Downside Risk and the Design of CEO Incentives:
Evidence from a Natural Experiment
David De Angelis, Gustavo Grullon, and Sébastien Michenaud*
May 13, 2013
Abstract
This paper examines the causal effects of downside risk on the design of CEO incentive
contracts. Using an experiment that relaxed short-selling constraints for a random sample
of US firms (pilot firms), we find that these firms’ stock prices display a greater sensitivity
to negative market-wide and firm-specific news and that the volatility skew of their put
options (downside risk) increases. Firms in the pilot group respond to this change in
downside risk by increasing the convexity of the compensation payoff of their CEOs and
other top managers. While pilot and control firms exhibit similar equity grant structures
before exposure to the treatment, we find that the proportion of stock options in new
equity grants increases significantly for pilot firms during the experiment and that this
increase reverses immediately after the repeal of the experiment. We also find that the
changes in the structure of new equity grants are related to changes in the downside risk
profile of the pilot firms. In addition, pilot firms also modify non-pecuniary forms of
compensation by adopting new anti-takeover provisions and thus further insuring their top
managers against downside risk. Finally, we find suggestive evidence that the increases in
the use of stock options are related to increases in subsequent investment. Overall, our
results show that protecting managers from downside risk is an important goal in the
design of incentive contracts, and shed light on the implications of stock market regulation
on the design of corporate governance mechanisms.
De Angelis ([email protected]), Grullon ([email protected]), and Michenaud ([email protected]) are at
Rice University. We greatly appreciate the comments of Kerry Back, Alan Crane, François Degeorge, François
Derrien, Laurent Frésard, Yaniv Grinstein, Thomas Hemmer, Ambrus Kecskés, James Weston, and seminar
participants at Rice University. All remaining errors are our own.
*
Equity-based compensation (restricted stocks and stock options) is widely used to
help align CEO’s interests with those of dispersed shareholders of publicly listed firms. Yet,
this type of compensation exposes CEOs to risks that may lie outside of their control. 1 In
this setting, principal-agent theory predicts that firms will trade-off CEO incentives
provision with CEO risk exposure (Holmstrom (1979), and Holmstrom and Milgrom
(1987)). Despite extensive research, however, the empirical evidence with respect to this
prediction is inconclusive and controversial (e.g., Aggarwal and Samwick (1999), Core and
Guay (2001), and Prendergast (2002)). One important difficulty in studying this tradeoff is
that empiricists cannot easily disentangle the effect of compensation on risk from the effect
of risk on compensation.
In this paper, we investigate how one specific form of risk – downside risk –
influences the design of CEOs’ incentives. We address the identification challenge by
exploiting a randomized natural experiment that exogenously increased downside equity
risk through the relaxation of short-selling constraints. Because the removal of short-
selling constraints may cause an increase – or the fear of an increase - in bear raids and
market manipulation by short-sellers (Goldstein and Guembel (2008)), this increase in
downside risk potentially exposes managers to losses that are beyond their control. In this
scenario, CEOs may sub-optimally reduce the risk of their firms to protect their personal
wealth and firm-specific human capital (Amihud and Lev (1981), May (1995)). Consistent
with this view, firms and their CEOs display an acute aversion to short-sellers, and go to
great lengths to fight them and reduce their influence on stock prices (Lamont (2012)). As
1 Even though managers arguably have control over the operational risk of their firms, they may have little
control over some of their firm’s stock price risk. As a result, equity-based compensation is expected to be
more costly to shareholders in the presence of increased idiosyncratic risk (Aggarwal and Samwick, 1999), or
if CEOs are more risk-averse (Becker, 2006).
2
a result, firms that maximize shareholder value should respond to an exogenous increase in
short selling activity by increasing their CEOs’ risk-taking incentives to avoid sub-optimal
risk reduction policies, and/or by immunizing their CEOs against the downside risk that
lies outside of their control and does not reflect their performance.
Consistent with the notion that firms take downside risk into consideration when
designing CEO incentive compensation contracts, we find that the firms affected by this
exogenous shock include relatively more stock options in the compensation packages of
their CEO and other top managers. Furthermore, we find that these firms adopted other
pecuniary and non-pecuniary forms of compensation (severance packages, anti-takeover
provisions) to protect their CEOs from the increase in downside risk. Overall, our evidence
reveals that there is a causal effect of downside risk on the design of CEO incentives.
Our experiment is based on the SEC’s approval of Regulation SHO (Reg SHO) in 2004,
which removed the “uptick rule” for a randomly selected sample of firms (pilot firms).
Since the “uptick rule” prevents investors from short selling stocks when prices decline, the
firms selected for the Reg SHO experiment became more susceptible to downside risk. 2 As
documented by Grullon, Michenaud, and Weston (2011), the increase in short-selling
activity after the announcement of Reg SHO led to an increase in the sensitivity of stock
returns to negative news. Consistent with these results, we find that firms in the pilot
group exhibit more negative returns on bad-market days, become more sensitive to large
negative earnings surprises, and display an increase in the volatility-skew 3 of puts on their
2 Rule 10a-1 of the Exchange Act (1938), the “uptick rule”, only allowed short sales on plus ticks or zero plus
ticks on the NYSE, while NASD Rule 3350 (1994) prohibited short sales below the bid if the last bid was a
down bid on NASDAQ.
3 We define volatility-skew as the difference between the implied volatility of out-of-the-money put (call)
options and that of in-the-money put (call) options. This measure or variants of it have been shown to proxy
3
stocks, suggesting that investors anticipate large negative jumps in price levels. Moreover,
this shock to equity risk appears to be asymmetric: there are no significant differences in
stock price reactions between the two groups for large positive news, and no increase in
the volatility-skew of calls on the stocks. Taken together, these findings suggest that the
volatility of pilot firms stock prices has only increased on the downside. Hence, we use the
Reg SHO experiment to investigate whether an exogenous shock to the downside risk
affects CEO compensation.
Since granting more stock options relative to restricted stocks increases the
convexity of CEOs’ compensation payoff and provides increased protection on the
downside (stock-options granted at the money are less affected by increases in downside
risk than restricted stocks), one would expect firms experiencing an exogenous increase in
downside risk to use relatively more stock options in their compensation packages. One
potential reason for this is that the increase in downside risk leads to an increase in the
relative cost of granting stocks as risk-averse managers demand a premium for the
exposure to downside risk. Using a difference-in-differences approach, we find evidence
consistent with this prediction. In particular, we find that the shock to downside risk does
not affect the total value of equity grants awarded by the firm to its CEO, but it affects the
structure of the equity grants, which consist of stock options and restricted stocks. Firms
in the pilot group respond to the announcement of Reg SHO by increasing the proportion of
stock options grants in the new equity grants awarded to CEOs by 7% to 8%.
Additional difference-in-differences tests show that the difference in the structure of
new equity grants between pilot and control firms persists over the 2-year period following
for large expected negative jumps in individual stocks (Xing, Zhang and Zhao (2010)) and in indices (Bollen
and Whaley (2004), Bates (2003), Gârleanu, Pedersen, and Poteshman (2007)).
4
the announcement and the implementation of the experiment. The difference disappears
immediately following the repeal of the uptick rule on all US stock markets in 2007. 5
Furthermore, we also observe that the change in the structure of new equity grants is
significantly larger for pilot firms that exhibit the largest increase in their sensitivity to
negative news around the announcement date of Reg SHO. This finding suggests that the
increase in downside risk is the primary driver of our main results. In addition, we find
that this change in the structure of new equity grants extends to other top executives of the
firm. We also find evidence that pilot firms further protect CEOs from downside risk by
adopting new anti-takeover provisions such as staggered boards, and supermajority rules,
and by providing severance packages.
Finally, we investigate the interaction between the design of CEO incentives and
investment policies.
While Grullon et al. (2011) find that pilot firms reduce their
investment activity after the adoption of Reg SHO, we find evidence that the provision of
risk-taking incentives via stock options grants potentially mitigates this effect. Specifically,
we find that the pilot firms that responded the most to changes in downside equity risk by
increasing stock option grants experience the largest increase in capital expenditures and
research and development expenses.
Although these results shed light on the potential
real effects of CEO incentive contracts, we cannot rule out that firms provide more risk-
taking incentives via stock options because they have more investment opportunities, and
thus we are cautious not to draw any causality inferences from this analysis.
Our results are related to several predictions from principal-agent theories. First,
our findings are overall consistent with the trade-off between risk and incentives
5
We stop our analysis before the financial crisis to avoid any confounding effect related to this event.
5
(Holmstrom (1979)). By changing the structure of new equity grants and insuring their
managers against the adverse effects associated with the increased probability of hostile
takeovers and dismissals, firms reduce the amount of risk borne by their managers and
thus the expected compensation costs. Also related to our results, the model proposed by
Hemmer, Kim, and Verrecchia (2000) shows that the convexity in the compensation payoff
will be related to the skewness of the price distribution, arguably a measure of downside
risk. Second, our evidence is consistent with the view that options potentially induce more
risk-taking incentives (Jensen and Meckling (1976)). 6 Risk-averse managers may sub-
optimally lower firm risk when exposed to more risk they cannot control, i.e. stock price
risk. By providing more risk-taking incentives in managerial contracts, firms may be able
to offset this adverse effect. Finally, our results are also related to a recent contracting
model proposed by Dittmann, Maug, and Spalt (2010) who show that the presence of
options in an optimal contract can be justified by CEOs’ loss-aversion. 7 To the extent that
downside risk is observationally equivalent to loss-aversion, our results would be
consistent with their arguments.
We perform a number of robustness tests. Given the randomized nature of our
experimental setting, endogeneity should not be an issue because firms cannot have caused
their inclusion in the pilot program. Nevertheless, we examine whether our findings are
the result of chance. To evaluate this possibility, we randomize inclusion of firms in the
pilot group and bootstrap an empirical distribution of our main results. Out of 5,000
This view is controversial. Ross (2004) shows that a convex compensation payoff does not necessarily
induce greater risk-taking incentives. In particular, it depends on the type of the agent’s utility function. See
also Carpenter (2000).
7 Using structural estimation of a standard principal-agent model, Dittmann and Maug (2007) finds that it is
difficult to explain the presence of stock-options in the compensation contract.
6
6
simulations, we do not find a single instance in which all our main variables experience
statistically significant changes. Thus, it is unlikely that our results are generated by
methodology choices or sample selection.
Furthermore, we test alternative channels that could explain our main findings. The
first alternative channel is related to a change in stock prices. For example, Grullon et al.
(2011) find that firms in the pilot, especially small firms, experience price declines after the
announcement of Reg SHO. Thus, our results could be driven by firms simply reloading
managers’ incentives after these price declines. First, in our sample of large firms, we
neither observe any significant and persistent effect on stock price nor on call option price.
Second, we show that firms that exhibit large negative announcement returns around the
announcement date do not drive our main results, thus confirming that the effect is not
coming from a decrease in stock prices. We also re-run our entire analysis using the
number of options and stocks (instead of their grant value) to verify that our results are not
mechanically driven by changes in stock and option prices.
Another potential channel is related to a change in the informativeness of stock
prices. Incorporation of negative information into stock prices may have improved for the
pilot firms as a result of the removal of short-sales constraints (see Holmstrom and Tirole
(1993)). 8 Nevertheless, if firms were changing CEO incentives contracts to take advantage
of the negative information impounded into stock prices, they should use more restricted
stocks, which expose managers to negative stock price reactions, and fewer stock options,
which insulate managers from negative outcomes. Therefore, we believe that our results
are unlikely to be primarily driven by an increase in the informativeness of stock prices.
Consistent with that argument, the results in Karpoff and Lou (2010) suggest that short sellers detect firms
that misrepresent their financial statements and thus help to improve price efficiency.
8
7
Our paper makes a number of contributions to the literature. First, we provide
causal evidence that risk is an important determinant of CEO incentives design. As noted
earlier, identification of a causal relationship between incentives and risk is problematic
due to the fundamental endogeneity between these two variables.
While incentive
contracts may be the outcome of firm’s risk environment, it is also possible that managers
may change firm risk because of the incentive contracts in place. In addition, the variables
typically used to proxy for firm risk are prone to controversial causal interpretations. For
instance, firm risk is usually proxied by the volatility of stock returns. Measures based on
stock prices are subject to reverse causality interpretations because stock prices may
incorporate information about unobserved variables such as investment opportunities or
managerial skills. Not surprisingly, the empirical evidence using this measure is mixed.
Aggarwal and Samwick (1999) find a negative relationship between firm risk and CEO
incentives whereas Guay (1999) and Core and Guay (2001) argue that the relationship is
positive and is due to reverse causality. Prendergast (2002) summarizes findings on this
issue and finds that the evidence inconclusive. More recently, Cuñat and Guadalupe (2009)
find that changes in operational risk, measured by exogenous shocks to competition,
influence CEO compensation design.
Second, our paper contributes to the literature on CEO incentives by providing
evidence that boards move quickly to readjust CEO incentives following an exogenous
shock to the environment of the firm. Core and Guay (1999) find that firms often readjust
CEO incentives in response to deviations from the “optimal” incentive package. In contrast,
Gormley, Matsa, and Milbourn (2012) find that boards move slowly to adjust CEO
incentives in response to exogenous shocks to liability risk. Our results complement the
8
ones in Hayes, Lemmon, and Qiu (2012), who show that firms readjust compensation
packages after the adoption of FAS 123R, which changed the accounting benefits of
granting stock options. In our paper, Reg SHO creates an economic cost to granting
restricted stock (relative to granting stock options), leading firms to readjust the structure
of their new equity grants.
Finally, our paper contributes to the literature that links stock prices to corporate
decisions. For instance, Chen, Goldstein and Jiang (2007) and Grullon et al. (2011) show
that the stock market influences real investment. Our study complements their results by
uncovering the importance of stock markets in the design of CEO incentives and corporate
governance mechanisms, as was first suggested in Holmstrom and Tirole (1993).
The remainder of the paper is organized as follows. Section I discusses the Reg SHO
experiment and our data. Section II discusses our identification strategy and the impact of
Reg SHO on downside equity risk. Section III analyzes how firms adjust CEO incentives in
response to an unanticipated change in downside equity risk. In Section III, we also
provide additional findings on how these changes in compensation are related to changes
in firm behavior. Section IV analyzes the robustness of our results. Section V concludes.
I. Sample, Data, and Variable Definitions
On July 28, 2004, the SEC announced the removal of restrictions on short sales for a
randomly selected sample from the Russell 3000 index.
The list of pilot firms was
approved by the SEC board at least a month earlier on June 23, 2004. The SEC selected
firms from the Russell 3000 index listed on NYSE, NASDAQ and AMEX and ranked them
separately for each stock exchange by average daily traded volume. In each stock market,
9
the SEC would then take 3 stocks and pick only the second one to be part of the pilot study.
It would then repeat the process by moving down the rankings to ensure representation of
the three stock markets, and to get consistent average trading volume between pilot and
control firms in each stock market. The objective of the pilot study was to test the impact
of removing short sales restrictions induced by the price tests on stock market volatility,
liquidity, and price efficiency. Figure 1 provides a detailed timeline of the experiment. 9, 10
{Insert Figure 1 here}
We construct the main dataset from the Center for Research on Security Prices
(CRSP).
We build the Russell 3000 index based on the rankings of stock market
capitalizations as of May 28, 2004 and May 31, 2005. 11 We follow Diether, Lee and Werner
(2009) who keep firms that were in the Russell 3000 index in 2004 and 2005 and eliminate
firms that are deleted from the index due to acquisitions, mergers or bankruptcies during
the year. We merge this list with the list of pilot securities announced on July 28, 2004 by
the SEC. Out of the 968 pilot securities in the initial list, 946 pilot securities remain in the
sample after the first filter.
Merging with Compustat, Execucomp, Risk Metrics, and
excluding banks and financial firms leaves 1,442 firms (935 control / 507 pilot). Our final
sample is an unbalanced panel of 4,036 firm-year observations. We define all variables
used in the paper in Appendix 2.
The Securities Exchange Act Release No 48709A first announced on October 28, 2003 the SEC’s intention to
run the experiment and requested external comments. The Securities Exchange Act Release No 50104 on July
28, 2004 announced the final design of the experiment, the list of all firms in the pilot group, the group of
firms for which all price tests were suspended.
10 Rule 202T (the pilot program) was part of Reg SHO, which aimed at testing a broader set of rules. Both
rules were announced on July 28, 2004, and adopted on August 6, 2004 (Release No 34-50103). Reg SHO
included provisions concerning location and delivery of short sales (Rule 203) to reduce naked short selling,
and new marking requirements for equity sales (Rules 200 and 201.)
11 Consistent with the definition of the Russell 3000 at the reconstitution date, we exclude stocks with prices
below $1, pink sheet and bulletin board stocks, closed-end mutual funds, limited partnerships, royalty trusts,
foreign stocks and American Depositary Receipts (ADRs).
9
10
Table I provides summary statistics for all the firms in the sample, with a
breakdown between pilot and control firms. We find no differences between the two
groups, suggesting that our filtering process has not created any obvious sample selection
bias to the random selection by the SEC. Both groups of firms have about the same size,
compensation levels, equity grants structure, governance quality, corporate spending,
payout, and capital structure. None of the differences in characteristics are statistically
significant. Therefore, the data support the hypothesis that our pilot group firms represent
a random draw from our overall sample.
{Insert Table I here}
II. Regulation SHO, Short Interest, and Downside Risk
In this section we examine the impact of Reg SHO on short-selling activity and on
the sensitivity to realized and anticipated negative news to show that the randomized
natural experiment represents a shock to the downside equity risk faced by CEOs with
equity based incentive contracts.
We follow the methodology in Grullon, Michenaud and Weston (2011) who focus on
event windows around the announcement date. The authors argue that under rational
expectations, investors should incorporate the future impact of the change in short sales
regulation at the time of the announcement. Theory by Allen, Morris, and Postlewaite
(1993) suggests that the announcement of Reg SHO should have the same effect on shortselling activity as the immediate removal of short-selling constraints. 12, 13 Investors should
12 Allen, Morris, and Postlewaite (1993) show that stock price bubbles may arise if investors face short sale
constraints either now or in the future, in spite of all agents being rational and fully informed about future
dividends. In their model, the belief that investors will be able to sell the stock at a high price in the future
causes the bubble. In this setting, the announcement of the removal of short-selling constraints in the future
11
find it profitable to sell short the Reg SHO pilot stocks as long as the benefits from doing so
are larger than the costs. In addition, the Reg SHO experiment could increase short-selling
activity around the announcement date because of the increased incentives of bear raiders
to manipulate the value of those firms that are easier to sell short (Goldstein and Guembel
(2008)).
A.
Short Selling Activity
The SEC’s Office of Economic Analysis (OEA, 2007), Alexander and Peterson (2008),
Diether, Lee, and Werner (2009) document an increase in short sales after the
implementation of the pilot experiment on May 2, 2005. In this paper, we replicate the
results in Grullon et al. (2011), who find that short sales increase around the
announcement of the pilot program.
As argued earlier, short sellers may anticipate an effect of the suspension of price
tests on firms in the pilot group. If this is the case, then they should increase short sales on
these stocks after the disclosure of the list of pilot firms on July 28, 2004. We test this
hypothesis by running a difference-in-differences analysis using time-series of Short
Interest from the monthly short interest reported by NASDAQ and NYSE. Short Interest is
the monthly short interest as a percentage of previous calendar month shares outstanding
(from CRSP) over the period 2001-2007. 14 NASDAQ and NYSE report the number of all
open short positions on the last business day on or before the 15th of each calendar month.
should immediately reduce current stock prices: investors realize that they will not be able to sell the stocks
at inflated prices to other investors in the future.
13 Scheinkman and Xiong (2003) show that stock prices should incorporate the option value of reselling to
optimistic investors in the presence of short-selling constraints. The expected removal of short-selling
constraints should therefore lead to an increase in short selling activity after the announcement.
14 Our difference-in-differences methodology requires data for the period preceding the experiment (20052007). Therefore, we cannot use actual short sales data that are only available for the period of the
experiment. We use Short Interest and Abnormal Short Interest as proxy variables for short sales.
12
We build a proxy for the unexpected component of short interest, Abnormal Monthly Short
Interest, by computing the residual of a firm fixed effect regression in which Short Interest
is regressed on month dummies, market-to-book, lagged total assets, logarithm of lagged
return on assets, trading volume, and a dummy variable for listing on the NYSE.
Table II presents the average Short Interest and Abnormal Monthly Short Interest for
a period of three years before and after the announcement of the pilot test on July 28, 2004.
We stop the analysis on July 7, 2007 when price tests are suspended for all US stocks. We
find that both Short Interest and Abnormal Monthly Short Interest increase more for firms in
the pilot group than firms in the control group in a difference-in-differences analysis. We
compare the difference in these two measures of short interest between the pilot and
control groups and the difference from before to after the announcement. When we
examine the effect of Reg SHO on Short Interest and Abnormal Short Interest, we find that
the difference-in-differences is +0.37% and +0.29% respectively.
These changes are
statistically significant and represent a relative increase of about 8% of the average
monthly Short Interest or about 9% of the standard deviation of Abnormal Short Interest.
Overall, our findings confirm that short-selling activity increases for the firms in the pilot
group around the announcement of Reg SHO.
B.
{Insert Table II here}
Sensitivity to Negative News
We now test whether stock prices for the firms in the pilot group become more
sensitive to bad news. If the removal of short selling constraints increases the trading
activity of pessimistic investors in the stock market, then stock prices of firms in the pilot
group should become more sensitive to realized or anticipated bad news after the
13
announcement of the Reg SHO experiment relative to before. First, we measure the daily
returns of pilot and control group firms during bearish stock market days and around
negative (positive) earnings announcements to test whether the firms in the pilot group
become more sensitive to negative information. Second, we also test that options markets
anticipate the effects of the removal of short-sales constraints in the stock price behavior of
pilot vs. control stocks in case of negative news. The objective of these tests is to provide
evidence that Reg SHO generates an asymmetric shock to stock price risk. By becoming
more sensitive to negative news, we argue that stocks become more risky on the downside,
a feature that will expose stock and put option investors to more risk, but will expose call
stock option holders much less so to this risk due to the asymmetric payoff of call options
on the stock. These results are central to our identification strategy and provide the
foundations for using Reg SHO as a reduced-form instrument for increased downside
equity risk.
We first test firms’ stock price reactions to bad market-wide news. We resort to
difference-in-differences analyses in which we sort daily market-wide returns into five
quintiles to test whether the returns of firms in the pilot group become more negative in
very bad market days (first quintile of market returns) after the announcement of the pilot
program than before relative to the control group.
{Insert Table III here}
Panel A of Table III presents the results of this analysis. The two groups of firms do
not display different returns on bad market days before the announcement of Reg SHO.
However, firms in the pilot group have more negative returns than the control firms after
14
the announcement during the worst market days (lowest quintile). The difference-indifferences coefficient is statistically significant at the 1% level.
Second, we measure changes in the sensitivity of pilot stock returns to firm-specific
news. We test for differences in stock returns after large negative and large positive
earnings news using earnings surprises relative to the I/B/E/S quarterly consensus analyst
forecasts. We report the results of this analysis in Panels B and C of Table III. On average,
firms in the pilot group do not show any significant differences before the announcement of
Reg SHO relative to firms in the control group. After the announcement of Reg SHO, firms
in the pilot group have significantly larger negative CARs when reporting large negative
earnings news than the firms in the control group. Importantly for our identification
strategy, we do not find any increase in the stock price reaction to large positive earnings
news for the pilot firms after the announcement of Reg SHO (Panel C of Table III).
Finally, we measure changes in the volatility skew of put and call options on the
stocks of pilot and control firms. We follow standard practices and define volatility skew as
the difference between the implied volatility of out of the money stock options (strike price
to stock price ratio is less than .9 and more than .7) and at the stock options (strike price to
stock price ratio is less than 1.05 and more than .95) (see Xing, Zhang and Zhao (2010)).
Our estimation window represents the two-month period before and after July 28, 2004
(i.e. the Reg SHO announcement). 16 The volatility-skew of puts capture the anticipation of
large negative jumps in price levels. As illustrated in Figure 2, we observe that the volatility
skew of put options is similar across both groups of firms before the experiment while it
Due to data limitations, we use a restricted subsample of firms that have options traded on options market
with a large enough trading volume. Only 490 such firms (pilot and control) meet our requirements, thus
resulting in a sample that is about 1/3 of the size of our original sample.
16
15
increases after the announcement for firms in the pilot group relative to the ones in the
control group. In addition, the statistical tests in Panel D of Table III show that the increase
in the volatility skew of the puts is significant. We also perform the same exercise using call
options (see Figure 2 and Panel E of Table III) and find no significant change in the
difference of volatility skew between the two groups. These results confirm that the change
in the risk profile of the firm is asymmetric: only the downside component of risk is
affected by the reduction in short selling constraints.
{Insert Figure 2 here}
In general, all our results point to a significant increase in downside risk for the
firms in the pilot group. Since this increase in the sensitivity of stock returns to negative
news represents a shock to CEO exposure to equity risk when the CEO has equity-based
incentive contracts, we use Reg SHO as an exogenous shock to the downside equity risk
faced by the CEO.
III. The Effects of Downside Equity Risk on the Design of CEO Incentives
We now move to the analysis of the impact of this exogenous shock to downside
equity risk on the design of CEO incentives. We first look at the changes in the structure of
new equity grants around the announcement of Reg SHO. We then investigate whether
firms change their governance structure around this regulatory change.
A.
The Structure of New Equity Grants awarded to the CEO
Our first set of tests examines whether the structure of the new equity grants
awarded to the CEO changes around the removal of short selling constraints. Since Reg
SHO creates a shock to downside equity risk, we investigate the effects of this shock on the
16
convexity of the new compensation package. Following the existing literature, we use stock
options awards to capture the convexity of the compensation payoff (see, e.g., Hayes et al
(2012)). Guay (1999) uses vega as a measure of the convexity of the compensation payoff
and shows that the vega associated with stock options is considerably larger than the vega
associated to restricted stock. 17 As a result, subsequent studies such as Knopf et al (2002)
and Coles et al (2006) approximate the total vega of CEOs’ stock and option portfolios by
the vega of their option portfolio.
In this paper, we study the change of convexity in the compensation contract by
examining the trade-off between awarding stock options and restricted stock in new CEO
equity grants. Everything else equals, granting more stock options relative to restricted
stock in new equity grants will lead to higher convexity in the compensation payoff. Our
main measure of interest is the portion of options in new equity awards (i.e. the sum of
option and stock awards). 18
One alternative approach to study the change of convexity in CEO incentives would
be to compute the vega of the CEO’s equity portfolio. However, the computation of the
portfolio vega relies on the stock-return distribution of the underlying stock. Hence, even
without any change in compensation practices, there would be a mechanical change in the
vega since Reg SHO impacts the return distribution of the underlying stock. As a
consequence, this would not be a reliable measure in our empirical setting.
Vega captures the sensitivity of a change in dollar value of a financial claim as a function of a change in
annualized standard deviation of stock returns.
18 This measure is similar to the one employed in Kadan and Swinkels (2008).
17
17
A.1.
The Structure of New Equity Grants in the 2001-2007 period
In Figure 3 we first compare the evolution of the structure of CEO equity grants for
firms in the pilot group and in the control group over time. Panel A plots the average ratio
of the value of stock options granted to the total value of equity grants between 2001 and
2007. The proportion of stock options in new CEO equity grants decreases over the entire
period for both pilot and control firms. Before the start of the experiment, the difference in
the structure of new equity grants between the two groups is very small. The difference (in
dollars) ranges between -2.3% and 0% before the experiment (see Panel C), increases to
+4.5% during the experiment, and goes back to 0.7% when the uptick rule is repealed for
all US firms.
{Insert Figure 3 here}
We also study the number of stock options and restricted stock to verify that our
results are not mechanically driven by a relative change in the stock price of pilot firms
relative to control firms. This analysis is useful in confirming that we indeed capture a
change in contracting behavior. Panel B plots the average ratio of the number of stock
options granted to the total number of stock options and shares of restricted stock granted
to the CEO over the same period. Consistent with the previous analysis, we find that before
the experiment, the difference of the new equity grant structure ranges between -1% and
0.4% (see Panel C). This difference increases during the experiment to reach +3.3% in 2004
and +4.3% in 2006, while it decreases to +2.2% after the repeal of the uptick rule for all US
firms in 2007.
In Panel D we plot the difference-in-differences of the structure of new CEO equity
grants between pilot firms and control firms over the same period. The difference-in18
differences coefficient (DiD) measures the change in the difference of the ratio of stock
options granted to total equity grants (in value and in number of shares) between pilot and
control firms from year t-1 to year t. The graphs show that there are almost no changes in
the difference of the structure of new equity grants between the two groups during all the
years covered except in 2005 - the year following the announcement of Reg SHO - and in
2007 – the year of the repeal of the uptick rule for all US stocks. In 2005, the DiD is +5.7%
(Option/Equity($))
and
+4.3%
(Option/Equity(#)).
-3.8% (Option/Equity($)) and -2.2% (Option/Equity(#)).
In
2007,
the
DiD
is
These results suggest that the increase in downside equity risk associated with the
implementation of Reg SHO causes pilot firms to use more stock options in their new CEO
equity grants, and this leads to an increase in the convexity of the CEOs’ compensation
payoffs.
A.2.
Difference-in-Differences Analysis
Our empirical strategy relies on the exogenous shock created by the announcement
on July 28, 2004 of the list of firms in the pilot experiment implemented in 2005. We thus
employ a difference-in-differences technique to gauge the effect of the treatment (e.g. Reg
SHO) on the affected group (e.g. pilot firms). The sample period is from June 2002 to May
2007. The treatment years are fiscal year 2005 and 2006 (so unaffected years are fiscal
year 2003 and 2004). Indeed, firms in Compustat with a 2005 fiscal year have a fiscal year
start date between June 1, 2004 and May, 31 2005. Therefore, considering that equity
grants are in general decided at the beginning of the fiscal year (Lie (2005)), we implicitly
assume that firms decisions regarding the structure of new equity grants occur either
immediately following the announcement date of Reg SHO (July, 28 2004), or up to 12
19
months after the announcement date. 19 We will consider other timing classification in the
robustness tests section and reach similar conclusions. The dependent variable is the ratio
of the value of stock options granted to the CEO to the total value of equity grants
(Option/Equity ($)). Panel A of Table IV shows results for OLS, fixed-effect and Tobit
regressions (left censored at 0 and right censored at 1).
{Insert Table IV here}
In those regressions, the coefficient of Pilot (dummy variable equal to one if the firm
is in the Pilot Group of Reg SHO) is not significant. This confirms that there is no pre-
treatment effect for pilot firms, and that pilot and control firms exhibit similar equity grant
structures before exposure to the treatment. The coefficient of Treatment Years is negative
and significant, suggesting a negative trend in the use of stock options in new CEO equity
grants. Firms use fewer stock options across the board due to changes in the expensing
and regulation of stock options in CEO compensation (Hayes, Lemmon and Qiu (2012)).
Finally, our coefficient of interest, Treatment Years*Pilot, is positive and significant. This
coefficient indicates that the pilot firms include more stock options in their new CEO equity
grants during the experiment than the control firms. We reach similar conclusions using
our alternative regression specifications. 20 These results are consistent with our graphical
analysis in Figure 3 and suggest that Reg SHO causes pilot firms to use more stock options
in new CEO equity grants.
See, for instance, Core and Guay (1999). In their empirical framework, they assume that the design of
executive incentives is decided at the beginning of the fiscal year.
20 As exposed in Puhani (2012), the interacted term Treatment Years*Pilot in the Tobit regression correctly
identifies the sign of the treatment effect in a difference-in-differences model, even though Tobit is a nonlinear model.
19
20
The economic magnitude of our results is large. The point estimates from the first
column in Panel A suggest that the change in the proportion of stock options in new equity
grants increases by 5.95 percentage points during the treatment years. This represents an
increase of 7.66% relative to the ex-ante mean proportion of stock options in new equity
grants (i.e. in 2003 and 2004 – during the control period before the Reg SHO experiment),
or a 17.79% increase relative to the ex-ante standard deviation of the variable.
We also replicate our analysis using the ratio of the number of stock options granted
to the total number of stock options and shares of restricted stock granted to the CEO
(Option/Equity (#)) as a dependent variable. We find similar results, thus confirming that
we capture a change in contracting behavior that is not driven by changes in stock prices.
In Panel B of Table IV, we extend the sample period by including fiscal years 2001,
2002 and 2007. We create dummy variables for each fiscal year separately and interact
these with our pilot dummy to precisely identify when changes in the equity grant
structure occur. Consistent with the previous analyses, we find that the difference in the
equity grant structure between pilot and control firms is only significant in 2005 and 2006.
These results confirm that there is no pre-treatment effect (i.e. both groups are similar
before the experiment), that pilot firms use more stock options during the treatment
period, and that this difference disappears at the end of the experiment around the time of
the repeal of the uptick rule for all US stocks.
The economic magnitude of these results is similar to the one measured in Panel A.
Using the point estimates from the first column in Panel B, the change in the proportion of
stock options in new equity grants increases by 5.69 percentage points in 2005. This
represents an increase of 7.62% relative to the ex-ante mean proportion of stock options in
21
new equity grants (i.e. in 2004 – the benchmark year in this regression), or a 16.35%
increase relative to the ex-ante standard deviation of the variable.
A.3.
Difference-in-Difference-in-Differences Analysis
We use a difference-in-difference-in-differences technique to explore whether our
results are more pronounced for pilot firms that exhibit larger changes in their sensitivity
to negative news (i.e. downside risk). For that purpose, we create a dummy variable equal
to 1 if the firm is in the top quintile of changes in stock price returns sensitivity to negative
market returns around the announcement date (High Downside Risk). We measure changes
in stock price returns sensitivity to negative market returns as changes in firms’ stock
returns when the daily stock market returns fall into the lowest quintile of stock market
return days (as shown in Table III). The change is measured over a one-year period before
and after the Reg SHO announcement date. The results are reported in Table V.
{Insert Table V here}
The coefficient for High Downside Risk*Treatment Years*Pilot is positive and
significant in all specifications. Changes in the structure of new equity grants are more
pronounced for the pilot firms with the largest increases in the sensitivity of their stock
prices to negative market-wide news. This result suggests that changes in downside equity
risk are driving the effects on the changes in the structure of new CEO option grants.
B.
The Structure of New Equity Grants awarded to all Firm Executives
We also investigate the change in the structure of new equity grants awarded to all
top executives present in the Execucomp database. In addition to using OLS, firm fixed-
effect and Tobit specifications, we also use an executive fixed–effect specification. The
results are reported in Table VI.
22
{Insert Table VI here}
The results are similar to the ones regarding the CEO. In all regression
specifications, we find a significant increase in the proportion of stock options in new
equity grants for the Pilot firms relative to the Control firms (Panel A). In addition, when
extending the sample period and including dummy variables for each fiscal year, we find
that the difference in the structure of new equity grants is only significant in 2005 and
2006, i.e. during the experiment (Panel B). Also consistent with the results for the CEO
equity awards, the coefficient of the interaction of the Pilot dummy and the 2007 fiscal year
dummy term is not significant. This last result confirms that the difference in the structure
of new equity grants disappears at the end of the experiment.
C.
Additional Results regarding the Design of CEO incentives
We also study changes in other pecuniary and non-pecuniary forms of incentives in
response to the implementation of Reg SHO. More precisely, we investigate changes in the
provision of severance package and in anti-takeover provisions. We examine three specific
anti-takeover provisions: if the board of the company is classified (cboard), if the firm has a
blank check preferred provision (blankcheck), and if the firm requires supermajority to
approve a merger (supermajor). We employ logit regressions and report the results in
Table VII.
{Insert Table VII here}
The coefficient for Treatment Years*Pilot is positive for all provisions, although only
significant at the usual significance level for classified board and blank check. Lower power
is expected given that we only have one observation per firm every other year. These
results suggest that firms insure CEOs against the adverse effects associated with increased
23
probability of hostile takeovers and dismissal due to the increase in downside equity risk. 21
These results also complement the results related to the changes in the structure of new
CEO equity grants and confirm that firms react to a change in the firm’s risk environment
by redesigning CEO incentives.
D.
New Incentive Contracts and Investment Outcomes
In this section, we investigate the interaction between the design of CEO incentives
and investment policies. We explore whether pilot firms that change the structure of their
equity grants the most also tend to invest more. The motivation for this test comes from
Grullon et al (2011) who find that pilot firms exhibit a large decrease in their investment
following Reg SHO.
To proxy for firms that exhibit a large change in grant structure, we create a dummy
variable equal to 1 if the increase in Option/Equity ($) from the 2003-2004 to the 2005-
2006 period falls in the top decile of the sample distribution (High Equity Change). For this
part of the analysis, the sample firms are restricted to non-utilities firms in the Pilot group.
We use two different measures of investment: one based on capital expenditure (CAPX) and
another one including capital expenditure and research and development expenses
(CAPX+R&D). The results are reported in Table VIII.
{Insert Table VIII here}
The coefficient for Treatment Years* High Equity Change is positive and significant
for both specifications. In other words, pilot firms that responded the most to changes in
downside equity risk by increasing stock option grants also increase investment in capital
21 One other way to further insure CEO pay would be to simply increase base salary. We explore that venue
and do not find any significant change in the difference of base salary between both groups. Tax-deductibilityrelated reasons (e.g. Internal Revenue Code Section 162(m)) might significantly affect firm incentives to
increase base salary and thus might explain this non-result.
24
expenditures and research and development expenses the most. These results provide
suggestive evidence of the interplay between the design of CEO incentives and investment
outcomes.
IV. Robustness Analysis
We first run placebo regressions to check the validity of our results. The results are
reported in Table IX. The sample period is fiscal year 2001 to 2004. The placebo treatment
years are 2003 and 2004. Confirming that our results are not spurious, we find that the
coefficient of Placebo Treatment Years*Pilot is not significant.
{Insert Table IX here}
We perform additional robustness tests that we report in Table X. We first examine
whether our results are robust to a different classification of the treatment period. In our
empirical framework, we assume that the decision regarding the structure of the equity
awards is made at the beginning of the fiscal year (see, e.g., Core and Guay, 1999). Yet,
since the Reg SHO experiment was announced on July 28, 2004, it is possible that some
firms already re-contracted in fiscal year 2004 if the design of CEO incentives contract
occurs at the end of the fiscal year. This potential measurement error would reduce our
ability to find a significant effect of the regulation or reduce the economic magnitude of the
impact of Reg SHO on the change in the equity grant structure.
To address this concern, we re-run our main regressions using only firms with
fiscal-year month ending after the month of July (Panel A). We also exclude fiscal year
2004 (Panel B). In both specifications, we find similar results to the ones presented in our
main analysis. In addition, the point estimates in Panel A are greater than in our main
25
regressions, confirming that the potential measurement error would work against us
finding a significant effect. It is therefore unlikely that a timing mismatch affects our
conclusions.
{Insert Table X here}
An alternative channel that can explain our results is related to a change in stock
price. Since stock prices of firms in the pilot might be negatively affected by the experiment
(Grullon et al (2011)), it is possible that the pilot firms could be simply reloading managers’
incentives. We test this alternative explanation by examining whether the firms that
exhibit a large negative announcement returns around the announcement date (i.e. firms
more impacted by a change in stock price – variable Low CAR) also exhibit a larger change
in the structure of new equity grants. The results are reported in Panel C. The coefficient
for LowCAR*TreatmentYears*Pilot has the wrong sign and is not statistically significant,
suggesting that a large drop in stock prices is not the driving force behind our results. In
addition to this test, we do not find a large persistent effect on stock prices in our sample.22
We conclude that this alternative channel is unlikely to drive our results.
Another potential channel is related to a change in the informativeness of stock
prices. Incorporation of negative information into stock prices may have improved for pilot
firms as a result of the removal of short-sales constraints (see Holmstrom and Tirole
(1993) for a model of market monitoring). However, if firms were changing CEO incentives
contracts to take advantage of the negative information impounded into stock prices, they
should use more restricted stock and less stock options, which insulate managers from
22
Results are not reported but available upon request.
26
negative outcomes. As a consequence, our results are unlikely to be primarily driven by an
increase in the informativeness of stock prices.
Our final robustness test is related to the randomized nature of our experimental
framework. As mentioned earlier, endogeneity is unlikely to be an issue since firms cannot
possibly have caused their inclusion in the pilot program. Yet, we test whether our results
could have been the result of chance. We randomize inclusion of firms in the pilot group
and bootstrap an empirical distribution of our main results. Out of 5,000 simulations, there
is not a single sample exhibiting a joint increase in short sales, in the sensitivity to negative
news, and in the proportion of options in new equity grants that are independently
statistically significant at the 10% level. Thus, it is unlikely that the results we document
are generated by methodology choices or sample selection.
{Insert Table XI here}
In addition, this robustness test validates the level of significance of our main tests.
In Table XI, for our main tests we provide the bootstrapped distribution of T-statistics from
the randomized samples. According to the bootstrapped distribution of T-statistics, the
change in the structure of new CEO equity grants is significant at the 1% level. In addition,
the change in the antitakeover provisions classified board and blank check is significant at
the 5% level.
V. Conclusion
In this paper, we investigate whether risk affects the design of CEO incentives. We
use a randomized natural experiment that exogenously increased downside equity risk
through the relaxation of short-selling constraints on a random sample of US stocks (Reg
27
SHO). Using difference-in-differences tests around the pilot program, we find that firms in
the treatment group reacted swiftly to the change in the firm’s risk environment by
increasing the proportion of stock options granted in new CEO equity grants. In addition,
we also find that this effect is significantly more pronounced for firms with larger changes
in the sensitivity of their stock prices to negative market news. Our evidence also indicates
that firms redesign the contracts of the other top executives as well as adopt anti-takeover
provisions after the adoption of Reg SHO. Finally, we find suggestive evidence that these
changes in incentive contracts influence corporate investment. Our results contribute to
the literature on CEO compensation by pointing to a causal effect of risk in the design of
CEO incentive contracts.
28
Bibliography
Aggarwal, Rajesh K., and Andrew A. Samwick, 1999, The other side of the trade-off: The
impact of risk on executive compensation, Journal of Political Economy, 107, 65-105.
Alexander, G.J., and M. A. Peterson, 2008, The effect of price tests on trader behavior and
market quality: An analysis of Reg SHO, Journal of Financial Markets 11, 84-111.
Allen, F., S. Morris, and A. Postlewaite, 1993, Finite Bubbles with Short Sale Constraints and
Asymmetric Information, Journal of Economic Theory 61, 206-229.
Amihud, Y., and B. Lev, 1981, Risk Reduction as a Managerial Motive for Conglomerate
Mergers, Bell Journal of Economics 12(2), 605–617.
Bates, D. S. “Empirical Option Pricing: A Retrospection.” Journal of Econometrics, 116
(2003), 387–404.
Becker, B. 2006, Wealth and executive compensation, Journal of Finance, 61(1): 379–397.
Berle, A.A. and G.C. Means, 1932, “The Modern Corporation and Private Property,” New York,
Macmillan.
Bollen, N. P. B., and R. E. Whaley, 2004, Does Net Buying Pressure Affect the Shape of
Implied Volatility Functions? Journal of Finance, 59, 711–753.
Carpenter, J., 2000, Does Option Compensation Increase Managerial Risk Appetite? Journal
of Finance 55: 2311-2331.
Chen, Q., I. Goldstein, and W. Jiang, 2007, Price Informativeness and Investment Sensitivity
to Stock Price, Review of Financial Studies 20(3), 619–50.
Coles, L.J., N.D. Daniel, and L. Naveen, 2006, “Managerial incentives and risk-taking.” Journal
of Financial Economics 79: 431-468.
Core, J, and W Guay, 1999, “The use of equity grants to manage optimal equity incentive
levels.” Journal of Accounting and Economics, 28(2): 151–184.
Core, John, and Wayne Guay, 2001, The other side of the tradeoff: the impact of risk on
executive compensation—a comment, Working paper, University of Pennsylvania.
Cuñat, V., and M. Guadalupe, 2009, “Globalization and the Provision of Incentives inside the
Firm: The Effect of Foreign Competition,” Journal of Labor Economics, 27(2), 179-212.
Diether, M., J. Lee, and J. Werner, 2009, It’s SHO Time! Short-Sale Price Tests and Market
Quality, Journal of Finance 64(1), 37-73.
29
Dittmann, I., and E. Maug, 2007, “Lower Salaries and No Options? On the Optimal Structure
of Executive Pay.” Journal of Finance 62: 303-343.
Dittmann, I, E. Maug, and O. Spalt, 2010, “Sticks or Carrots? Optimal CEO Compensation
when Managers Are Loss Averse.” The Journal of Finance 65(6): 2015–2050.
Ferreira, D, MA Ferreira, and CC Raposo, 2011, “Board structure and price
informativeness.” Journal of Financial Economics.
Gârleanu, N., Pedersen, L. H., & Poteshman, A. M., 2009, Demand-Based Option Pricing.
Review of Financial Studies, 22(10), 4259–4299.
Goldstein, I., and A. Guembel, 2008, Manipulation and the Allocational Role of Prices,
Review of Economic Studies 75(1), 133–64.
Gormley, T., D. Matsa, and T. Milbourn, 2012, “CEO Compensation and Corporate RiskTaking: Evidence from a Natural Experiment,” Working Paper.
Grullon, G., S. Michenaud, and J. Weston, 2011, The Real Effects of Short-Selling Constraints,
Working Paper.
Hayes, R., M. Lemmon, and M. Qiu, 2012, “Stock options and managerial incentives for risk
taking: Evidence from FAS 123R”, Journal of Financial Economics 105, 174-190.
Hemmer, T., O. Kim, and R. E. Verrecchia, 2000, Introducing Convexity into Optimal
Compensation Contracts, Journal of Accounting and Economics 28: 307-327.
Holmström, B. R., 1979, “Moral hazard and observability,” Bell Journal of Economics 10, 7491.
Holmström, B. R., and P. Milgrom, 1987, “Aggregation and linearity in the provision of
intertemporal incentives,” Econometrica 55: 303-328.
Holmström, B. R., and J. Tirole, 1993, “Market Liquidity and Performance Monitoring.”
Journal of Political Economy 101(4): 678–709.
Jensen, M.C. and W.H. Meckling, 1976, “Theory of the firm: Managerial behavior, agency
costs and ownership structure,” Journal of Financial Economics 3, 305-360.
Kadan, O., and J. M. Swinkels, 2008, Stocks or Options? Moral Hazard, Firm Viability, and the
Design of Compensation Contracts, Review of Financial Studies 21: 451-482.
Karpoff, J.M., and X. Lou, 2010, Short Sellers and Financial Misconduct, Journal of Finance
65: 1879-1913.
30
Knopf, J., J. Nam, and J. Thornton, 2002, “The volatility and price sensitivities of managerial
stock option portfolios and corporate hedging.” Journal of Finance 57: 801-814.
Lie, E., 2005, On the Timing of CEO Stock Option Awards, Management Science 51(5), 802–
12.
Lamont, O., 2012, Go Down Fighting: Short Sellers vs. Firms, Review of Asset Pricing Studies
2(1), 1–30.
May, D., 1995, Do Managerial Motives Influence Firm Risk Reduction Strategies? Journal of
Finance 50(4), 1291–1308.
Prendergast, C., 2002, The tenuous trade-off between risk and incentives, Journal of
Political Economy 110, 1071-1102.
Puhani, P. A., 2012, The treatment effect, the cross difference, and the interaction term in
nonlinear “difference-in-differences” models, Economic Letters 115: 85-87.
Ross, S., 2004, Compensation, Incentives, and the Duality of Risk Aversion and Riskiness,
Journal of Finance 59: 207-225.
Scheinkman, J., and W. Xiong, 2003, Overconfidence, Short-Sale Constraints, and Bubbles,
Journal of Political Economy 111, 1183–1219.
US Securities and Exchange Commission, 2007, Economic Analysis of the Short Sale Price
Restrictions Under the Regulation SHO Pilot, Office of Economic Analysis, Washington,
DC.
Xing, Y., X. Zhang, and R. Zhao, 2010, What does the individual option volatility smirk tell us
about future equity returns? Journal of Financial and Quantitative Analysis 45, 641-662.
31
Appendix 1
Construction of the sample of Pilot and Control group firms
The various steps in the sample selection process and the remaining firms in the sample are
detailed in the table below.
Total #
Firms left
after
selection
Selection process
Russell 3000 on May 31, 2004
# Firms in
Control
Group
# Firms in
Pilot Group
2,747
1,801
946
2,040
1,349
691
3,000
Only firms listed on Nasdaq national market
securities market, (NNM), AMEX and NYSE
2,968
Russell 3000 in 2004 and 2005
Compustat merge
Banks and financial services firms are excluded
Execucomp and RiskMetrics merge (Final
Sample)
32
2,565
1,442
1,685
935
883
507
Appendix 2
Definition of Main Variables
Abnormal Monthly Short
Interest
blankcheck
Cash flow
Cash Holdings
CAPX
CAPX+R&D
cboard
CEO Tenure
Control
Debt Issues
Dividends
Equity Issues
High Downside Risk
High Equity Change
Leverage
Low CAR
The residual of a firm fixed effect regression where Short Interest the monthly
mean ratio of net short positions outstanding reported on the 15th of each month
to shares outstanding at the start of the month is regressed on month dummies,
market-to-book, lagged total assets, logarithm of lagged Return on Assets,
Trading Volume, and a dummy variable for listing on the NYSE
Dummy variable equal to 1 if the firm has a blank check preferred provision
(blankcheck)
Net income before extraordinary Items (IB) + depreciation and amortization
expenses (DP) scaled by start-of-year total assets x 100
Cash and Short Term Investment (CHE) scaled by start-of-year total assets (AT) x
100
Capital expenditures (Compustat CAPX) scaled by start-of-year total assets (AT)
x 100
Capital expenditures (CAPX) plus Research and Development Expenses (XRD)
scaled by start-of-year total assets (AT) x 100
Dummy variable equal to 1 if the board of the company is classified (RiskMetrics:
cboard)
The difference between fiscal year and the year in which the CEO became the
CEO
Dummy variable equal to 1 if the company is not in the Pilot Group of REG SHO
Long-term debt Issues (DLTIS) scaled by start-of-year Total Assets (AT) x 100
Common Shares Dividends (DVC) plus Preferred Shares Dividends (DVP) scaled
by start-of-year total assets (AT) x100
Sale of Common and Preferred Shares (SSTK) scaled by start-of-year Total Assets
(AT) x 100
Dummy variable equal to 1 if the firm is in the top quintile of changes in stock
price returns sensitivity to negative market returns around the announcement
date. We measure changes in stock price returns sensitivity to negative market
returns as changes in firms’ stock returns when the daily stock market returns
fall into the lowest quintile of stock market return days (as shown in Table III).
The change is measured over a one-year period before and after the Reg SHO
announcement date.
Dummy variable equal to 1 if the increase in Option/Equity ($) from the 20032004 to the 2005-2006 period is in the top decile of the sample distribution
Long term debt (DLTT) plus debt in current liabilities (DLC) scaled by the sum of
long term debt, debt in current liabilities, and total stockholders’ equity (SEQ)
x 100
Dummy variable equal to 1 if firm’s CAR around the SHO announcement is below
the median
33
Market-to-Book ratio
Monthly Short Interest
Options ($)
Options (#)
Options/Equity ($)
Options/Equity (#)
Past profitability
Pilot
Placebo Treatment Years
Restricted Stock ($)
Restricted Stock (#)
severance
Short Interest
Supermajr
Total assets
Treatment Years
Market value of equity (PRCC x CSHO) plus book value of assets minus book value
of equity minus deferred taxes (when available) (AT-CEQ-TXDB), scaled by book
value of total assets (AT). Variable is lagged one year
Monthly short interest reported to NASDAQ or NYSE on the 15th of each calendar
month scaled by the total number of shares outstanding (from CRSP) at the start
of the month.
The value of stock options granted to the CEO (Execucomp – before 2006:
option_awards_blk_value – starting 2006: option_awards_fv)
The number of stock options granted to the CEO (option_awards_num)
Ratio of the value of stock options granted to the total value of equity grants in %
(100 x Options ($)/(Options ($)+Restricted Stock ($))
Ratio of the number of stock options granted to the total number of stock options
and shares of restricted stock granted in% (100 x Options (#)/(Options
(#)+Restricted Stock (#))
Ratio of operating income before depreciation and amortization (OIBDP) to startof-year total assets (AT) x 100. Variable is lagged one year
Dummy variable equal to 1 if the company is in the Pilot Group of REG SHO
Dummy variable equal to 1 if fiscal year is 2003 or 2004
The value of restricted stock granted to the CEO (before 2006: rstkgrnt – starting
2006: stock_awards_fv)
The number of shares of restricted stock granted to the CEO (Restricted Stock
($)/prcc_f)
Dummy variable equal to 1 if the firm uses severance packages (severance)
Average reported monthly short interest during the fiscal year, where monthly
short interest reported to NASDAQ or NYSE is scaled by the total number of
shares outstanding (from CRSP)
Dummy variable equal to 1 if the firm requires supermajority to approve a
merger (supermajor)
Start-of-year total assets (AT) (in million USD)
Dummy variable equal to 1 if fiscal year is 2005 or 2006
34
Figure 1
Timeline of the Reg SHO Experiment
10/28/2003
Proposed
Regulation SHO,
Pilot Test.
Consultation by
SEC
07/28/2004
Announcement of
SHO Pilot test, and
publication of the list
of Russell 3000 firms
in the Pilot
01/03/2005
Initial start date
of SHO Pilot test
35
05/02/2005
Start date of SHO
Pilot test:
Suspension of
price tests for
firms in the Pilot
04/28/2006
Initial end
date of SHO
Pilot test
07/06/2007
Actual end date of
SHO Pilot test, and
suspension of
price tests for all
firms in the US
stock markets
Figure 2
The Increase in Downside Risk in the Options Markets
This figure plots the average difference in implied volatility skew between pilot firms and control firms, both for puts and calls options. The implied
volatility skew is defined as the difference between the implied volatility of out-of-the-money puts (calls) on the stock of a firm and the implied volatility
of in-the-money puts (calls) on the stock of a firm and is measured at the daily level. We calculate the mean implied volatility skew for the one-month
period before the announcement of the RegSHO experiment on July 28, 2004 (Pre-announcement), and the one-month period following the
announcement.
0.90%
0.80%
Pre-Announcement
0.70%
Post-Announcement
0.60%
0.50%
0.40%
0.30%
0.20%
0.10%
0.00%
Pilot - Control Volatility Skew (Puts)
36
Pilot - Control Volatility Skew (Calls)
Figure 3
The Structure of Equity Grants: Pilot versus Control Group
This figure compares the evolution of the structure of CEO equity grants measured by the ratio of stock options granted to total equity grants (in value
and in number of shares) for firms in the pilot group and in the control group. Panel A plots the average ratio of the value of stock options granted to the
total value of equity grants. Panel B plots the average ratio of the number of stock options granted to the total number of stock options and shares of
restricted stock granted to the CEO. Panel C plots the difference of the structure of CEO equity grants between the pilot firms and the control firms in
any given year. Panel D plots the annual difference in differences of the structure of CEO equity grants between the pilot firms and the control firms
between year t-1 and year t.
Panel A: Option / Equity ($)
Start of the
Experiment
100%
Panel B: Option / Equity (#)
Repeal of the
Experiment
100%
90%
90%
80%
80%
70%
70%
60%
60%
50%
50%
40%
2001
2002
2003
Pilot
2004
2005
2006
Start of the
Experiment
Repeal of the
Experiment
2004
2006
40%
2007
2001
Control
2002
2003
Pilot
37
2005
Control
2007
Panel C: Differences between the two groups
(Pilot-Control)
Start of the
Experiment
Panel D: Annual Difference in Differences
Start of the
Experiment
Repeal of the
Experiment
6%
6%
4%
4%
2%
2%
0%
Repeal of the
Experiment
0%
2001
2002
2003
2004
2005
2006
2007
2002
-2%
-2%
-4%
-4%
-6%
-6%
Option / Equity ($)
2003
2004
Option / Equity ($)
Option / Equity (#)
38
2005
2006
Option / Equity (#)
2007
Table I
Summary Statistics
Data are collected from the merged CRSP/Compustat Industrial database, Execucomp, and RiskMetrics in the fiscal year that is the closest to July 28, 2004, the
announcement date of the SHO pilot test. We exclude firms that are not in the Russell 3000 index in 2004 and 2005, and financial services firms (SIC code 60006999). All variables are described in Appendix 2.
Pilot group
Total assets
Market-to-Book ratio
CAPX
CAPX+R&D
Cash flow
Leverage
Dividends
Cash Holdings
Past profitability
Equity Grant ($)
Options/Equity ($)
Options/Equity (#)
cboard
blankcheck
supermajor
severance
G Index
CEO Tenure
Control group
N
Mean
Median
Std. Dev
N
Mean
Median
Std. Dev
Diff.
T-stat
471
471
471
466
470
469
471
471
470
442
353
357
473
473
473
473
473
438
4,635
2.11
5.85
9.82
10.79
29.08
1.01
20.92
12.66
2,388
73.23
80.47
0.57
0.90
0.16
0.06
9.12
6.92
1,132
1.80
3.99
7.06
10.48
28.22
0.00
13.35
13.07
1,338
100
100
1.00
1.00
0.00
0.00
9.00
5.00
13,064
1.06
6.18
8.58
10.01
24.61
1.70
23.15
9.71
2,926
35.20
31.65
0.50
0.30
0.37
0.24
2.66
6.30
878
878
877
872
877
873
876
877
874
807
660
667
860
860
860
860
860
804
5,263
2.13
5.33
9.60
10.80
29.29
0.97
22.81
12.39
2,579
75.50
81.44
0.60
0.91
0.15
0.07
9.17
6.46
1,199
1.75
3.50
7.23
10.72
26.96
0.00
13.48
12.44
1,386
100
100
1.00
1.00
0.00
0.00
9.00
4.00
14,738
1.08
5.61
10.37
12.17
27.12
1.73
23.92
10.86
3,122
34.58
31.27
0.49
0.29
0.36
0.25
2.45
5.76
-629
-0.02
0.52
0.23
0.56
-0.21
0.04
-1.90
0.27
-191
-2.27
-0.96
-0.03
-0.01
0.01
0.01
-0.05
0.47
-0.78
-0.25
1.56
0.48
0.72
-0.14
0.42
-1.41
0.46
-1.06
-0.99
-0.47
-0.92
-0.45
0.66
0.28
-0.37
1.32
39
Table II
SHO Pilot and Short Interest
This table presents mean values of Short Interest and Abnormal Monthly Short Interest for firms that were part
of the pilot group and control group three years before and after the announcement date (July 28, 2004).
Short Interest is the monthly mean ratio of net short positions outstanding reported on the 15th of each
month to shares outstanding at the start of the month. Abnormal Monthly Short Interest is the residual of a
firm fixed effect regression where Short Interest is regressed on firm fixed effects, controlling for month
dummies, market-to-book lagged total assets, logarithm of lagged Return on Assets, Trading Volume, and a
dummy variable for listing on the NYSE. Averages are computed for all firms that are in the Pilot Group and in
the Control Group. T-statistics are constructed with Newey-West standard errors (8 lags). c, b, a indicate a
significance level of less than 10%, 5%, and 1% respectively.
Short Interest
Abnormal Short Interest
Before
After
Diff.
Pilot Group
4.26
6.08
+1.82
Control Group
4.50
5.95
+1.45
Difference
-0.24
T-stat
(-1.59)
Difference-in-differences
T-stat
Before
After
Diff.
a
(10.64)
+0.20
-0.01
-0.21
a
(11.29)
+0.34
-0.16
-0.50
+0.12
-0.14
+0.15
(1.41)
(-1.41)
(1.68)
c
+0.37
(1.73)
40
T-stat
b
(-2.03)
a
(-6.50)
c
+0.29
b
(2.21)
Table III
Downside Risk: Sensitivity to Realized and Anticipated Negative News
Panel A presents the mean daily raw returns for all firms in the sample that were part of the pilot experiment, and firms
that were part of the control group. We sort the observations by quintiles based on the value-weighted daily market
returns (from CRSP), and then compute the average daily market returns for the pilot and control firms for each
quintile. Quintile 1 of the value-weighted daily market returns is the lowest quintile of market daily returns while
quintile 5 is the largest. The difference-in-differences measures the change in mean daily returns after the
announcement of the Pilot (versus before the announcement of the Pilot) for the pilot group relative to the control
group. Point estimates are based on OLS regressions where the daily returns are regressed on a dummy for firms in the
Pilot, a dummy variable equal to 1 after the experiment is announced (July 28, 2004) and the interaction term of these
two variables. Before is the one-year period before July 28, 2004. Panel B reports the cumulative abnormal returns
computed one day before up to one day after the date of announcement of the negative earnings news for all firms in the
Pilot Group. Before is the two-year period before July 28, 2004. After is the two-year period after July 28, 2004.
Quarterly Negative Earnings Surprises are negative surprises of quarterly earnings relative to the last analyst consensus
forecast from I/B/E/S. We restrict our analysis to negative earnings surprises that are below the median negative
earnings surprises. Panel C presents the same results for the positive earnings surprises that are above the median
positive earnings surprises. Panel D reports the average daily volatility skew of put options on stocks of for all firms in
the Pilot Group and the Control Group. Volatility Skew is computed as the difference between the implied volatility of
out of the money puts (strike price to stock price ratio is less than .9 and more than .7) and at the money puts (strike
price to stock price ratio is less than 1.05 and more than .95). Before is the two-month period before July 28, 2004. After
is the two-month period after July 28, 2004. Standard errors are clustered at the firm and date level. c, b, a indicate a
significance level of less than 10%, 5%, and 1% respectively.
Panel A: Sensitivity to Daily Market Returns
Before
After
Quintile
Pilot
Control
Diff.
T-stat
Pilot
Control
Diff.
T-stat
Diff.-inDiff.
1
-1.46
-1.48
0.03
(1.56)
-1.38
-1.33
-0.05
(-1.44)
-0.07
a
(-2.63)
2
-0.31
-0.30
-0.01
(-0.65)
-0.37
-0.34
-0.03
(-1.64)
-0.01
(-0.62)
3
0.20
0.19
-0.01
(-0.80)
0.11
0.11
-0.00
(-0.09)
-0.01
(-0.66)
4
0.73
0.75
-0.03
(-1.17)
0.59
0.57
0.02
(1.03)
0.05
c
(1.95)
5
1.61
1.61
-0.00
(-0.06)
1.34
1.29
0.05
(1.05)
0.05
(1.56)
41
T-stat
Panel B: Cumulative Abnormal Returns after Large Negative Earnings News
Before
After
a
-5.06
Difference
T-stat
a
-1.72
a
(-3.23)
a
+0.12
(0.22)
b
(-2.55)
Pilot Group
-2.74
Control Group
-3.33
a
-3.21
Difference
+0.59
-1.83
T-stat
(0.98)
(-3.51)
a
Difference-in-differences
-1.85
Panel C: Cumulative Abnormal Returns after Large Positive Earnings News
Before
Difference
T-stat
a
+0.22
(0.56)
a
-0.46
(-1.56)
0.68
(1.38)
Difference
T-stat
a
3.48
3.53
a
3.07
-0.28
0.40
(-0.67)
(1.33)
Pilot Group
3.25
Control Group
Difference
T-stat
After
Difference-in-differences
Panel D: Volatility Skew on Put Options
Before
Pilot Group
After
a
a
7.30
8.85
a
a
Control Group
7.25
8.05
Difference
+0.05
+0.80
(+0.12)
(+1.64)
T-stat
Difference-in-differences
a
(+4.29)
a
(+2.38)
b
(+2.16)
+1.55
+0.80
+0.75
Panel E: Volatility Skew on Call Options
Before
After
Difference
T-stat
Pilot Group
0.14
0.08
-0.06
(-0.19)
Control Group
-0.16
0.07
+0.22
(+0.66)
Difference
+0.27
+0.01
(+1.19)
(+0.03)
-0.29
(-1.18)
T-stat
Difference-in-differences
42
Table IV
The Impact of Downside Equity Risk on the Structure of Equity Grants awarded to the CEO
This table shows results of OLS, fixed-effect (FE) and Tobit regressions. Tobit regressions are left censored at 0 and right censored at 1. The sample
period is fiscal year 2003 to 2006 for Panel A, and fiscal year 2001 to 2007 for Panel B. The dependent variables are the ratio of the value of stock
options granted to the CEO to the total value of equity grants (Option/Equity ($)), and the ratio of the number of stock options granted to the total
number of stock options and shares of restricted stock granted to the CEO (Option/Equity (#)). Pilot is a dummy variable equal to 1 if the company is in
the Pilot Group of REG SHO. Treatment Years is a dummy variable equal to 1 if fiscal year is 2005 or 2006. Standard errors are clustered at the firm level.
T-statistics are reported in parenthesis. c, b, a indicate a significance level of less than 10%, 5%, and 1% respectively.
Panel A: DiD Analysis (2003-2006)
VARIABLES
Pilot
Treatment Years
Treatment Years*Pilot
Constant
Observations
2
2
Adjusted R / Pseudo R
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
-1.96
(-1.01)
a
-20.17
(-15.71)
a
5.95
(2.91)
-0.86
(-0.52)
a
-17.05
(-13.51)
b
4.68
(2.38)
a
FE
Option/
Equity ($)
a
-15.57
(-12.34)
c
3.50
(1.84)
a
-18.74
(-14.58)
b
4.78
(2.40)
a
FE
Option/
Equity (#)
a
a
Tobit
Option/
Equity ($)
Tobit
Option/
Equity (#)
-5.46
(-0.93)
a
-53.84
(-14.00)
b
14.34
(2.55)
-3.46
(-0.66)
a
-47.38
(-13.12)
b
11.60
(2.26)
a
a
78.32
(69.60)
84.14
(85.79)
77.14
(154.34)
83.30
(172.01)
122.79
(31.03)
126.20
(36.82)
4,004
0.058
4,036
0.049
4,004
0.477
4,036
0.478
4,004
0.012
4,036
0.011
43
Panel B: DiD Analysis By Year and Extended Sample Period (2001-2007)
VARIABLES
Pilot
Year 2001
Year 2002
Year 2003
Year 2005
Year 2006
Year 2007
Year 2001 * Pilot
Year 2002 * Pilot
Year 2003 * Pilot
Year 2005 * Pilot
Year 2006 * Pilot
Year 2007 * Pilot
Constant
Observations
2
2
Adjusted R / Pseudo R
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
-2.27
(-0.98)
a
13.26
(9.21)
a
10.49
(7.54)
a
5.68
(4.47)
a
-9.61
(-6.69)
a
-25.24
(-13.65)
a
-29.22
(-15.57)
1.87
(0.78)
2.33
(0.97)
0.90
(0.39)
b
5.69
(2.56)
b
6.79
(2.23)
2.97
(0.95)
-0.96
(-0.47)
a
10.67
(8.28)
a
8.01
(6.44)
a
5.46
(4.66)
a
-8.05
(-5.72)
a
-20.77
(-11.72)
a
-24.89
(-13.60)
1.35
(0.62)
1.12
(0.50)
0.44
(0.21)
b
4.25
(1.97)
c
5.30
(1.85)
3.15
(1.03)
a
a
FE
Option/
Equity ($)
FE
Option/
Equity (#)
a
11.73
(8.82)
a
8.68
(6.76)
a
4.98
(4.47)
a
-8.13
(-6.00)
a
-19.16
(-11.08)
a
-23.56
(-13.17)
1.43
(0.66)
2.91
(1.32)
1.95
(0.98)
b
4.10
(2.07)
c
5.07
(1.88)
3.49
(1.19)
a
14.75
(10.09)
a
11.71
(8.30)
a
5.40
(4.52)
a
-9.87
(-7.19)
a
-23.57
(-12.90)
a
-27.71
(-14.97)
1.52
(0.65)
3.78
(1.60)
2.11
(1.00)
a
5.65
(2.78)
b
6.17
(2.10)
3.36
(1.10)
a
a
Tobit
Option/
Equity ($)
Tobit
Option/
Equity (#)
-6.16
(-0.97)
a
43.80
(8.91)
a
34.46
(7.58)
a
18.29
(4.85)
a
-25.48
(-6.72)
a
-61.98
(-12.53)
a
-69.83
(-13.81)
4.87
(0.61)
5.41
(0.69)
2.69
(0.40)
b
13.27
(2.30)
b
16.44
(2.16)
7.15
(0.92)
-3.75
(-0.65)
a
39.30
(8.74)
a
30.56
(7.34)
a
17.62
(5.09)
a
-22.56
(-6.27)
a
-53.75
(-11.64)
a
-60.76
(-12.80)
3.96
(0.54)
2.41
(0.34)
1.51
(0.24)
c
10.45
(1.93)
c
13.11
(1.88)
6.87
(0.95)
a
a
75.50
(56.06)
81.44
(67.22)
73.89
(99.24)
80.37
(114.91)
112.38
(27.84)
116.91
(32.56)
6,809
0.163
6,883
0.129
6,809
0.488
6,883
0.465
6,809
0.033
6,883
0.031
44
Table V
Downside Equity Risk & CEO Incentive Contracts – Difference-in-Difference-in-Differences Analysis
This table shows results of OLS, fixed-effect (FE) and Tobit regressions. Tobit regressions are left censored at 0 and right censored at 1. The sample
period is fiscal year 2003 to 2006. The dependent variables are the ratio of the value of stock options granted to the CEO to the total value of equity
grants (Option/Equity ($)), and the ratio of the number of stock options granted to the total number of stock options and shares of restricted stock
granted to the CEO (Option/Equity (#)). Pilot is a dummy variable equal to 1 if the company is in the Pilot Group of REG SHO. Treatment Years is a
dummy variable equal to 1 if fiscal year is 2005 or 2006. High Downside Risk is a dummy variable equal to 1 if the firm is in the top quintile of changes in
stock price returns sensitivity to negative market returns around the announcement date. We measure changes in stock price returns sensitivity to
negative market returns as changes in firms’ stock returns when the daily stock market returns fall into the lowest quintile of stock market return days
(as shown in Table III). The change is measured over a one-year period before and after the Reg SHO announcement date. Standard errors are clustered
at the firm level. T-statistics are reported in parenthesis. c, b, a indicate a significance level of less than 10%, 5%, and 1% respectively.
VARIABLES
Pilot
Treatment Years
Treatment Years*Pilot
High Downside Risk
High Downside Risk *Pilot
High Downside Risk *Treatment Years
High Downside Risk *Treat. Years*Pilot
Constant
Observations
2
2
Adjusted R / Pseudo R
OLS
Option/
Equity ($)
-0.94
(-0.42)
a
-19.68
(-13.42)
3.64
(1.54)
0.27
(0.09)
-2.18
(-0.46)
0.16
(0.05)
c
9.54
(1.84)
a
79.43
(63.59)
3,654
0.059
OLS
Option/
Equity (#)
-0.60
(-0.31)
a
-15.99
(-11.14)
2.52
(1.10)
-0.41
(-0.16)
0.21
(0.05)
-0.43
(-0.12)
c
8.18
(1.67)
a
85.00
(77.40)
3,682
0.047
45
FE
Option/
Equity ($)
a
FE
Option/
Equity (#)
a
-18.77
(-12.95)
2.72
(1.19)
-15.22
(-10.81)
1.55
(0.71)
1.07
(0.31)
c
9.86
(1.90)
a
78.60
(155.76)
3,654
0.468
0.62
(0.18)
c
8.75
(1.80)
a
84.39
(173.39)
3,682
0.468
Tobit
Option/
Equity ($)
-2.21
(-0.32)
a
-51.29
(-12.30)
7.01
(1.10)
4.16
(0.45)
-9.24
(-0.65)
-2.04
(-0.22)
b
28.84
(2.06)
a
123.08
(28.75)
3,654
0.012
Tobit
Option/
Equity (#)
-1.44
(-0.24)
a
-44.35
(-11.37)
4.97
(0.86)
2.48
(0.31)
-5.53
(-0.44)
-2.06
(-0.24)
b
25.67
(2.02)
a
126.06
(33.80)
3,682
0.011
Table VI
The Impact of Downside Equity Risk on the Structure of Equity Grants awarded to all Firm Executives
This table shows results of OLS, firm fixed-effect (Firm FE), executive fixed-effect (Exec FE) and Tobit regressions. Tobit regressions are left censored at
0 and right censored at 1. The sample period is fiscal year 2003 to 2006 for Panel A, and fiscal year 2001 to 2007 for Panel B. The dependent variables
are the ratio of the value of stock options granted to the CEO to the total value of equity grants (Option/Equity ($)), and the ratio of the number of stock
options granted to the total number of stock options and shares of restricted stock granted to the CEO (Option/Equity (#)). Pilot is a dummy variable
equal to 1 if the company is in the Pilot Group of REG SHO. Treatment Years is a dummy variable equal to 1 if fiscal year is 2005 or 2006. Standard errors
are clustered at the firm level. T-statistics are reported in parenthesis. c, b, a indicate a significance level of less than 10%, 5%, and 1% respectively.
Panel A: DiD Analysis (2003-2006)
VARIABLES
Pilot
Treatment Years
Treatment Years*Pilot
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
-0.43
(-0.25)
a
-20.48
(-18.83)
a
5.57
(3.14)
0.15
(0.11)
a
-18.40
(-17.74)
a
4.99
(3.01)
a
a
Firm FE
Option/
Equity ($)
Firm FE
Option/
Equity (#)
a
-17.60
(-16.76)
a
4.55
(2.79)
a
-19.99
(-18.30)
a
4.98
(2.86)
Exec FE
Option/
Equity ($)
a
-18.16
(-16.77)
a
5.08
(2.98)
a
Exec FE
Option/
Equity (#)
a
-16.56
(-15.94)
a
4.45
(2.82)
a
a
Tobit
Option/
Equity ($)
Tobit
Option/
Equity (#)
-1.41
(-0.25)
a
-55.78
(-16.30)
b
12.79
(2.51)
-0.28
(-0.06)
a
-50.97
(-16.10)
b
10.89
(2.44)
Constant
77.80
(74.53)
84.90
(102.28)
77.51
(184.05)
84.66
(227.80)
76.59
(184.38)
84.20
(231.34)
a
123.82
(33.50)
a
129.07
(41.57)
Observations
2
2
Adjusted R / Pseudo R
22,322
0.062
24,549
0.060
22,322
0.559
24,549
0.547
22,322
0.492
24,549
0.479
22,322
0.012
24,549
0.014
46
a
Panel B: DiD Analysis By Year and Extended Sample Period (2001-2007)
VARIABLES
Pilot
Year 2001
Year 2002
Year 2003
Year 2005
Year 2006
Year 2007
Year 2001 * Pilot
Year 2002 * Pilot
Year 2003 * Pilot
Year 2005 * Pilot
Year 2006 * Pilot
Year 2007 * Pilot
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
-0.88
(-0.41)
a
14.32
(10.88)
a
11.79
(9.38)
a
7.01
(6.16)
a
-9.00
(-7.72)
a
-24.88
(-15.72)
a
-27.99
(-16.94)
0.82
(0.36)
0.38
(0.17)
0.93
(0.46)
a
5.22
(2.81)
b
6.73
(2.54)
2.44
(0.90)
0.08
(0.04)
a
11.15
(10.18)
a
9.01
(8.62)
a
6.58
(6.77)
a
-8.57
(-7.85)
a
-21.55
(-14.74)
a
-25.18
(-16.04)
0.10
(0.05)
-0.47
(-0.25)
0.22
(0.13)
a
4.47
(2.65)
b
5.62
(2.33)
3.24
(1.27)
a
Firm FE
Option/
Equity ($)
Firm FE
Option/
Equity (#)
a
11.92
(10.77)
a
9.34
(8.82)
a
6.79
(7.31)
a
-8.98
(-8.50)
a
-20.13
(-13.87)
a
-24.24
(-15.79)
0.63
(0.34)
0.76
(0.41)
0.24
(0.15)
a
4.78
(3.02)
b
5.14
(2.19)
3.15
(1.27)
a
15.54
(12.12)
a
12.67
(10.30)
a
7.09
(6.74)
a
-9.88
(-8.89)
a
-23.90
(-15.13)
a
-27.50
(-16.97)
1.00
(0.47)
0.88
(0.42)
1.02
(0.55)
a
5.74
(3.34)
b
6.25
(2.40)
2.91
(1.10)
Exec FE
Option/
Equity ($)
a
14.31
(11.28)
a
12.22
(10.17)
a
6.60
(6.35)
a
-9.54
(-8.54)
a
-23.40
(-14.67)
a
-26.75
(-16.03)
1.46
(0.69)
0.61
(0.29)
1.37
(0.75)
a
5.76
(3.33)
b
6.56
(2.51)
2.20
(0.80)
a
Exec FE
Option/
Equity (#)
a
11.80
(10.72)
a
9.40
(8.97)
a
6.69
(7.24)
a
-9.07
(-8.57)
a
-19.94
(-13.63)
a
-24.06
(-15.35)
0.63
(0.34)
0.70
(0.37)
0.37
(0.22)
a
4.88
(3.10)
b
5.32
(2.28)
2.39
(0.93)
a
a
Tobit
Option/
Equity ($)
Tobit
Option/
Equity (#)
-2.61
(-0.43)
a
49.10
(10.58)
a
41.23
(9.56)
a
24.13
(6.86)
a
-24.15
(-7.55)
a
-61.90
(-14.08)
a
-68.60
(-14.97)
3.90
(0.50)
-1.29
(-0.17)
2.94
(0.48)
b
11.57
(2.28)
b
15.84
(2.28)
5.12
(0.73)
-1.25
(-0.24)
a
43.92
(11.07)
a
37.76
(10.15)
a
22.82
(7.64)
a
-23.25
(-8.05)
a
-55.31
(-13.91)
a
-61.10
(-14.67)
2.43
(0.37)
-3.69
(-0.60)
2.60
(0.50)
b
9.99
(2.25)
b
13.45
(2.22)
5.85
(0.95)
Constant
74.30
(59.23)
a
81.52
(76.79)
73.56
(112.83)
81.02
(142.33)
73.62
(113.94)
81.02
(142.97)
a
111.35
(29.47)
a
117.58
(36.68)
Observations
2
2
Adjusted R / Pseudo R
38,156
0.163
43,184
0.143
38,156
0.536
43,184
0.501
38,156
0.527
43,184
0.492
38,156
0.034
43,184
0.035
47
a
Table VII
The Impact of Downside Equity Risk on Antitakeover Provisions and Severance Packages
This table shows results of Logit regressions. The dependent variables are dummy variables equal to 1 if the board of the company is classified (cboard),
the firm has a blank check preferred provision (blankcheck), the firm requires supermajority to approve a merger (supermajor), and the firm uses
severance packages (severance). Pilot is a dummy variable equal to 1 if the company is in the Pilot Group of REG SHO. Treatment Years is a dummy
variable equal to 1 if fiscal year is 2005 or 2006. Standard errors are clustered at the firm level. T-statistics are reported in parenthesis. c, b, a indicate a
significance level of less than 10%, 5%, and 1% respectively.
VARIABLES
Pilot
Treatment Years
Treatment Years*Pilot
Logit
cboard
Logit
blankcheck
Logit
supermajor
Logit
severance
-0.11
(-0.92)
a
-0.16
(-3.51)
b
0.14
(1.99)
-0.09
(-0.45)
0.02
(0.27)
c
0.19
(1.74)
0.10
(0.66)
-0.06
(-1.18)
0.09
(1.27)
-0.07
(-0.28)
a
-0.75
(-3.80)
0.40
(1.35)
a
a
a
a
Constant
0.40
(5.76)
2.29
(19.40)
-1.73
(-18.12)
-2.63
(-19.31)
Observations
2
Pseudo R
2,616
0.0008
2,616
0.0006
2,616
0.0009
2,616
0.012
48
Table VIII
Contracting and Investment Outcomes
This table shows results of OLS regressions. The sample period is fiscal year 2003 to 2006 and the sample firms are restricted to non-utilities firms in
the Pilot Group of REG SHO. The dependent variables are the ratio of capital expenditures to start-of-year total assets multiplied by 100 (CAPX), and the
ratio of the sum of capital expenditures and research and development expenses to start-of-year total assets multiplied by 100 (CAPX+R&D). High
Equity Change is a dummy variable equal to 1 if the increase in Option/Equity ($) from the 2003-2004 to the 2005-2006 period is in the top decile of the
sample distribution. Option/Equity ($) is the ratio of the value of stock options granted to the CEO to the total value of equity grants. Treatment Years is
a dummy variable equal to 1 if fiscal year is 2005 or 2006. Standard errors are clustered at the firm level. T-statistics are reported in parenthesis. c, b, a
indicate a significance level of less than 10%, 5%, and 1% respectively.
VARIABLES
High Equity Change
Treatment Years
Treat. Years*High Equity Change
Constant
Observations
R-squared
49
OLS
CAPX
OLS
CAPX+R&D
0.97
(0.84)
0.08
(0.39)
b
1.90
(2.04)
-0.64
(-0.48)
0.35
(1.19)
b
2.59
(2.18)
5.62
(16.33)
a
9.41
(17.87)
760
0.014
760
0.001
a
Table IX
Placebo Tests
This table shows results of OLS, fixed-effect (FE) and Tobit regressions. Tobit regressions are left censored at 0 and right censored at 1. The sample
period is fiscal year 2001 to 2004. The dependent variables are the ratio of the value of stock options granted to the CEO to the total value of equity
grants (Option/Equity ($)), and the ratio of the number of stock options granted to the total number of stock options and shares of restricted stock
granted to the CEO (Option/Equity (#)). Pilot is a dummy variable equal to 1 if the company is in the Pilot Group of REG SHO. Placebo Treatment Years is
a dummy variable equal to 1 if fiscal year is 2003 or 2004. Standard errors are clustered at the firm level. T-statistics are reported in parenthesis. c, b, a
indicate a significance level of less than 10%, 5%, and 1% respectively.
VARIABLES
Pilot
Placebo Treatment Years
Placebo Treatment Years*Pilot
Constant
Observations
2
2
Adjusted R / Pseudo R
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
-0.13
(-0.08)
a
-9.03
(-8.54)
-1.83
(-1.04)
0.31
(0.25)
a
-6.62
(-7.16)
-1.17
(-0.73)
a
a
FE
Option/
Equity ($)
FE
Option/
Equity (#)
a
-7.70
(-8.12)
-1.13
(-0.70)
a
-10.51
(-9.70)
-1.55
(-0.86)
a
a
Tobit
Option/
Equity ($)
Tobit
Option/
Equity (#)
-0.87
(-0.14)
a
-30.65
(-8.08)
-4.56
(-0.75)
-0.37
(-0.07)
a
-24.86
(-7.68)
-2.83
(-0.55)
a
a
87.35
(96.85)
90.75
(120.08)
88.03
(194.43)
91.42
(228.05)
152.89
(32.28)
146.82
(36.66)
3,793
0.025
3,829
0.017
3,793
0.482
3,829
0.439
3,793
0.006
3,829
0.005
50
Table X
Tests of Alternative Timing and Channel
This table shows results of OLS regressions. The dependent variables are the ratio of the value of stock options granted to the CEO to the total value of
equity grants (Option/Equity ($)), and the ratio of the number of stock options granted to the total number of stock options and shares of restricted
stock granted to the CEO (Option/Equity (#)). Pilot is a dummy variable equal to 1 if the company is in the Pilot Group of REG SHO. Treatment Years is a
dummy variable equal to 1 if fiscal year is 2005 or 2006. Low CAR is a dummy variable equal to 1 if firm’s CAR around the SHO announcement is below
the median. Panel A shows results for a restricted sample of firms with fiscal end month ending after the month of July (Fiscal month end>July) and a
sample period from fiscal year 2003 to 2006. Panel B shows results for a restricted sample period: fiscal year 2003, 2005 and 2006 (Drop fiscal year
2004). In Panel C, the sample period is fiscal year 2003 to 2006. Standard errors are clustered at the firm level. T-statistics are reported in parenthesis. c,
b, a indicate a significance level of less than 10%, 5%, and 1% respectively.
VARIABLES
Pilot
Treatment Years
Treatment Years*Pilot
Panel A: Alternative Timing
Fiscal month end>July
Panel B: Alternative Timing
Drop fiscal year 2004
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
OLS
Option/
Equity ($)
OLS
Option/
Equity (#)
OLS
Option/
Equity ($)
-2.36
(-1.04)
a
-20.87
(-14.06)
a
7.30
(3.13)
-1.01
(-0.51)
a
-17.89
(-12.23)
a
5.93
(2.61)
-1.37
(-0.63)
a
-23.03
(-15.84)
b
5.36
(2.18)
-0.52
(-0.28)
a
-19.81
(-14.10)
c
4.34
(1.86)
-5.20
(-1.77)
a
-20.77
(-11.23)
a
9.07
(3.02)
0.75
(0.33)
c
7.59
(1.95)
1.70
(0.65)
-6.25
(-1.48)
Low CAR
Low CAR*Pilot
Low CAR*Treatment Years
Low CAR*Treatment Years*Pilot
Constant
Observations
2
Adjusted R
a
a
a
a
Panel C: Pricing
OLS
Option/
Equity (#)
c
-3.68
(-1.45)
a
-17.49
(-9.65)
b
6.65
(2.30)
-0.05
(-0.03)
c
6.29
(1.88)
1.89
(0.73)
-4.01
(-1.00)
a
a
76.69
(57.36)
82.94
(70.31)
81.18
(65.77)
86.89
(81.50)
79.12
(47.90)
84.94
(58.74)
3,129
0.058
3,154
0.049
2,991
0.067
3,012
0.057
3,739
0.062
3,769
0.050
51
Table XI
Bootstrapped Distribution of T-statistics for Randomized Samples
This table presents the distribution of t-stats of the OLS regressions when we randomize the selection of firms in the Pilot and Control Group using
5,000 simulations. The t-stats correspond to the DiD coefficient or the interaction variable between the Treatment dummy variable and the Pilot
dummy variable in all the differences-in-differences analyses.
Short
Interest
Abnormal
Short
Interest
Sensitivity to
Daily Market
Returns
Option/
Equity ($)
Option/
Equity (#)
Logit
cboard
Logit
blankcheck
1
2
3
4
5
6
7
1%
-2.54
-3.47
-2.34
-2.33
-2.32
-2.31
-2.47
5%
-1.76
-2.47
-1.72
-1.66
-1.67
-1.61
-1.74
10%
-1.41
-1.98
-1.33
-1.29
-1.30
-1.27
-1.33
50%
-0.01
+0.02
-0.01
+0.03
+0.04
-0.02
+0.00
90%
+1.37
+1.87
+1.33
+1.32
+1.32
+1.29
+1.26
95%
+1.72
+2.41
+1.71
+1.70
+1.70
+1.65
+1.62
99%
+2.40
+3.27
+2.45
+2.30
+2.34
+2.40
+2.29
DiD
DiD
DiD
Treatment
Years*Pilot
Treatment
Years*Pilot
Treatment
Years*Pilot
Treatment
Years*Pilot
Location
Table
II
Table
II
Table
III.A.Q.1
Table
IV.A.1
Table
IV.A.2
Table
VII.1
Table
VII.2
Reported T-stat
1.73
2.21
2.63
2.91
2.38
1.99
1.74
Significance level
5%
10%
1%
1%
1%
5%
5%
Percentiles
Coefficient
52