Professional Documents
Culture Documents
ANALYSIS/MODERATOR
VARIABLES
(INSIDER TRADING,
EARNINGS AND STOCK
R.Ravichandran
MODERATOR MEDIATOR
VARIABLE (MO-ME)
Social Psychological Research
Conceptual, Strategic and Statistical Considerations
Distinguish between Moderator &Mediator Varaiable
Effect of Third variable
Focal Independent variable (FIV) sub group (domains of maximal effectiveness on Dependent variable of
Interest (DVI))
Generative mechanism of influence of FIV on DVI
Impact of identifiability on social loafing (identifiablity is important mediator of social loafing)
Gender, age, race and social economic level mediators of the relation between locus of control and academic
achievement
Contrasting the Mo-ME functions that delineate the implications of this distinctions for theory and research
Experimental Design, Research operations and plan of statistical analysis
Nature of causal mechanism and integrate seemingly irreconcilable theoretical positions
Problem areas of disagreement about mediators can be resolved by treating certain variables as
moderators
LITERATURE REVIEW
Unexpected changes in earnings are associated with unexpected
changes in firm value (Ball &Brown,1968)
Market agents learn about earnings and valuation- relevant events form
many information sources throughout the fiscal period
(Foster,1986/Watts & Simmerman,1989)
Significant changes in firm value follow the trading activity of corporate
insiders (Fama,1991)
Association arises because market acts as if insider tarding is important
source of information about long term prospects of the firm (Givoly and
Palmon (1985))
PURPOSE OF STUDY
To investigate if
the information captured b y insider trading modifies the responsiveness of
returns to AUE
the information captured by insider trading differs from that captured by annual
unexpected earnings
Findings
Insider Trading conveys information not fully captured by the annual earnings
ISSUES
Research Design & Measurement
Sample Selection Procedure
Test Procedure of empirical results
Concluding remarks
RESEARCH ISSUES
Information &Insider Trading
Earnings announcements & Insider trading
Earnings Measurement & Insider Trading
SEC Rule Insiders to Disclose material information prior to trading in
their firms securities or abstain from trading
Courts: extraordinary information reasonably certain to have a
substantial effect on the market price of the security if disclosed meets
materiality standard of the statues
Much of information available to insiders is less dramatic and they
remain relatively free to trade even when the future performance of their
firm in their assessment differs from that expected by the insiders
MEASUREMENT OF
ABNORMAL RETURN
15 month period from the beginning of the fiscal year
Period of relatively little to complete public dissemination of earnings
information
AR for firm j for the month t subsequent to the first of the year
Arjt = Rjt (aJ +bj Rmt ) where
Rjt = the actual return for firm j in the t month subsequent to the
beginning of the year
Rmt = the CRSP equally weighted market index for month t
aJ & bj = estimated OLS parameters for market model for firm j using
monthly returns over months -66 to 6 relative to beginning of the year
MEASUREMENT OF
UNEXPECTED EARNINGS
AE is difference between EPS number in COMPUSTAT for year Y and
a prediction of EPS for the year Y based on a random walk walk
with drift model
Drift term is computed as the mean year to year change in EPS
during prior 5 years
UE is standardized by the firms security price at the end of year Y-1
Firms with greater than 0 are classified as UE+ and less than 0 as
UE-
MEASUREMENT OF INSIDER
TRADING
Reported IT aggregated to first quarter of each year (assumptions)
In comparison to later quarter this interval is characterized by relatively low
generation of publicly available information about forth coming annual earnings
Interval precedes the announcement of first quarterly earnings report (analysts
are unlikely to revise the forecast in first quarter) & Information of advantage of
insiders is maximized during this period
Information about trading by corporate insiders may take as long as 50 days to
become publicly available (Stickel,1989)
Information reflected by insider trading may take more than 6 months to be fully
impounded in prices(Lee & Solt,1986)
15 month cumulative abnormal return are more likely to capture the information
underlying insider trading which occurred during first quarter of the year
compared to insider trading occurred later in the year
CLASSIFICATION PROCESS
2step process sensitive to direction and magnitude of IT activity
3 groups (Buy, No Trade, Sell) bases on Net number of shares (NNS)
transacted (Shares purchased Shares sold) through private and open
market transactions by the firms insiders during first quarter
NNS >0 =Buy NNS <0 = Sell NNS =0 No Trade
For each year firms in the buy and Sell groups are assigned to either the
maximum or moderate trading categories based on the relative intensity
of the insider trading
NNS for buy and sell are standardized by common shares traded in Prior
year and this metric is ranked
Observations in Top 50 of Buy &Sell are assigned MAX BUY /MAX SELL
Other half is classified as Moderate Buy &Sell
SAMPLE
All December 31st firms fro 1978 to 1987 listed in COMPUSTAT Annual
Tape
Dec.31st fro cross sectional consistency in the effect (year end
tax/portfolio adjustment and exercise and receipt of managerial stock
options on association of PE)
Monthly security return data on CRSP Tape
66 months prior to and 15 months subsequent to the first trading day of the year
Primary annual earnings per share data on the COMPUSTAT Annual Tape (Current year
and prior 5 years)
Common shares traded during the prior year and closing security price for the prior
year (8301 observations)
SAMPLE.
Official summary of security transactions and holdings published by SEC
Reports of the transactions disclosed by corporate officers, Directors and beneficial
owners under the provisions of the SEC Act of 1934
Private and open market purchases and sales reported by insiders for the firms
corresponding to 8301 observations (Jan 1978 to Jul,1987)
Reported IT Transactions were aggregated over the first quarter of each year
Each of 8301 observations were assigned to one of the five trading categories by
employing 2 step classification procedure
Buy (Max Buy/Moderate Buy & Sell (Max Sell/Moderate sell)
ANALYSIS OF INTERACTIVE
EFFECTS
If insider trading and unexpected earnings capture similar value relevant
a0 (t-stat)
B0(t-stat)
R^2 (%)
Interaction
Hypothesis
(Fstatistics)
0.08 (3.98)
1.07(6.21)
8.9
9.65
MaxBuy/UE-
-0.10(-3.39)
0.29(1.40)
0.7
0.02(1.4)
0.80(5.51)
4.1
Max sell/UE-
-0.11(-4.82)
0.60(3.84)
3.6
255
398
0.09
Coefficient in
Model
Estimated
Coefficient
T-statitic
Intercept
a0
-0.028
-2.53
UE Class
a1
0.078
7.14
IT class
a2
0.016
1.52
UE Class*IT Class
a3
0.013
1.1.8
Mean UE
b0
0.687
8.37
UE Class*UE
b1
0.246
2.99
IT Class*UE
b2
-0.009
-0.11
UE Class*IT
Class*UE
b3
0.145
1.77
RESULTS
Descriptive Analysis IT intensity, Standardized UE &Cumulative
abnormal Returns (CAR) over 15 months
43% of sample of 6484 observations show no net trading by insiders
When IT occurs net selling by insiders is more prevalent than net buying
(net sell to net buy ratio 1.56)
Intensity of Trading is very small (0.1% )(Insiders trade 10000 shares out
of total trading of 10 million shares during first quarter)
Mean of UE are not significantly different (F value 1.5) among 5 trading
categories
IT is poor predictor of direction of UE (Givoly& Palmon,1985)
Mean CAR for the Max buy group is the largest and the Mean CAR for the
Max sell group is smallest among 5 groups
HYPOTHETICAL PORTFOLIO
Plot of cumulative abnormal returns to a Ball and Brown type startegy
conditional on perfect foreknowledge of insider trading and sign of un
expected earnings
6 portfolio formed one of 6 groups on both the sign of UE and direction
of trading activity by IT
Rapid divergence of CARs associated with the portfolios with positive UE
Relationship between positive UE and AR is also affected by the
information conveyed by IT
This effect appears rapidly around the end of I Quarter
Differences in the patterns of CAR are attributable to differences in
magnitude of annual UE
QUALITATIVE INSIGHT TO
ADDITIVE EFFECTS
Negative UE the regression line of confirmatory group Max sell/UEis below regression line of contradictory group Max Buy/UE- in the
relevant range of UE
Since neither mean of UE nor the intercept terms for these 2 groups
are significantly different an additive effect of trading cannot be
substantiated
Difference between Car plots between Max sell/UE- and Max
Buy/UE- groups attributable solely to interactive effect
CONCLUSIONS
Insider buying &selling is associated average positive abnormal returns
Buying & Selling activity by insiders does not always precede the public
release of good/bad news by firms
Likely that abnormal returns associated with IT result from security market
acting as if insider conveys information about changes in long term
prospects of the firm which is not conveyed by specific news events
Market seems to view Positive UE as capturing more favorable information
compared to the unfavorable in formation captured by negative
unexpected earnings
Ability of earnings to capture information increases when the sign of UE
and the direction of Intensive IT have similar valuation implications (Max
Buy/UE+ and Max Sell/UE- pairs)
CONCLUSIONS
Inherent uncertainty in the information captured by sign of unexpected earnings appears to
be reduced by confirmatory insider trading and vice versa
IT may be non accounting source of information to explain cross sectional and inter temporal
difference in the marginal response to PE
Annual UE captures only some but all of the info underlying IT
Intensive IT is associated (Max Buy group) is associated with favorable info not fully
captured by that years UE+ (additive effect)
Pattern and magnitude of negative CAR associated with negative UE do not differ
significantly with IT classifications
For all observation with negative UE unfavorable info appears to arrive fairly uniformly over
the holding period
Magnitude of UE for these observations is only weakly associated with CAR
There may be fundamental differences in the ability of IT and accounting earnings to jointly
capture favorable versus un favorable information
THANK YOU
DIFFICULT TO PROVE
First, it can be hard to determine what the accused actually knew at the time
the trades were made. Second, it can be challenging to establish that a
particular individual was responsible for a trade, because knowledgeable
traders can hide behind a variety of proxies and complete their trades over
a number of international markets, many of which do not cooperate with the
authorities. Third, wealthy insiders can afford to retain distinguished
attorneys who can drag out cases at significant cost to the U.S. taxpayer.
Fourth, direct evidence of insider trading is rare. Unless the defendant
confesses or the prosecutor has access to testimony from an eyewitness
whistleblower, cases are almost entirely circumstantial. Fifth, burgeoning
swaps and options markets afford insiders more sophisticated tools for
avoiding detection. Finally, the details of insider trading cases can be difficult
to grasp by non-experts, thereby making it more difficult for prosecutors to
convince juries that an actionable crime has been committed
RECENT CHANGES
Insider trading was not commonly prosecuted until the second half of the
20th century. Between 1966 and 1980, the SEC only filed an average of
2.6 cases per year, while between 1982 and 1986, it filed an average of
17.2 cases per year (Seyhun, 1992). After the spike in insider trading in
the late 1980s, highlighted by the indictments of Boesky and Levine,
there were no insider trading prosecutions of Wall Street professionals by
the SEC between 1990 and 1995, and only ten such prosecutions
between 1995 and 2000 (Thomsen, 2006). Insider trading actions have
been much more prevalent in this first decade of the 21st century. The
SEC had 106 successful insider trading convictions between 2001 and
2006. In the first half of the year 2007 alone, the SEC brought
enforcement actions against over 20 defendants for insider trading, most
of which involved insider trading in advance of mergers and acquisitions
(Gorman, 2007).
LEGAL DEVELOPMENTS
First, the SEC has broadened the liability of employers for employee actions (SEC, 2006).
Second, the agency not only publicly announced (Thomsen, 2006) that it was prioritizing hedge
fund insider trading, but it also proceeded to back up (Thomsen, 2008) this commitment. In
2009, three particularly significant hedge fund-related cases were brought within a three-week
period, including the largest such case in the agencys history (Clark, 2010). This focus on
hedge funds was also evident in Europe (Financial Services authority (FSA), 2005). Third, the
SEC announced that an agreement had been reached among the major securities selfregulatory
organizations to centralize insider trading regulation (SEC, 2008). Prior to this announcement,
each equity exchange was responsible for surveillance of trading on its market and any
investigations and enforcement actions involving its members. This centralization of
surveillance should improve detection of insider trading across the equities markets by focusing
expertise and eliminating gaps and duplication among the markets. Fourth, the SEC announced
a policy designed to encourage individuals to cooperate with the agency in its insider trading
investigations (SEC, 2010). Fifth, the agency tried to expand the scope of insider trading law in
the first case to allege insider trading in credit default swaps (CDS) (Clark, 2010). While the case
has yet to be decided, the court refused to dismiss the case (Weidlich, 2009) on the grounds
that it was an issue of fact, not law, as to whether the CDS met the definition of security under
existing insider trading laws (McGrath, 1993)
IT
if insider trades are informative, then markets are not strong-form efficient. One set of studies examines the
relationship between the information contained in earnings announcements and that revealed by the trades
of insiders. For instance, Penman viewed insider trading as a signal of managements assessments of firms
future prospects, and compares the information content of these trades to that in managements earnings
forecasts (Penman, 1985). He found that the insider trading measures that account for the timing of the
trades relative to the release date of the forecast are informative (See also Allen & Ramanan (1995)). A
similar set of studies analyzes the relationship between the information contained in a variety of other
corporate announcements and that revealed by insider trades. In an examination of corporate sell-offs,
Hirschey and Zaima (1989) found that insider trading and ownership structure information were used by the
market in the classification of sell-off decisions as favorable or unfavorable for investors. Thus, they
concluded that this voluntary announcement was informative. Similar results have been found for the
informational role of corporate dividend announcements (John & Lang, 1991). A final set of studies in this
area concerns the predictive content of insider trades for outsiders who seek to secure abnormal returns by
mimicking these trades. Givoly and Palmon (1985) established that the abnormal returns gained by insiders
could be largely due to price changes arising from the disclosure of the trade itself, rather than to
subsequent disclosure of specific news about the company to which the insiders might be privy. Other
authors have confirmed that outsiders can garner abnormal returns by using publicly reported insider trading
data as a leading indicator (Rozeff & Zaman, 1988). This implies that the mere occurrence of insider trading,
whether or not it is informationbased, may generate abnormal returns. Hence, it is not surprising that many
business magazines report information on individual- and aggregate-level insider trading.
IT
However, Seyhun (1992) provided evidence that the abnormal returns
were actually due to the informativeness of the trades by documenting
that, from 1975 to 1989, aggregate insider trading predicted stock
returns. In particular, the aggregate net number of open market
purchases and sales by insiders in their firms predicted up to 60% of the
variation in one-year-ahead aggregate stock returns (Seyhun, 1992).
Lakonishok and Lee (2001) extended the results through 1995 and
indicated that the results were driven by insiders abilities to predict the
returns of smaller firms. These results have also been found to hold in
the United Kingdom (U.K.), where insider trades are even more
informative than those in the U.S (Fidrmuc, 2006). Adding robustness to
these findings is the fact that trades by insiders in options markets have
also been shown to be information-based (Chen & Zhao, 2005).
IT
I did find one study that surmised that insider trades were not
informative (Chakravarty & McConnell, 1999). However, the
persuasive power of this study is limited by the fact that one of the
co-authors later co-authored a paper that argued in favor of there
being an important informational role for options (Chakravarty,
Gulen, & Mayhew, 2004). Thus, given the results discussed above
from all three sets of informativeness studies, the evidence in
support of the contention that insiders trades are informationbased is quite compelling.
IT
Given that insiders trades are informative, is it also clear that they
are abnormally profitable? This is a key question because if insiders
do not, on average, earn abnormal returns (inflation-adjusted
returns in excess of the return that the average investor could have
expected to earn on a similar trade involving a firm with the same
level of systematic risk) on their trades, then one could offer three
compelling reasons that regulators should not concern themselves
with such activities: (1) if, in expectation, insiders cannot profitably
exploit their information, then they would no longer have an
incentive to engage in such trading; (2) there would be no
increased likelihood of harm
IT
because counterparties to insider trades would not be any more likely to lose money
on their average trades; and (3) the information conveyed by the trades would tend
to improve market efficiency. Studies of abnormal profitability involve the event
study methodology. Therefore, a finding of abnormal profitability is conditional on
the model used by the researcher to calculate expected returns being correctly
specified. Also, for parametric model-based estimators other than Generalized
Method of Moments estimators (that only yield asymptotically valid inferences), the
results are conditional on the correct specification of the data generating process for
the employed test statistics. Consequently, the results of these studies should be
scrutinized for possible violations of the following assumptions: (1) normality of the
prediction errors; (2) contemporaneous correlation in the prediction errors; (3) serial
correlation in the prediction errors; (4) parameter stability in the estimates; (5) no
event-induced volatility changes; (6) homoscedasticity of the prediction errors; and
(6) use of prediction errors, not residuals. Hence, I will now examine whether
insiders are, in fact, able to profit from their information-based trades.