737
(-4.60)
0.080
(3.25)
-0.011
(-2.93)
3.252
(1.44)
-0.018
(-2.61)
0.000
(0.19)
0.059
(1.87)
0.065
(0.06)
0.012
(0.47)
-0.000
(-0.04)
0.004
(0.62)
IY
1,173,814
-0.095
(-4.16)
0.011
(0.50)
-0.018
(-2.90)
0.001
(0.31)
0.050
(2.00)
0.681
(0.77)
0.014
(0.71)
0.001
(0.57)
0.001
(0.17)
IY
1,376,567
4.0%
3.6%
TABLE 5 Continued
LR
Variable
News CoverageITLR
LR
News CoverageIT
Size
MB
Return
STD
R&D
Trade Size
Trade Frequency
Fixed Effects
Obs
R2Adj
Model
(1)
Non-senior
Officers
Model
(5)
Inside
Directors
Model
(6)
Outside
Directors
Model
(7)
-0.680
(-2.14)
0.296
(0.83)
0.003
(0.26)
-0.007
(-0.98)
0.002
(0.61)
0.027
(2.01)
1.117
(1.56)
-0.019
(-1.01)
0.000
(0.17)
-0.000
(-0.06)
IY
1,260,880
-0.020
(-4.79)
-0.014
(-2.26)
0.002
(0.87)
0.061
(2.25)
-0.596
(-0.56)
0.006
(0.26)
0.003
(1.03)
0.006
(0.78)
IY
1,239,420
0.037
(1.09)
-0.007
(-1.06)
-0.008
(-3.23)
0.001
(0.04)
4.097
(4.68)
0.008
(0.24)
0.001
(0.19)
-0.016
(-1.47)
IY
137,147
-0.017
(-2.96)
-0.010
(-1.44)
0.001
(0.63)
0.064
(2.15)
0.126
(0.09)
0.014
(0.57)
-0.001
(-0.32)
-0.008
(-0.80)
IY
485,231
-0.017
(-3.44)
-0.011
(-2.05)
-0.001
(-0.43)
0.019
(1.02)
1.102
(1.34)
0.018
(0.72)
-0.001
(-0.53)
0.006
(1.05)
IY
414,367
-0.012
(-1.03)
-0.006
(-0.48)
0.009
(1.82)
-0.023
(-0.75)
1.362
(0.58)
0.048
(0.84)
-0.006
(-0.75)
-0.015
(-1.33)
IY
80,192
0.016
(0.66)
-0.021
(-2.95)
0.000
(0.16)
0.060
(1.87)
0.808
(0.79)
-0.027
(-0.97)
0.002
(0.38)
0.002
(0.27)
IY
396,777
3.3%
3.6%
6.0%
6.3%
2.9%
12.1%
4.7%
35
TABLE 5 Continued
Variable
News CoverageITCR
-0.066
-0.055
-0.080
-0.098
(-3.18)
(-2.33)
(-2.62)
(-3.56)
News CoverageIT
0.578
0.481
0.066
0.083
(3.19)
(2.34)
(2.34)
(3.34)
CR
0.032
0.035
0.024
0.024
(2.10)
(2.27)
(1.21)
(1.32)
Size
-0.008
-0.011
-0.001
-0.002
(-0.86)
(-1.10)
(-0.14)
(-0.21)
MB
0.008
0.008
0.004
0.004
(2.09)
(2.19)
(1.06)
(1.10)
Return
0.000
0.001
0.009
0.008
(0.02)
(0.06)
(0.49)
(0.42)
STD
1.385
1.123
1.518
1.458
(1.26)
(1.03)
(1.26)
(1.23)
R&D
-0.042
-0.041
-0.020
-0.015
(-1.50)
(-1.56)
(-0.87)
(-0.67)
Trade Size
0.003
0.003
0.003
0.003
(1.28)
(1.19)
(1.03)
(0.94)
Trade Frequency
-0.010
-0.013
-0.015
-0.016
(-1.63)
(-2.16)
(-2.23)
(-2.39)
Fixed Effects
IY
IY
IY
IY
Obs
521,157
557,478
838,216
838,216
R2Adj
6.7%
6.2%
3.6%
4.0%
This table presents a regression of insiders transaction profits measured by alpha (Alpha) on news coverage related to
insider trading (News CoverageIT), its interaction with information asymmetry variables (IA) in Panel A, its interaction
with litigation risk (LR) in Panel B, and its interaction with capital-at-risk variables (CR) in Panel C, as well as firm-level
control variables and unreported industry- and year-fixed effects (IY). The firm-level control variables include firm size
(Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research & development
(R&D), trade size (Trade Size), and trade frequency (Trade Frequency). In Panel A, the information asymmetry variables
include analyst forecast dispersion (Analyst Forecast Dispersion) and the Big N auditors (Big N) indicator. In Panel B, we
use Kim and Skinner's (2012) litigation risk measure (LR), and further split insiders transactions based on the types of
trades (sales and purchases) and the classifications of insiders, including senior officers, non-senior officers, executive
insiders, and non-executive insiders. Senior officers are insiders holding roles of CEO, CFO, and/or Chairman. Non-senior
officers are pure officers excluding CEO, CFO and/or Chairman with no director roles. Inside directors are those with both
director and officer roles excluding CEO, CFO and/or Chairman. Outside directors are pure directors excluding Chairman
with no officer roles. In Panel C, capital-at-risk variables include CEO's equity compensation (CEO Equity Compensation),
executives' equity compensation (Executive Equity Compensation), and two measures of corporate social responsibility
(Corporate Social Responsibility1, Corporate Social Responsibility2). The construction of these variables is detailed in the
Appendix. Key results are highlighted in bold. The t-statistics shown in parentheses are based on standard errors adjusted
for heteroskedasticity and firm-level clustering. Obs denotes the number of transaction observations, and R2Adj is adjusted
R2. The sample period is between January 2001 and June 2012 based on insider trade dates.
36
TABLE 6
Decrease in Profits or Increase in Losses
Alpha
News CoverageIT
Mean
z-statistics
0.000
0.022
(2.20)
[0.002
0.042]
0.010
0.022
(2.19)
[0.002
0.041]
0.080
0.020
(2.10)
[0.001
0.040]
0.160
0.019
(1.99)
[0.000
0.038]
0.250
0.017
(1.86)
[-0.002
0.036]
0.350
0.015
(1.71)
[-0.003
0.034]
0.490
0.012
(1.49)
[-0.006
0.031]
0.690
0.008
(1.15)
[-0.010
0.026]
1.030
0.000
(0.54)
[-0.019
0.018]
2.080
-0.051
(-1.25)
[-0.085
-0.018]
High News CoverageIT
This table presents the results of marginal effects at the means analysis of the relationship between expected insiders
trading profits measured by alpha (Alpha) and news coverage related to insider trading (News CoverageIT). Conditional on
means of firm-level control variables as well as unreported industry- and year-fixed effects, we report the expected
insiders trading profits estimated based on the decile-ranking values of News CoverageIT. The firm-level control variables
include firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research &
development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction of these variables
is detailed in Appendix A. Mean denotes the average value of expected Alpha. The z-statistics shown in parentheses are
based on standard errors of Alpha. 95% confidence interval of Alpha is reported in brackets. The sample period is between
January 2001 and June 2012 based on insider trade dates.
37
TABLE 7
News Dissemination or News Exploration
Variable
Adjusted News CoverageIT
Filing Frequency
Model
(1)
Model
(2)
Model
(3)
-0.056
(-1.65)
-0.014
(-1.76)
-0.034
(-0.76)
-0.018
(-3.70)
Model
(4)
-0.022
(-4.43)
Model
(5)
Model
(6)
-0.039
(-5.03)
First News CoverageIT
-0.004
0.128
(-0.10)
(2.10)
Size
-0.017
-0.019
-0.017
-0.017
-0.019
-0.017
(-2.91)
(-3.06)
(-3.00)
(-2.87)
(-2.93)
(-2.88)
MB
0.001
0.001
0.001
0.001
-0.000
0.001
(0.35)
(0.52)
(0.60)
(0.59)
(-0.02)
(0.45)
Return
0.049
0.046
0.048
0.049
0.047
0.049
(1.93)
(1.76)
(1.88)
(1.94)
(1.78)
(1.96)
STD
0.440
0.363
0.419
0.476
0.368
0.699
(0.48)
(0.39)
(0.45)
(0.52)
(0.40)
(0.80)
R&D
0.009
0.011
0.011
0.010
0.006
0.005
(0.46)
(0.56)
(0.59)
(0.53)
(0.30)
(0.24)
Trade Size
0.001
0.001
0.001
0.001
0.001
0.000
(0.34)
(0.48)
(0.46)
(0.40)
(0.22)
(0.02)
Trade Frequency
-0.002
0.002
0.002
-0.000
-0.006
-0.006
(-0.34)
(0.28)
(0.28)
(-0.01)
(-0.97)
(-0.97)
Fixed Effects
IY
IY
IY
IY
IY
IY
Obs
1,376,567 1,376,567 1,376,567 1,376,567 1,376,567 1,376,567
R2Adj
3.1%
3.0%
3.1%
3.1%
2.9%
3.3%
This table presents results of regressions of insiders trading profits measured by alpha (Alpha) on adjusted news coverage
related to insider trading (Adjusted News CoverageIT), filing frequency (Filing Frequency), repeated news coverage
related to insider trading (Repeated News CoverageIT), first news coverage related to insider trading (First News
CoverageIT), as well as firm-level control variables and unreported industry- and year-fixed effects (IY). The firm-level
control variables include firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility
(STD), research & development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction
of these variables is detailed in the Appendix. Key results are highlighted in bold. The t-statistics shown in parentheses are
based on standard errors adjusted for heteroskedasticity and firm-level clustering. Obs denotes the number of transaction
observations, and R2Adj is adjusted R2. The sample period is between January 2001 and June 2012 based on insider trade
dates.
38
TABLE 8
Insiders Propensity and Timing to Trade
Variable
Abnormal Transactions
Trade VolumeAbn
Trade SizeAbn
Model
Model
(1)
(2)
News CoverageIT
-4.220
-4.357
-0.024
-0.023
(-17.72)
(-14.45)
(-4.33)
(-4.16)
Size
0.294
0.306
0.009
0.009
(12.19)
(9.84)
(10.63)
(10.76)
MB
-0.004
-0.025
-0.001
-0.001
(-0.31)
(-1.57)
(-2.62)
(-2.84)
Return
-0.270
-0.362
0.002
0.002
(-3.45)
(-3.73)
(1.03)
(0.90)
STD
-2.428
-6.558
-0.145
-0.142
(-1.02)
(-2.17)
(-1.87)
(-1.83)
R&D
0.062
0.160
-0.003
-0.003
(0.51)
(1.02)
(-0.67)
(-0.66)
Trade Frequency
-0.838
-0.927
0.018
0.018
(-33.17)
(-28.72)
(24.86)
(24.61)
Fixed Effects
IY
IY
IY
IY
Obs
37,341
37,341
37,341
37,341
R2Adj
6.3%
5.9%
4.3%
4.3%
This table presents the results of a regression of abnormal insider trading measured by abnormal trade volume (Trade
VolumeAbn) and abnormal trade size (Trade SizeAbn), and the timing of insider trading measured by trade volume ratio
(Trade VolumeRatio) and trade size ratio (Trade SizeRatio) on news coverage related to insider trading (News CoverageIT), as
well as firm-level control variables and unreported industry- and year-fixed effects (IY). The firm-level control variables
include firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research &
development (R&D), and trade frequency (Trade Frequency). The construction of these variables is detailed in the
Appendix. Key results are highlighted in bold. The t-statistics shown in parentheses are based on standard errors adjusted
for heteroskedasticity and firm-level clustering. Obs denotes the number of transaction observations, and R2Adj is adjusted
R2. The sample period is 2001 to 2011 based on insider trade dates, where the year 2012 is dropped because our main
sample of insider trades ends in June 2012.
39
TABLE 9
Endogeneity Tests
Variable
ProximityRank
Alpha
Stage 2
Model
(4)
0.509
(3.17)
ProximityCategory
0.288
(2.35)
MB
Return
STD
R&D
Trade Size
Trade Frequency
Fixed Effects
Obs
R2Adj
Alpha
Stage 2
Model
(6)
0.639
(2.53)
ProximityDummy
Size
News CoverageIT
Stage 1
Model
(5)
0.148
(1.49)
0.063
(2.46)
0.158
(1.34)
2.979
(0.88)
0.268
(2.18)
0.033
(1.45)
0.386
(5.32)
IY
1,363,540
-0.099
(-2.05)
-0.003
(-0.26)
0.006
(1.44)
0.061
(2.45)
0.759
(0.81)
0.035
(1.31)
0.003
(1.00)
0.032
(1.53)
IY
1,363,540
48.8%
3.0%
40
0.140
(1.52)
0.063
(2.42)
0.156
(1.29)
2.301
(0.72)
0.286
(2.34)
0.032
(1.31)
0.385
(5.64)
IY
1,363,540
-0.080
(-2.11)
-0.006
(-0.76)
0.005
(1.53)
0.058
(2.41)
0.690
(0.74)
0.030
(1.25)
0.003
(0.98)
0.024
(1.59)
IY
1,363,540
0.150
(1.51)
0.061
(2.50)
0.156
(1.33)
2.737
(0.77)
0.293
(2.43)
0.031
(1.35)
0.388
(5.37)
IY
1,363,540
-0.107
(-2.29)
-0.002
(-0.14)
0.007
(1.61)
0.062
(2.45)
0.786
(0.80)
0.038
(1.40)
0.003
(1.09)
0.035
(1.60)
IY
1,363,540
49.3%
3.0%
48.9%
3.0%
TABLE 9 Continued
Panel B: A Natural Experiment with Dow Jones Formal Coverage of Insider Trading Filings
Variable
ProximityRank
2001-2003
2004-2007
2001-2003
2004-2007
2001-2003
2004-2007
2001-2003
2004-2007
Model
Model
Model
Model
Model
Model
Model
Model
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
-0.025
-0.070
(-0.42)
(-1.35)
ProximityDummy
-0.010
-0.048
(-0.31)
(-1.67)
ProximityCategory
-0.006
-0.045
(-0.22)
(-1.89)
News CoverageIT
0.022
-0.022
(0.30)
(-3.69)
Difference
-0.045
-0.038
-0.039
-0.044
[P-value]
[0.42]
[0.22]
[0.10]
[0.09]
Size
MB
Return
STD
R&D
Trade Size
Trade Frequency
Fixed Effects
-0.008
-0.014
-0.008
-0.013
-0.008
-0.014
-0.008
-0.013
(-0.76)
(-1.39)
(-0.75)
(-1.39)
(-0.74)
(-1.38)
(-0.73)
(-1.41)
0.000
0.002
0.000
0.002
0.000
0.002
0.000
0.003
(0.09)
(0.64)
(0.10)
(0.61)
(0.10)
(0.81)
(0.12)
(1.26)
-0.008
0.069
-0.008
0.069
-0.008
0.068
-0.007
0.076
(-0.33)
(1.53)
(-0.33)
(1.53)
(-0.33)
(1.54)
(-0.28)
(1.77)
-0.862
0.225
-0.878
0.342
-0.878
0.437
-1.016
0.022
(-0.74)
(0.13)
(-0.76)
(0.19)
(-0.74)
(0.25)
(-0.88)
(0.01)
-0.009
0.010
-0.010
0.008
-0.010
0.006
-0.012
0.014
(-0.18)
(0.38)
(-0.20)
(0.30)
(-0.20)
(0.21)
(-0.25)
(0.51)
-0.003
0.001
-0.003
0.002
-0.003
0.001
-0.004
0.004
(-0.46)
(0.33)
(-0.46)
(0.40)
(-0.47)
(0.32)
(-0.56)
(1.06)
0.006
-0.015
0.006
-0.015
0.006
-0.015
0.006
-0.006
(0.53)
(-1.67)
(0.52)
(-1.65)
(0.53)
(-1.63)
(0.47)
(-0.61)
IY
IY
IY
IY
IY
IY
IY
IY
Obs
164,443
741,779
164,443
741,779
164,443
741,779
165,162
747,895
R2Adj
4.9%
4.2%
4.9%
4.2%
4.9%
4.4%
4.8%
4.6%
This table provides tests to address endogeneity issues. Panel A presents regression results of an instrumental variable
approach. In Models (1), (3), and (5), news coverage related to insider trading (News CoverageIT) is predicted by the
instrumental variables, respectively geographic proximity rank (ProximityRank), geographic proximity dummy
(ProximityDummy), and geographic proximity category (ProximityCategory), and firm-level control variables as well as
unreported industry- and year-fixed effects (IY). In Models (2), (4), and (6), insider trading profit measured by alpha
(Alpha) is regressed on the predicted news coverage related to insider trading (Predicted News CoverageIT). The firm-level
control variables include firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility
(STD), research & development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction
of these variables is detailed in the Appendix. Panel B explores a natural experiment based on the date when Dow Jones
formally initiated coverage of insider trading filings - January 2004. The period preceding systematic coverage is 20012003 in Models (1), (3), (5), and (7). The period of systematic coverage is 2004-2007 in Models (2), (4), (6), and (8). Key
results are highlighted in bold. The t-statistics shown in parentheses are based on standard errors adjusted for
heteroskedasticity and firm-level clustering. Obs denotes the number of transaction observations, and R2Adj is adjusted R2.
The sample period is between January 2001 and June 2012 based on insider trade dates.
41
DEFINITION TABLE
Variable
Acronym
Definition
Data Source
SEC Investigation
Dummy variable that equals one if the firm is investigated by the SEC in a year.
RavenPack
Litigation indicator
Litigation Defendant
Dummy variable that equals one if the firm is sued for legal, insider trading, and/or fraud issues in a year.
RavenPack
Firm-level alpha
AlphaFirm
Average firm-level abnormal return for the transactions made by all the insiders of the firm in a year.
Market-adjusted return
ReturnADJ
AlphaIndex
Buy and hold market adjusted return in a window [1, 180] following the transaction. For sales, return is multiplied by -1.
Annualized abnormal daily return based on Cremers, Petajisto, and Zitzewitzs [2013] index-based four-factor model in
Index-based alpha
Three-month alpha
Alpha90
a window [1, 180] following the transaction. For sales, return is multiplied by -1.
Annualized abnormal daily return based on the four-factor model in a window [1, 90] following the transaction.
Trade Volume
Thomson Reuters
Trade price
News coverage related to insider trading over the
Trade Price
News CoverageIT [-2Y,0]
Thomson Reuters
RavenPack
Number of news releases related to insider trading in a window [-1080, -1] prior to the transaction scaled by 100.
RavenPack
Number of news releases related to insider trading in a window [-720, -361] prior to the transaction scaled by 100.
RavenPack
Number of news releases related to insider trading in a window [-1080, -721] prior to the transaction scaled by 100.
RavenPack
Number of high impact news releases related to insider trading scaled by 100.
RavenPack
Number of low impact news releases related to insider trading scaled by 100.
RavenPack
Number of news releases related to insider trading in a window [-180, -1] prior to the transaction scaled by 100.
RavenPack
Number of news releases related to insider trading in a window [-360, -181] prior to the transaction scaled by 100.
RavenPack
Routine Trade
Insider trades made by routine traders, who trade in the same month over the previous three years.
Thomson Reuters
Non-Routine Trade
Insider trades made by non-routine traders, who are not identified as routine traders.
Thomson Reuters
ROBUSTNESS TESTS
1. Alternative Specifications
In Table IA1, we run robustness tests with varying dependent variables in Models (1)
to (4) using market-adjusted returns (ReturnADJ), index-based alpha (AlphaIndex), three-month
alpha (Alpha90), and the product of trade size and alpha (Trade SizeAlpha). We cluster
standard errors by year in Model (5), by industry in Model (6), by insider in Model (7), and
by firm and year in Model (8). In Models (9) to (11), we utilize alternative samples by
excluding trivial trades, using as cut-off points trade price (ten dollars), trade volume (1,000
shares), or trade size (10,000 dollars). Our results survive in all of these specifications.
The score is a measure of confidence that the story will have a value impact on the market
over the two-hour period following its release, and is given a value between zero and 100.
We define High Impact News CoverageIT (Low Impact News CoverageIT) as the number of
relevant news releases with NIP equal to or greater than (below) the mean NIP in the sample
(equal to 40). To measure the time relevance of specific news related to insider trading, we
define Recent News CoverageIT as news articles published within 180 days of the insider
transaction. The results suggest that high impact and recent news coverage plays an important
role in attenuating insiders profits.
TABLE IA1
Alternative Specifications
Alternative measures of insiders' trading profits
Variable
News CoverageIT
Size
MB
Return
STD
R&D
Trade Size
Trade Frequency
Fixed Effects
Obs
R2Adj
Alternative samples
ReturnADJ
AlphaIndex
Alpha90
Trade SizeAlpha
CYear
CIndustry
CInsider
CFirmYear
Trade Price$10
Trade Volume1,000
Trade Size$10,000
Model
(1)
Model
(2)
Model
(3)
Model
(4)
Model
(5)
Model
(6)
Model
(7)
Model
(8)
Model
(9)
Model
(10)
Model
(11)
-0.007
(-3.73)
-0.003
(-0.85)
0.000
(0.46)
0.027
(2.61)
1.048
(2.38)
-0.006
(-0.58)
0.001
(0.96)
0.002
(0.58)
IY
1,376,567
3.3%
-0.017
(-3.56)
-0.019
(-3.13)
0.001
(0.50)
0.051
(2.06)
-0.048
(-0.05)
0.003
(0.16)
0.002
(0.65)
0.002
(0.34)
IY
1,268,228
3.2%
-0.015
(-2.93)
-0.018
(-2.66)
0.001
(0.32)
0.038
(1.40)
0.597
(0.52)
-0.007
(-0.27)
0.002
(0.68)
-0.003
(-0.45)
IY
1,376,567
2.0%
-0.407
(-4.24)
-0.203
(-2.22)
0.012
(0.46)
0.434
(1.55)
5.183
(0.40)
0.088
(0.26)
-0.017
(-3.29)
-0.017
(-1.99)
0.001
(0.29)
0.049
(1.91)
0.433
(0.37)
0.010
(0.68)
0.001
(0.47)
-0.000
(-0.02)
IY
1,376,567
3.1%
-0.017
(-4.99)
-0.017
(-3.12)
0.001
(0.45)
0.049
(1.98)
0.433
(0.59)
0.010
(0.63)
0.001
(0.41)
-0.000
(-0.01)
IY
1,376,567
3.1%
-0.017
(-5.21)
-0.017
(-4.20)
0.001
(0.79)
0.049
(2.91)
0.433
(0.64)
0.010
(0.73)
0.001
(0.50)
-0.000
(-0.02)
IY
1,376,567
3.1%
-0.017
(-4.95)
-0.016
(-1.77)
0.001
(0.27)
0.049
(1.77)
0.561
(0.46)
0.010
(0.76)
0.001
(0.38)
-0.000
(-0.05)
IY
1,376,567
3.1%
-0.020
(-4.54)
-0.015
(-2.22)
0.002
(0.76)
0.063
(2.06)
-0.192
(-0.16)
0.006
(0.29)
0.003
(1.27)
0.002
(0.31)
IY
1,192,228
3.9%
-0.024
(-3.85)
-0.006
(-1.42)
0.001
(0.69)
0.014
(1.27)
1.282
(2.07)
-0.012
(-0.71)
0.001
(0.51)
-0.003
(-0.76)
IY
651,208
1.9%
-0.017
(-3.76)
-0.009
(-1.97)
0.003
(1.59)
0.025
(1.62)
0.117
(0.14)
-0.004
(-0.22)
0.003
(1.33)
-0.005
(-0.95)
IY
927,693
2.9%
-0.029
(-0.30)
IY
1,376,567
3.3%
This table reports results of robustness tests on the relationship between news coverage and insiders trading profits. The table presents a panel regression of insiders trading
profits measured by market-adjusted returns (ReturnADJ) in Model (1), index-based alpha (AlphaIndex) in Model (2), three-month alpha (Alpha90) in Model (3), and the product
between trade size and alpha (Trade SizeAlpha) in Model (4), on news coverage related to insider trading (News CoverageIT) as well as firm-level control variables and
unreported industry- and year-fixed effects (IY). The firm-level control variables include firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock
return volatility (STD), research & development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction of these variables is detailed in the
Appendix of the paper and in the first section of this Internet Appendix. Key results are highlighted in bold. t-statistics shown in parentheses are based on standard errors
adjusted for heteroskedasticity and firm-level clustering except in Models (5)-(8). The standard errors in Models (5)-(8) are clustered by year, industry, insider, and both firm
and year, respectively. Models (9)-(11) show the results for alternative samples: stock price is more than $10 in Model (9); trade volume is more than 1,000 shares in Model
(10); Trade size is more than $10,000 in Model (11). Obs denotes the number of transaction observations, and R2Adj is adjusted R2. The sample period is between January
2001 and June 2012 based on insider trade dates.
Internet Appendix, Page8
TABLE IA2
Falsification Test on Stale News
Variable
News CoverageIT [-2Y,0]
Model
(1)
Model
(2)
Model
(3)
Model
(4)
-0.000
(-0.53)
STD
R&D
Trade Size
Trade Frequency
Fixed Effects
Obs
-0.004
(-0.31)
0.000
(0.59)
News CoverageIT
Return
-0.001
(-0.09)
-0.000
(-1.49)
MB
Model
(6)
-0.000
(-1.86)
Size
Model
(5)
-0.016
(-2.50)
0.000
(0.22)
0.045
(1.69)
0.593
(0.59)
0.010
(0.48)
0.001
(0.43)
-0.002
(-0.21)
IY
1,337,209
-0.016
(-2.24)
0.000
(0.10)
0.044
(1.63)
0.682
(0.61)
0.010
(0.52)
0.001
(0.42)
-0.001
(-0.08)
IY
1,292,631
-0.018
(-2.59)
-0.000
(-0.16)
0.044
(1.69)
0.516
(0.50)
0.006
(0.32)
0.001
(0.27)
-0.006
(-0.74)
IY
1,337,209
-0.021
(-2.59)
-0.000
(-0.23)
0.047
(1.72)
0.575
(0.51)
0.005
(0.26)
0.000
(0.18)
-0.006
(-0.73)
IY
1,292,631
0.020
(1.25)
-0.017
(-3.86)
-0.016
(-2.49)
0.001
(0.36)
0.047
(1.86)
0.563
(0.55)
0.011
(0.53)
0.001
(0.45)
-0.000
(-0.04)
IY
1,337,209
-0.018
(-4.23)
-0.018
(-2.45)
0.001
(0.41)
0.049
(1.88)
0.730
(0.66)
0.011
(0.54)
0.001
(0.40)
0.002
(0.24)
IY
1,292,631
R2Adj
3.0%
3.1%
2.9%
2.9%
3.1%
3.2%
This table presents a panel regression of insiders trading profits measured by alpha (Alpha) on news coverage
related to insider trading over the last two years (News CoverageIT [-2Y,0]), news coverage related to insider
trading over the last three years (News CoverageIT [-3Y,0]), news coverage related to insider trading within the
second last year (News CoverageIT [-2Y,-1Y]), news coverage related to insider trading within the third last year
(News CoverageIT [-3Y,-2Y]), news coverage related to insider trading (News CoverageIT), and firm-level control
variables as well as unreported industry- and year-fixed effects (IY). The firm-level control variables include
firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research
& development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction of
these variables is detailed in the Appendix of the paper and in the first section of this Internet Appendix. Key
results are highlighted in bold. The t-statistics shown in parentheses are based on standard errors adjusted for
heteroskedasticity and firm-level clustering. Obs denotes the number of transaction observations, and R2Adj is
adjusted R2. The full sample period is between January 2001 and June 2012 based on insider trade dates.
TABLE IA3
Value Impact and Timeliness of News
Variable
High Impact News CoverageIT
Low Impact News CoverageIT
Model
(1)
News impact
Model
(2)
Model
(3)
-0.015
(-0.61)
-0.026
(-2.65)
0.029
(0.74)
-0.022
(-3.66)
Model
(4)
-0.033
(-3.16)
News period
Model
(5)
Model
(6)
-0.030
(-1.63)
Old News CoverageIT
-0.017
-0.005
(-1.37)
(-0.25)
Size
-0.017
-0.019
-0.018
-0.017
-0.018
-0.017
(-2.94)
(-2.92)
(-3.04)
(-2.89)
(-2.92)
(-2.85)
MB
0.001
0.000
0.001
0.001
0.001
0.001
(0.55)
(0.05)
(0.49)
(0.41)
(0.29)
(0.49)
Return
0.049
0.047
0.049
0.050
0.047
0.050
(1.93)
(1.80)
(1.93)
(1.97)
(1.83)
(1.99)
STD
0.457
0.369
0.480
0.500
0.368
0.489
(0.50)
(0.39)
(0.54)
(0.55)
(0.39)
(0.55)
R&D
0.011
0.006
0.010
0.011
0.008
0.011
(0.55)
(0.33)
(0.52)
(0.57)
(0.39)
(0.58)
Trade Size
0.001
0.001
0.001
0.001
0.001
0.001
(0.42)
(0.26)
(0.38)
(0.43)
(0.31)
(0.44)
Trade Frequency
0.000
-0.006
-0.000
-0.000
-0.003
0.000
(0.07)
(-0.70)
(-0.01)
(-0.00)
(-0.45)
(0.06)
Fixed Effects
IY
IY
IY
IY
IY
IY
Obs
1,376,567 1,376,567 1,376,567
1,376,567 1,376,567 1,376,567
R2Adj
3.1%
2.9%
3.2%
3.1%
3.0%
3.1%
This table presents a panel regression of insiders trading profits measured by alpha (Alpha) on high impact
news coverage related to insider trading (High Impact News CoverageIT), low impact news coverage related to
insider trading (Low Impact News CoverageIT), recent news coverage related to insider trading (Recent News
CoverageIT), old news coverage related to insider trading (Old News CoverageIT), and firm-level control
variables as well as unreported industry- and year-fixed effects (IY). The firm-level control variables include
firm size (Size), market-to-book ratio (MB), annual stock return (Return), stock return volatility (STD), research
& development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency). The construction of
these variables is detailed in the Appendix of the paper and in the first section of this Internet Appendix. Key
results are highlighted in bold. The t-statistics shown in parentheses are based on standard errors adjusted for
heteroskedasticity and firm-level clustering. Obs denotes the number of transaction observations, and R2Adj is
adjusted R2. The sample period is between January 2001 and June 2012 based on insider trade dates.
TABLE IA4
Routine and Non-Routine Transactions
Variable
News CoverageIT
Method one
Non-Routine
Routine
Model
Model
(1)
(2)
-0.044
(-1.95)
0.014
(1.53)
Method two
Non-Routine
Routine
Model
Model
(3)
(4)
-0.042
(-1.80)
-0.058
[0.01]
0.014
(1.65)
Method three
Non-Routine
Routine
Model
Model
(5)
(6)
-0.043
(-1.87)
-0.056
[0.01]
0.014
(1.52)
Method four
Non-Routine
Routine
Model
Model
(7)
(8)
-0.042
(-1.84)
-0.057
[0.01]
0.014
(1.65)
-0.056
[0.01]
0.003
(0.35)
0.008
(1.67)
0.031
(1.21)
1.917
(1.44)
-0.037
(-1.26)
0.004
(1.23)
0.011
(1.25)
IY
204,090
-0.042
(-2.77)
-0.010
(-2.70)
0.081
(2.27)
-0.492
(-0.26)
0.029
(0.62)
0.007
(1.14)
0.048
(2.55)
IY
126,938
0.001
(0.09)
0.008
(2.13)
0.033
(1.33)
1.851
(1.44)
-0.035
(-1.23)
0.004
(1.12)
0.010
(1.20)
IY
207,754
-0.040
(-2.58)
-0.012
(-3.37)
0.076
(1.86)
-0.544
(-0.27)
0.030
(0.66)
0.008
(1.18)
0.048
(2.49)
IY
123,274
0.002
(0.28)
0.008
(1.85)
0.052
(2.27)
2.185
(1.67)
-0.036
(-1.25)
0.004
(1.11)
0.008
(0.99)
IY
199,978
-0.041
(-2.70)
-0.010
(-2.64)
0.078
(2.09)
-0.536
(-0.27)
0.025
(0.55)
0.008
(1.18)
0.049
(2.55)
IY
125,612
0.001
(0.14)
0.009
(2.37)
0.052
(2.31)
2.171
(1.67)
-0.035
(-1.22)
0.004
(1.09)
0.008
(0.97)
IY
202,316
-0.040
(-2.58)
-0.012
(-3.37)
0.076
(1.86)
-0.544
(-0.27)
0.030
(0.66)
0.008
(1.18)
0.048
(2.49)
IY
123,274
7.5%
14.0%
7.5%
14.4%
8.2%
14.0%
8.2%
14.4%
This table presents a panel regression of insiders routine and non-routine transaction profits measured by alpha (Alpha) on news coverage related to insider trading (News
CoverageIT), and firm-level control variables as well as unreported industry- and year-fixed effects (IY). The firm-level control variables include firm size (Size), market-tobook ratio (MB), annual stock return (Return), stock return volatility (STD), research & development (R&D), trade size (Trade Size), and trade frequency (Trade Frequency).
The construction of these variables is detailed in the Appendix of the paper and in the first section of this Internet Appendix. Routine and non-routine transactions are
classified using four methods. Method One requires that to be included in this analysis, an insider needs to trade at least once in each of the last three years but places no
Internet Appendix, Page11
conditions on the identification of routine traders. To mitigate the impact of trivial trades in this identification process, the first condition is that the number of shares traded is
equal to or more than 100, and the second condition is that the transaction price is equal to or more than $2. Method Two applies both conditions to identify routine traders
i.e. identification based on those non-trivial trades executed in the same month over the three-year period preceding the inside transaction. In Method Three the two
conditions apply to the sample selection process we require an insider to make non-trivial trade at least once in each of the last three years. In Method Four we impose the
conditions to both the sample selection process and the identification of routine traders. Key results are highlighted in bold. The t-statistics shown in parentheses are based on
standard errors adjusted for heteroskedasticity and firm-level clustering. Obs denotes the number of transaction observations, and R2Adj is adjusted R2. The sample period is
between January 2001 and June 2012 based on insider trade dates.
TABLE IA5
SEC Investigation and Litigation Defendant
Variable
News CoverageITAlphaFirm
News CoverageIT
AlphaFirm
Size
MB
Return
STD
R&D
Trade Frequency
ROA
S&P 500
Fixed Effects
Obs
SEC Investigation
Model
(1)
Litigation Defendant
Model
(2)
Investigation or Defendant
Model
(3)
1.407
(3.14)
0.406
(0.99)
-0.030
(-0.09)
0.693
(7.75)
-0.021
(-0.83)
0.122
(0.54)
22.516
(3.00)
0.040
(0.08)
-0.122
(-1.90)
-1.258
(-2.21)
0.523
(1.62)
IY
40,130
0.437
(2.39)
0.203
(1.93)
0.057
(0.84)
0.622
(23.88)
-0.010
(-1.25)
-0.115
(-2.76)
17.158
(9.24)
0.233
(2.49)
0.040
(2.77)
-0.407
(-2.36)
0.276
(3.30)
IY
40,130
0.439
(2.41)
0.202
(1.91)
0.058
(0.86)
0.620
(24.06)
-0.010
(-1.27)
-0.111
(-2.67)
17.154
(9.30)
0.213
(2.28)
0.040
(2.78)
-0.432
(-2.53)
0.287
(3.44)
IY
40,130
R2Pseudo
21.1%
20.2%
20.3%
This table presents a panel (firm-year) logistic regression of SEC Investigation dummy in Model (1), Litigation
Defendant dummy in Model (2), or a combined SEC Investigation and Litigation Defendant dummy in Model
(3), on the interaction between firm-level trading profits (AlphaFirm) and news coverage related to insider trading
(News CoverageIT). We control for other prior year firm-level characteristics including firm size (Size), marketto-book ratio (MB), annual stock return (Return), stock return volatility (STD), research & development (R&D),
trade frequency (Trade Frequency), return-on-asset ratio (ROA), and S&P 500 index membership (S&P 500), as
well as unreported industry- and year-fixed effects (IY). The construction of these variables is detailed in the
Appendix of the paper and in the first section of this Internet Appendix. Key results are highlighted in bold. The
z-values shown in parentheses are based on standard errors adjusted for heteroskedasticity and firm-level
clustering. Obs denotes the number of firm-year observations, and R2Pseudo is Pseudo R2. The sample period is
from 2000 to 2011 based on news release dates (one year lagged to the main sample of insider trades).