Option Expensing and Executive Compensation

Document Sample
Option Expensing and Executive Compensation Powered By Docstoc
					Topic: http://www.isknow.com/compensation




                Option Expensing and Executive Compensation*



                                           Yi Feng
                                 Schulich School of Business
                                        York University
                                       4700 Keele Street
                              Toronto, Ontario, Canada M3J 1P3
                                   yfeng@schulich.yorku.ca
                                  (416) 736-2100 ext. 20635

                                                and

                                        Yisong S. Tian**
                                 Schulich School of Business
                                        York University
                                       4700 Keele Street
                              Toronto, Ontario, Canada M3J 1P3
                                   ytian@schulich.yorku.ca
                                  (416) 736-2100 ext. 77943



                                       September 4, 2007



*
     We thank Melanie Cao, Don Chance, Mark Huson, Mark Kamstra, Nadia Massoud, Moshe
     Milevsky, Debarshi Nandy and participants in the 2006 Northern Finance Association and
     2007 Asian Finance Association annual meetings and seminar participants at York University
     for helpful comments and suggestions. Financial support provided by the York University’s
     Schulich School of Business and the Social Sciences and Humanities Research Council of
     Canada is gratefully acknowledged.
**
     Corresponding author.




Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Abstract


     This paper examines the impact of mandatory option expensing on executive compensation. Using
CEO compensation data from 1993 to 2004, we document a sharp reversal in the use of option incentives
around 2001 while the use of stock incentives increases throughout the entire period. Median CEO option
incentives increases 25% a year before 2002 but declines 18% a year after 2001. We argue that this sharp
reversal is related to the recent change in option expensing rules. First, we develop a principal-agent
model that incorporates the impact of option expensing on executive compensation. The executive’s
perceived cost of option expensing alters the optimal contract and leads to a reduction in equity
incentives, option incentives in particular. This theoretical result provides a rational explanation for the
documented decline in the use of option incentives. Secondly, we find evidence that the decline in option
incentives is larger in firms using excessive levels of equity incentives prior to option expensing than in
control firms. This is consistent with the expected impact of option expensing across firms with different
levels of equity incentives prior to option expensing. Thirdly, firms make similar reductions to option
grants made to the CEO, other top executives and lower-level employees, again consistent with the
expected impact of option expensing. Finally, we find that other regulatory, business and market events
that coincide with the change in option expensing rules such as the Sarbanes-Oxley Act of 2002, the
option backdating scandal, and the 2000 stock market crash contribute to but do not fully explain the
documented decline in option incentives.


Keywords:       Option expensing; Executive compensation; Optimal contracting; Equity incentives;
                Option incentives; Stock incentives; Pay-performance sensitivity


JEL classification:     G30, M40, M52




Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



1. Introduction
         In this paper, we investigate the impact of mandatory option expensing on the use of equity
incentives, option incentives in particular, in executive compensation. As all U.S. firms are already
required to disclose the fair value of stock options in footnotes to financial statements since 1995, it is not
apparent why the recent change in option expensing rules should have any material effect on the way
firms compensate their executives. In an efficient market where all users of financial statements fully
understand and utilize all disclosed information, mandatory option expensing should have little or no
impact on firm value and performance as it merely moves the option expense from the footnotes to the
income statement. It is thus not unreasonable to argue that the recent change in the accounting treatment
of stock options should not have any material impact on executive compensation.1
         Contrary to the efficient market argument, many executives appear to believe that there are real
economic costs associated with option expensing perhaps due to the reduction in reported earnings and
other accounting concerns. A recent survey by Graham, Harvey, and Rajgopal (2005) finds that most
Chief Financial Officers (CFOs) believe that the market pays more attention to reported income than to
financial footnotes. Executives may thus object to mandatory option expensing because they believe that
it would lead to substantial reductions in reported income2 which might make it difficult to maintain the
pattern of increasing earnings, meet analyst expectations, or meet contractual agreements with
debtholders. Such difficulties may negatively affect their performance and compensation. In addition,
mandatory option expensing is also likely to make executive compensation more transparent and it thus
may become more difficult for executives to “camouflage” the true nature and magnitude of their
compensation (e.g., Bebchuk, Fried, and Walker, 2002; and Bebchuk and Fried, 2003). Such objection to
option expensing may reduce the attractiveness of stock options to executives and possibly alter the
optimal contracts between firms and their executives.
         To provide new insight on the relationship between option expensing and equity-based incentive
pay, we first develop a principal-agent model of executive compensation that incorporates the impact of
option expensing on equity incentives. In our model, there is no economic cost to the firm that is directly
associated with option expensing. Instead, the executive’s objection to option expensing is incorporated
into his utility function, making stock options less attractive to him than they are prior to option
expensing. This loss in utility is entirely due to the executive’s perceived cost of option expensing as
opposed to actual economic cost. Nevertheless, it alters the optimal contract between the firm and the

1
  Prior research by Aboody (1996), Dechow, Hutton, and Sloan (1996), Aboody, Barth, and Kasznik (2004a), and
Balsam, Bartov, and Yin (2004) lends support to the efficient market argument. They show that the stock market
correctly incorporates stock option disclosures and there is no evidence of any significant negative market reaction
around past events that change the likelihood of mandatory option expensing.
2
  The Standard and Poor’s estimates that the total option expense for the S&P 500 companies in fiscal 2002 is
approximately $6.00 per share or 17% of estimated earnings.


                                                                                                                       1
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



executive and leads to a reduction in optimal equity incentives, especially option incentives. Our
theoretical model thus provides clear predictions on the impact of option expensing on executive
compensation. In particular, if a favorable accounting treatment leads to the overuse of options in
executive compensation, the removal of the favorable treatment is expected to have the opposite effect
and leads to a reduction in the use of options.
           To test the empirical predictions of our theoretical model, we analyze Chief Executive Officer
(CEO) compensation data from 1993 to 2004. We first examine CEO equity incentives (i.e., the sum of
both option incentives and stock incentives) and investigate whether firms reduce such incentives upon
mandatory option expensing. As the likelihood of option expensing approaches certainty by the end of
2002,3 we hypothesize that most firms begin downward adjustments to CEO equity incentives after 2001.
Indeed, we find that the use of equity incentives by U.S. firms peaks in 2001 and declines every year
afterwards. For the median CEO in our sample, the equity incentives provided by his annual stock and
option grants in 2004 are approximately 31% lower than they are in 2001. This translates into an annual
decline of 11.7%, which is in contrast to an annual increase of 21.8% in CEO equity incentives from 1993
to 2001.
           In addition, the decline in equity incentives after 2001 is entirely driven by a decline in the use of
option incentives because stock incentives actually rise throughout the entire sample period. The median
CEO option incentives increase every year from 1993 to 2001 at an average annual rate of 24.6%. After
peaking in 2001, they decline every year afterwards at an average annual rate of 17.7%. In comparison,
the median CEO stock incentives are zero throughout the sample period. In fact, more than 75% of all
CEOs in our sample do not receive any restricted stock grant at all in any given year. Nevertheless, CEO
stock incentives do appear to rise in the sample period, with the mean CEO stock incentives rising 13.2%
a year during the 1993-2001 period and 16.3% during the 2002-2004 period. Both trends in option and
stock incentives are further verified using multivariate regression analysis after controlling for expected
equity incentives. These results are consistent with the hypothesis that mandatory option expensing
reduces the attractiveness of options relative to restricted stock, inducing a substitution effect between
options and restricted stock. However, the annual increase in stock incentives after 2001 is only
marginally higher than it is before, not nearly enough to offset the decline in option incentives.
           The changes in option and stock incentives we document are generally consistent with the
findings reported in other studies such as Brown and Lee (2007) and Carter, Lynch and Tuna (2007). An
innovative feature of our study is that we utilize multivariate regressions to control for the expected
incentive pay and analyze trends in the use of incentive pay over a longer sample period (1993–2004).
This allows us to document a sharp reversal in CEO option incentives around 2001 while other research

3
    We clarify this point subsequently in the next section.


                                                                                                               2
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



mostly focuses on changes in either the pre-2002 period or the post-2001 period. The contrasting patterns
around 2001 we document highlight a sharp break in the trend of option compensation that is consistent
with changes in the accounting treatment of stock options. Another distinction of our research is that we
focus on option and stock incentives as measured by Jensen and Murphy’s (1990) pay-performance
sensitivity while other studies tend to focus on dollar values of stock and option compensation. As Core
and Guay (1999) argue, equity-based pay has both a compensatory component and an incentive
component. By focusing on the incentive component, we can draw inferences on how option expensing
may influence the optimal contract between firms and their executives.4
        More importantly, we recognize the possibility that the sharp reversal in CEO option incentives
may not be related to mandatory option expensing at all. The relationship could be merely coincidental
and caused by other factors we have yet to identify. For example, the Sarbanes-Oxley Act of 2002
requires U.S. firms to disclose more detailed compensation data of their top executives in a more timely
manner, which may force some or most firms to cut back on option grants made to CEOs, other top
executives and lower level employees. The documented changes in option incentives are thus also
consistent with the expected impact of the Sarbanes-Oxley Act of 2002.
        To address this possibility, we evaluate competing hypotheses and provide further evidence on
the relationship between option expensing and the reduction in CEO option incentives. First, we examine
cross-sectional differences in CEO option incentives between firms that are more likely to be impacted by
option expensing and the remaining firms in the sample. We do this by dividing our sample of firms into
“high” and “low” incentive firms. High incentive firms provide their CEOs with equity incentives that are
larger than other firms do in peer control groups. We hypothesize that firms granting “excessive” levels of
equity incentives prior to option expensing (i.e., high incentive firms) might have underestimated the cost
of options (e.g., Hall and Murphy, 2003) and are expected to make larger reductions in the use of option
incentives in the years leading up to mandatory option expensing. Indeed, our empirical evidence supports
this hypothesis. After controlling for commonly used determinants of CEO equity incentives, we find that
high incentive firms make much larger reductions to CEO option incentives in the post-2001 period than
other firms do. The difference is both statistically and economically significant. In comparison, both types
of firms make similar changes in stock incentives, with the difference statistically insignificant in the pre-
2002 period, in the post-2001 period, and between the two subperiods. We interpret this finding as further
evidence that the decline in option incentives after 2001 is related to the recent change in option
expensing rules.



4
  To ensure robustness, we also examine changes in dollar values of CEO stock and option compensation and find
similar patterns around 2001.


                                                                                                                 3
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



        Secondly, we investigate the possibility that other concurrent regulatory, business or market
events are behind the documented decline in option incentives after 2001. In particular, we consider
several well-known important events such as the Sarbanes-Oxley Act of 2002, the option backdating
scandal, and the 2000 stock market crash. As discussed before, the passage of the Sarbanes-Oxley Act in
2002 may force some or most firms to cut back on CEO option incentives. The bulk of all identified
option backdating problems also happen to occur prior to the passage of the Sarbanes-Oxley Act (e.g.,
Heron and Lie, 2006), which might have inflated option values prior to 2002. The subsequent reduction in
option incentives may thus be related to the disappearance of option backdating problems after the
passage of the Sarbanes-Oxley Act. The 2000 stock market crash is one of the largest in history and has
had a prolonged bearish impact on stock prices for many years afterwards. All these events take place
around our event year and could have contributed to the documented decline in option incentives. We
perform extensive empirical tests and find that these events contribute to but do not fully explain the
downward adjustments in CEO option incentives.
        Another competing hypothesis is that the decline in the use of option incentives is driven by
voluntary expensing firms. Seethamraju and Zach (2004) and Carter, Lynch and Tuna (2007) find
evidence that voluntary expensing firms make larger reductions in option grants made to their CEOs than
control firms do. If these firms have other motives for beginning expensing earlier than required (e.g.,
signaling transparency or better corporate governance), then mandatory option expensing is perhaps just a
catalyst instead of the root cause of the decline in option incentives. Using a dummy variable to identify
firms that voluntarily begin to expense options by February 2004, we find no difference in the use of
option incentives between voluntary expensing firms and other firms. Both types of firms have similar
inverted U-shaped patterns in CEO option incentives. The decline in the use of option incentives is thus
not driven by voluntary expensing firms. Interestingly, we do find evidence that voluntary expensing
firms provide their CEOs larger (smaller) option pay (in dollar values estimated using the Black-Scholes-
Merton model) than other firms do before 2002 (after 2001). This is an indication that voluntary
expensing firms may have adopted poorly designed option plans before 2002 and provide the same level
of incentives at greater costs than other firms do. This problem appears to be resolved upon voluntary
expensing, with these firms making necessary adjustments to make their option plans as effective as those
in other firms.
        In addition, we examine whether firms make similar or different reductions in option incentives
to top executives and lower level employees upon option expensing. On the one hand, an option has the
same impact on reported earnings whether or not it is granted to the CEO or an assembly line worker.
Firms are thus likely to make similar reductions in option incentives to all recipients. On the other hand,
firms grant stock options for retention as well as incentive purposes. For lower level employees in



                                                                                                         4
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



particular, stock options are likely granted primarily for retention rather than incentive purposes (Kedia
and Mozumdar, 2002; and Oyer, 2004). In that case, firms are probably more constrained in their ability
to cut back option grants made to these lower level employees, especially relative to peers in control firms.
Empirically, we find evidence that firms make similar reductions to option grants made to the CEO, other
top executives and lower level employees. This is consistent with the expected impact of option
expensing, but inconsistent with the retention hypothesis.
        Finally, we perform further robustness analysis in order to ensure the validity of our findings. Our
key inferences are not materially affected if we (1) use alternative classifications of high and low
incentive firms; (2) use 2001 or 2000 as the event year instead of 2002; (3) use median regressions to re-
estimate our main empirical relationships; (4) include only firms without CEO turnovers; (5) exclude
firms in regulated industries.
        The rest of the paper proceeds as follows. Section 2 summarizes changes in the accounting
treatment of stock options and the potential impact on executive compensation. In Section 3, we develop
an optimal contracting model incorporating the impact of mandatory option expensing and discuss
empirical predictions of our model. Section 4 describes our sample selection and empirical methodology.
Section 5 documents the changes in executive compensation before and after our event year of 2002 and
examines the relationship between these changes and mandatory option expensing. Section 6 presents
robustness and additional analysis to ensure the validity of our main findings. The final section concludes.


2. Option expensing and its impact on executive compensation
        Accounting for stock options has been one of the most hotly debated issues in the history of the
Financial Accounting Standards Board (FASB). In 1993, the FASB issued an Exposure Draft proposing
to recognize the stock option expense using the fair value method instead of the previously adopted
intrinsic value method. The fair value method specifies that the cost of a stock option is equal to the
option value estimated using the Black-Scholes-Merton (1973) model or its binomial variations on grant
date. The fair value of the option is to be expensed in the income statement over the option’s service
period (usually the vesting period). As most options are granted at the money, the cost of the option is
mostly zero if the intrinsic value method is used. A switch to the fair value method would have resulted in
a large jump in the estimated option value, leading to a potentially large downward revision to reported
income. The recognition provision of the Exposure Draft was confronted with unprecedented opposition
and lobbying efforts from U.S. companies, especially those in the technology industries.
        As a compromise, the FASB issued Financial Accounting Standards 123 (SFAS 123) in 1995:
Accounting for Stock-Based Compensation, which requires firms to apply the fair value method to
evaluate their option grants. However, firms do have discretion to either recognize the fair value of option



                                                                                                          5
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



grants in the income statement or disclose it in footnotes. Not surprisingly, nearly all firms choose to
disclose the fair value in footnotes instead of expensing it in reported income. By July 2002, only 11
firms had voluntarily chosen to expense stock options according to a Bear Stearns report.5 Following
several high-profile corporate scandals (e.g., Enron, WorldCom, and Global Crossing), more U.S. firms
decided to voluntarily adopt option expensing, perhaps in an effort to signal transparency and good
corporate governance. In December 2002, the FASB issued Financial Accounting Standards 148 (SFAS
148): Accounting for Stock-Based Compensation – Transition and Disclosure, to provide guidance for the
transition from disclosure to recognition. The FASB’s resolve to imposing mandatory option expensing in
the near future was much more apparent at that point. Many U.S. firms appear to recognize this resolve
and prepare for the eventual adoption of option expensing. By the end of 2002, about 170 public traded
firms had voluntarily adopted the recognition provision under SFAS 123.6
        The FASB issued another Exposure Draft in March 2004, proposing the expensing of stock
options on grant date. The Exposure Draft requires the recognition of stock options in the income
statement at the grant-date fair value. It encourages using a lattice model and stipulates the recognition of
option costs over the requisite service period, generally the vesting period. The proposed effective date is
December 15, 2004 for all public firms. Finally, the FASB published in December 2004 the revised
FASB Statement No. 123 (SFAS 123R): Share-Based Payment, which requires companies to recognize
stock option costs based on the options’ grant-date fair value. However, the lattice model is no longer
preferred and companies generally have much flexibility in their choice of valuation models, stock
volatility estimation methods and other elements of option expensing. SFAS 123R became effective for
public firms as of the beginning of the first interim or annual reporting period that begins after June 15,
2005—the third quarter of 2005 for calendar year companies.7
        The strong opposition to mandatory option expensing by U.S. firms, which played a key role in
shaping the recognition requirement until its final adoption, is puzzling. In an efficient market, option
expensing should not induce any direct economic costs to firms as it merely moves the cost of stock
options from footnotes to the income statement. It does not appear to materially change the nature of the
information firms are required to disclose about their stock option grants. If the market has already


5
  These firms are Alabama National Bancorporation, Boeing Co., Entropin Inc., Level 3 Communications Inc.,
MacDermid Inc., Mercantile Bankshares, Protective Life, RCN Corp., Star Scientific Inc., TCF Financial Corp., and
Winn-Dixie Stores Inc.
6
  A Bear Stearns report dated February 12, 2004.
7
  In April 2005, SEC postponed the implementation date of SFAS 123R by 6 months for calendar year companies.
SEC stated that the mandatory compliance will be delayed until the first interim or annual reporting period of the
company’s first fiscal year beginning on or after June 15, 2005 – the first quarter of 2006 for calendar year firms.
But this postponement will not affect firms such as Cisco and Sun Microsystems, whose fiscal year ends in July 30th
and June 30th respectively. These firms have to adopt mandatory option expensing in the second half of 2005 for
their first quarter of fiscal year 2006.


                                                                                                                  6
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



incorporated all publicly available information in determining firm value, option expensing should not
have any economic impact on firm performance (either operationally or financially) other than reducing
reported earnings. There is some empirical evidence supporting the argument that the stock market
adequately incorporates disclosures associated with stock options.8 For example, Balsam, Bartov, and Yin
(2004), studying new quarterly disclosures filed by a sample of 302 firms, find that there is no significant
difference in market reactions whether the cost of stock options is disclosed in footnotes or recognized in
the income statement. In addition, a number of studies (e.g., Dechow, Hutton, and Sloan, 1996; Daniel,
Kale, and Naveen, 2003; Aboody, Barth, and Kasznik, 2004b; and Elayan, Pukthuanthong, and Roll,
2005) examine the announcement effect of option expensing decisions, voluntary or otherwise, and find
no evidence of any negative stock market reaction.
           Nevertheless, many executives appear to believe that there are real economic costs associated
with option expensing. A recent survey by Graham, Harvey, and Rajgopal (2005) finds that most CFOs
believe that the market pays more attention to the income statement than to financial footnotes. Although
there is empirical evidence that the market correctly incorporates the information disclosed in the
footnotes, many executives seem to behave as if the opposite is true.9
           One possible explanation for such executive behavior is the impact of option expensing on
reported earnings. As estimated by the Standard and Poor’s, mandatory option expensing would have
resulted in an immediate 17% decline in reported earnings for S&P 500 companies if it were implemented
in 2002. Executives may dislike such a sharp decline in reported earnings for several reasons. First,
performance pay such as bonuses and stock options is often contingent upon reaching certain target levels
of earnings. If executives believe that any reduction in reported earning will have a negative impact on
their current and future compensation, option expensing is clearly an undesirable development for them.
Secondly, executives may also dislike option expensing because of their concerns with greater visibility
and transparency of their compensation triggered by moving the cost of stock options from footnotes to
the income statement. Much of the campaign for option expensing has been fuelled by the public
concerns that excessive stock option grants caused executives to manipulate accounting numbers. Afraid
of further pressure from shareholders and other stakeholders, executives have incentives to “camouflage”
the true nature and magnitude of their compensation (e.g., Bebchuk, Fried, and Walker, 2002; and
Bebchuk and Fried, 2003). Finally, executives may object to option expensing due to other financial
reporting concerns that can negatively impact their compensation (e.g., Carter, Lynch and Tuna, 2007).
These concerns may arise from the desire to maintain the pattern of increasing earnings, the pressure to



8
    See Aboody (1996), Aboody, Barth, and Kasznik (2004a), and Balsam, Bartov, and Yin (2004).
9
    See Core, Guay, and Larcker (2003) and Oyer and Schaefer (2005b).


                                                                                                          7
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



meet analyst expectations, and the need to meet contractual agreements (covenants) with debtholders.
Any downward pressure on reported earnings due to option expensing may intensify these concerns.
        In this paper, we do not attempt to explain why many executives dislike option expensing. We
simply take such executive behavior as given and examine its impact on executive compensation. We
modify the standard principal-agent model by incorporating the executive’s perceived personal loss from
option expensing. Even if there is no economic cost to the firm that is directly associated with option
expensing, the executive’s perceived personal loss may in fact alter the optimal contract between firms
and their executives, leading to a decline in the use of option incentives in executive compensation.
        This perceived personal loss from option expensing is consistent with a related argument by Hall
and Murphy (2003) on executive beliefs prior to the current change in option expensing rules. Many
executives were convinced that stock options were cheap because of little or no accounting cost and no
cash outlay for the firm prior to mandatory option expensing and that such inaccurately perceived cost of
stock options is much lower than the real economic cost of options. When the new accounting regulation
removes the favorable accounting treatment of stock options, it brings executives’ perceived cost of
options much closer to options’ real economic costs. Our interpretation of the executives’ perceived
personal loss from option expensing is then equivalent to a jump in the executives’ perceived cost of
stock options upon mandatory option expensing. Hall and Murphy (2003) argue that the divergence
between executives’ perceived cost and the true economic cost of options (prior to option expensing)
largely accounted for the dramatic surge in option grants over the previous decade. Likewise, we expect
that the convergence of the executive’s perceived cost of options to the true economic cost upon
mandatory option expensing should lead to a decline in the corporate use of stock options after or in
preparation for mandatory option expensing.


3. An optimal contracting model with option expensing
        Consider a standard principal-agent model with a risk-neutral principal (owner) and a risk-averse
agent (manager), similar to the ones developed in Mirrlees (1976) and Holmstrom (1979). The owner
hires the manager to run the firm but cannot observe his actions, abstracted by a variable a ≥ 0 which we
call effort for convenience. The principal contracts with the agent on the outcome of the agent’s effort.
        The outcome is proxied by the change in firm value over an evaluation period which is influenced
by the agent’s effort as follows:
                                             ∆Vt +1 ≡ Vt +1 − Vt = Vt Rt ,                                  (1)

where Vt is the firm value on date t, Rt is the rate of return to firm value in period t:
                                                  Rt = f (a) + ε t ,                                        (2)




                                                                                                             8
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



a is the agent’s effort, f is the agent’s productivity function with f′(a) ≥ 0, and εt is white noise with zero
mean and variance σ2.
        The principal pays the agent both a fixed component (e.g., salary) and an incentive component
(e.g., stock) and the agent’s wealth on date t+1 is:
                                    Wt +1 = Wt + s + b∆Vt +1 = Wt + s + bVt Rt ,                           (3)

where s is the fixed component, b is the incentive component, and Wt is the agent’s wealth on date t.
Without loss of generality, we assume that Wt is a fixed amount while b captures all incentives in the
agent’s portfolio. Combining with Eq. (2), we have
                                         Wt +1 = Wt + s + bVt [ f (a) + ε t ] .                            (4)

        The agent’s welfare is further affected by his effort aversion and perceived loss associated with
option expensing and his net wealth on date t+1 is downward adjusted as follows:
                                         π t +1 = Wt +1 − c(a) − µδ (b∆Vt +1 ) ,                           (5)

where c(⋅) is disutility of effort, µ is the agent’s perceived loss for each dollar of options expensed, δ is
the fraction of the agent’s incentive pay that is being expensed. The disutility of effort should exhibit the
property c′(a) ≥ 0, c′′(a) ≥ 0, δ is always between 0 and 1, and µ should be either zero or a small positive
number. Combining with Eqs. (1)–(4), we can rewrite the agent’s net wealth on date t+1 as
                                 π t +1 = Wt + s + (1 − µδ )b∆Vt +1 − c( a)
                                                                                                           (6)
                                        = Wt + s + (1 − µδ )bVt [ f ( a) + ε t ] − c (a ).
        The agent is also risk averse and has negative exponential utility. Stated in certainty equivalent
terms, his expected utility can be written as follows:
                                                             1
                                        CEAt = Et (π t +1 ) − γ ⋅Vart (π t +1 ) ,                          (7)
                                                             2
where γ is the agent’s coefficient of absolute risk aversion. The agent’s expected utility, stated in certainty
equivalent amount, is thus
                                                                           γ
                        CEAt = Wt + s + (1 − µδ )bVt f ( a ) − c(a ) − [(1 − µδ )bVt ]2 σ 2 .              (8)
                                                                      2
        For a given package of fixed pay and incentives (s and b), the agent maximizes his expected
utility over his choice of effort. The first order condition (FOC) is
                                          (1 − µδ )bVt f ' ) − c ' ) = 0 .
                                                         (a      (a                                        (9)

The FOC for the agent’s optimal effort is quite intuitive as it simply states that the marginal benefit of
effort is equal to the marginal cost of effort at the optimal level of effort. The optimal effort cannot be
solved explicitly in general unless we make further assumptions about the agent’s productivity and effort
aversion.


                                                                                                             9
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



          In comparison, the principal wishes to maximize the increase in firm value and do so by choosing
the agent’s fixed pay and incentives (i.e., s and b). Aside from the inability to observe the agent’s effort,
the principal has full information on the agent’s utility and productivity functions, effort aversion (η),
perceived personal loss from option expensing (µ and δ), and reservation utility (certainty equivalent
value W0).
          The principal’s net payoff on date t+1 is
                                         ∆Vt +1 − ∆Wt +1 = (1 − b)∆Vt +1 − s .                           (10)

The principal is risk neutral and only cares about his expected payoff. In certainty equivalent terms, the
principal’s expected utility is thus
                                            CEPt = (1 − b)Vt f (a) − s .                                 (11)

The principal’s optimization problem is:
                                             max (1 − b)Vt f (a) − s ,                                  (12a)
                                              s ,b, a

subject to
                                          (1 − µδ )bVt f ' ) − c ' ) = 0 ,
                                                         (a      (a                                     (12b)

and
                                                                   γ
                            s + (1 − µδ )bVt f ( a ) − c ( a ) − [(1 − µδ )bVt ]2 σ ≥ W0 .              (12c)
                                                                2
Note that both the incentive compatibility (12b) condition and participation constraint (12c) are needed as
the agent maximizes his own utility and will not take the job unless his reservation utility is met. In
addition, the certainty equivalent value W0 (which defines the agent’s reservation utility) should be
interpreted as a cash amount net of the agent’s initial wealth from the beginning of the period (Wt).
          We also make the simplifying assumption that the agent has linear productivity and quadratic
effort:
                                                         f (a ) = ka ,                                   (13)
and
                                                                 1
                                                        c ( a ) = ηa 2 ,                                 (14)
                                                                 2
where all coefficients are nonnegative constants. This allows us to solve the optimal contract explicitly:
                                                                 1
                                         b* =                                     ,                      (15)
                                                                         γησ 2
                                                2 − (1 − µδ ) 1 −
                                                                            k2

                                                          (1 − µδ )Vt k
                                                a* =                       b* ,                          (16)
                                                               η


                                                                                                             10
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



and
                                                      1    2 1
                         s* = W0 − (1 − µδ )b*Vt ka* + ηa * + γ [(1 − µδ )b*Vt ]2 σ 2 .                  (17)
                                                      2      2
It is clear that under optimal contracting, the incentive component (b*) is positive (as long as the agent is
productive), and is increasing in the agent’s marginal productivity (k) and decreasing in the agent’s effort
aversion (η), risk aversion (γ) and firm volatility (σ).
        To see how option expensing influences the optimal contract, we rewrite the optimal contract
terms as
                                                           k2
                                      b* =                                    .                          (18)
                                             k 2 + γησ 2 + ( k 2 − γησ 2 ) µδ

It is straightforward to show that for a productive agent (i.e., k > γη σ ), the incentive component (b*) is

a negative function of both option expensing parameters (µ and δ). This result directly implies that the
executive’s perceived personal loss from option expensing alters the optimal contract by reducing the
incentive component of the executive’s compensation. Our model thus leads to a couple of interesting
testable hypotheses regarding the impact of mandatory option expensing on executive compensation.
        Hypothesis 1: Mandatory option expensing should lead to a decline in the use of equity-based
incentives in executive compensation. The decline should be greater (smaller) for firms using higher
(lower) levels of equity-based incentives (relative to comparable firms) prior to mandatory option
expensing.
        Prior to mandatory expensing, the optimal contract between the firm and the executive maximizes
firm value while satisfying the executive’s incentive and participation constraints. This is achieved by
offering an optimal level of equity incentives that induces the executive’s best effort at the lowest possible
cost to the firm. Mandatory option expensing renders the compensation package contracted previously
suboptimal and a new compensation package is contracted at a reduced level of equity incentives. For
firms using higher levels of equity incentives previously, the cost of option expensing is larger than for
other firms and we expect larger reductions in the use of equity incentives for these firms upon option
expensing.
        Hypothesis 2: Mandatory option expensing should lead to a decline (rise) in the use of stock
options (restricted stock) as equity-based incentives in executive compensation. The impact should be
greater (smaller) for firms using higher (lower) levels of equity-based incentives (relative to comparable
firms) prior to mandatory option expensing.
        As the executive’s perceived personal loss from option expensing is associated only with the
option component of the equity incentives, it is obvious that the reduction in equity incentives should
come mainly from a reduction in option incentives. For a given increase in the executive’s effort, the firm


                                                                                                           11
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



needs to offer more options after mandatory option expensing than it does before. On the other hand, the
cost of restricted or unrestricted stock grants is already required to be recognized even before mandatory
option expensing. For a given increase in the executive’s effort, the same number of shares is thus
required before or after mandatory option expensing. We thus expect firms to reduce the fraction of
equity incentives provided by stock options and simultaneously increase the fraction provided by
restricted stock.


4. Sample selection and empirical methodology
        We pool data from the CRSP (for stock returns), Standard and Poors’ ExecuComp (for executive
compensation) and Compustat (for accounting data) over the sample period from 1993 to 2004. We start
with a universe of 2,704 firms covered by the ExecuComp which compiles executive compensation data
for S&P 1500 firms. Some firms do not report CEO compensation, have no matching accounting data
(from Compustat) or stock return data (from the CRSP), or have the aggregate option grants for the top
five executives exceeding 100% of all options granted that year10 in one or multiple years of the sample
period. After removing these firms, 1,609 firms or roughly 60% of the original firms remain in the sample.
        We measure the strength of equity incentives provided by the CEO’s equity holdings using
Jensen and Murphy’s (1990) pay-performance sensitivity (PPS) which calculates the change in the value
of the CEO’s equity holdings for a $1,000 increase in firm value. We adopt this dollars-on-dollars
measure of equity incentives instead of Hall and Liebman’s (1998) dollars-on-percentage measure
because the former is directly related to our theoretical model (equivalent to the incentive pay parameter b
in Eq. (3)). It also can be loosely interpreted as a fractional equity ownership measure, which is less likely
to be affected by stock market movement (since a 1% ownership is still a 1% ownership even if the stock
price doubles or drops in half).
        We calculate separate incentive measures for the CEO’s stock holdings (STK_PPS) and option
holdings (OPT_PPS). The stock incentive measure is calculated as the ratio of restricted stock grants over
the total number of shares outstanding. The option incentive measure is calculated as the ratio of option
grants over the total number of shares outstanding, multiplied by the Black-Scholes-Merton hedge ratio.
We also calculate the CEO’s total equity incentives (TOT_PPS) by combining stock and option incentives
(STK_PPS + OPT_PPS).
        As in previous research, we hypothesize that firms optimally contract with executives in order to
maximize firm value given the particular business, economic and regulatory environment they operate in.

10
  Core and Guay (2001) discuss a few reasons that cause this problem: “… because of missing data or coding errors
in ExecuComp, because firms attribute grants made after the fiscal year-end to the current fiscal year, and because
options grants to the top-five executives include grants of options on subsidiary stock made to the top-five
executives.”


                                                                                                               12
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Such environment is abstracted by a set of control variables such as industry, firm size, growth
opportunities, leverage, past firm performance, CEO and board characteristics. When firms are required
to expense stock options, the executive’s perceived personal loss from option expensing may lead to a
new (lower) optimal level of equity incentives as our theoretical model predicts. This change in optimal
equity incentives can be estimated using a multivariate regression (with firm-fixed effects) based on a
piecewise linear specification of the time trend:
                           Compit = α i + β1TL1t + β 2TL2t + (control variables)it + eit ,                  (19)

where Comp is an incentive measure (e.g., the CEO’s total equity incentives), t is a year index (taking
values of 1, 2, 3, …), TL1 is equal to t before 2002 and zero after 2001, TL2 is equal to zero before 2002
and (t – t2001) after 2001, and i is a firm index.
        We choose 2002 as the event year for several reasons. Although U.S. firms are not required to
expense stock options until the first reporting period after June 15, 2005, many firms begin to prepare for
the new accounting regulation several years ahead of time. The passage of Sarbanes-Oxley Act in July
2002 and the release of SFAS 148 by the FASB in December 2002 are two important milestones in
accounting regulation and corporate governance. More and more firms realize that mandatory option
expensing is inevitable and begin or prepare to adopt it in the near future. Prior to July 2002, only 11
firms choose to expense stock options voluntarily. By the end of 2002, that number jumps to 170. We
thus choose the year 2002 as our event year and investigate the use of equity incentives before and after
the event year.
        Note that the piecewise linear specification in Eq. (19) is similar to the one used in Morck,
Shleifer and Vishny (1988). It is designed to capture a separate (linear) trend in the two periods before
and after the event year. While β 1 reflects the yearly change in executive compensation in the pre-2002
period (beyond what is explained by the control variables), β 2 captures the corresponding yearly change
in the subsequent time period. Our theoretical model predicts that mandatory option expensing would lead
to a downward shift in equity incentives (i.e., β 2 < β 1). The null hypothesis (i.e., mandatory option
expensing has no impact on equity incentives) is thus β1 = β 2.
        The piecewise linear specification in Eq. (19) requires a pre-specified event year (e.g., 2002 in
our case). Although we do evaluate the robustness of using 2002 as the event year subsequently, it is
desirable to consider an alternative specification that does not require an event year at all. The following
quadratic specification is thus used as well in our empirical analysis:
                              Compit = θ i + γ 1t + γ 2t 2 + (control variables)it + eit .                  (20)

The advantage of the quadratic specification in Eq. (20) is that it does not specify the location of the
inflection point (i.e., the timing of the shift). Instead, the timing of the change in optimal incentives, if any,



                                                                                                              13
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



is endogenized and estimated empirically. With this quadratic specification, the coefficient of the
quadratic term should be zero (i.e., γ2 = 0) if there is no change at all in the use of equity incentives.
Alternatively, CEO equity incentives may exhibit either a U-shaped (positive γ2) or an inverted U-shaped
(negative γ2) time trend. We should find γ1 > 0 and γ2 < 0 if CEO equity incentives initially rise and then
fall during the sample period.
        Following prior research (e.g., Demsetz and Lehn, 1985; Smith and Watts, 1992; Hall and
Liebman, 1998; Murphy, 1999; Gibbons and Murphy, 1992; Ittner, Lambert, and Larcker, 2003;
Milbourn, 2003; and Baker and Hall, 2004), we use a variety of control variables in our subsequent
regression analysis. In equilibrium, firms optimally determine executive compensation in order to
maximize firm value given the particular business, economic and regulatory environment they operate in.
Such environment is abstracted by a set of control variables reflecting firm, CEO and governance
characteristics, including firm size, growth opportunities, leverage, liquidity constraints, corporate taxes,
stock return volatility, past stock returns, CEO tenure, and corporate governance. By including these
control variables, we isolate the marginal impact of option expensing on equity incentives.
        Firm size (FIRM_SIZE) is measured as the logarithm of the market value of equity.11 Previous
research suggests that firm size is an important determinant of equity incentives. Large firms are more
challenging to manage than smaller firms and thus are expected to provide greater equity incentives (and
smaller cash pay) to their CEOs (e.g., Demsetz and Lehn, 1985; and Jensen and Murphy, 1990). This does
not necessarily translate into a positive relationship between firm size and PPS. The greater equity
incentives provided to CEOs at large firms may actually represent a smaller fractional equity ownership,
compared to CEOs at smaller firms. As PPS (our measure of equity incentives) is essentially a measure of
fractional equity ownership, it is likely to have a negative relationship with firm size (e.g., Morck,
Shleifer, and Vishny, 1988; McConnell and Servaes, 1990; Smith and Watts, 1992; Schaefer, 1998; and
Baker and Hall, 2004).
         Growth opportunities (GROWTH_OPP) are proxied by the market-to-book ratio. Firms with
abundant investment opportunities are expected to face a more complex and larger set of managerial
decisions. It is thus more difficult for shareholders to monitor managerial actions due to information
asymmetry. These firms are more likely to use equity-based compensation to motivate managers (e.g.,
Smith and Watts, 1992; Bizjack, Brickley and Coles, 1993; Gaver and Gaver, 1993; Ittner, Lambert, and
Larcker, 2003; and Murphy, 2003). Everything else being equal, the CEO’s equity incentives should be
positively related to growth opportunities.



11
  We also create a proxy for firm size with the logarithm of sales to ensure the robustness of the size variable. All
qualitative inferences from the multivariate analysis remain.


                                                                                                                 14
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



           Financial leverage (LEVERAGE) is defined as total debt divided by total assets. Another agency
problem is the conflict between shareholders and debtholders. If firms choose equity incentives to align
the interests of managers and shareholders, debtholders are likely to suffer the brunt of the agency costs as
managers make decisions to maximize shareholder value at the expense of debtholders. Such firms may
face higher interest rates and/or more restrictive covenants when they borrow in the future. In equilibrium,
equity incentives should be optimized to minimize agency costs that arise from both manager-shareholder
and bondholder-shareholder conflicts, with highly leveraged firms using lower levels of managerial
equity incentives and vice versa (e.g., John and John, 1993; and Yermack, 1995).
           Liquidity constraints are proxied by a measure of cash flow shortfall (CASH_SHORT), calculated
as the three-year average of the sum of common and preferred dividends plus the cash flow used in
investing activities minus the cash flow generated from operations, normalized by total assets. Equity-
based compensation does not involve any cash outflows on the grant date and subsequent option exercises
generate cash inflows on the exercise date. Firms facing liquidity constraints are thus more likely to use
stock options and restricted stock to compensate their managers (e.g., Smith and Watts, 1992; Yermack,
1995; Dechow, Hutton, and Sloan, 1996; Core and Guay, 1999, 2001; and Kedia and Mozumdar, 2002).
           Corporate tax rate is measured as the marginal tax rate (MTR) based on income before interest
expense calculated by John Graham, downloaded from his personal website at Duke University. 12
Corporate tax rates may influence the use of equity incentives in executive compensation. While cash pay
is generally immediately tax deductible, equity-based compensation only provide a deferred tax deduction
(when options are exercised or when restricted stock becomes unrestricted). Consequently, firms facing
higher marginal tax rate are likely to use less equity-based pay and more cash pay and vice versa (e.g.,
Matsunaga, 1995; Yermack, 1995; Dechow, Hutton, and Sloan, 1996; Core and Guay, 2001; and Kedia
and Mozumdar, 2002).
           Firm risk is proxied by stock return volatility (STK_VOL), calculated as the standard deviation of
monthly stock returns over the past 36 months. Managerial risk aversion and the need for diversification
may reduce the effectiveness of equity incentives. At high levels of firm risk, the marginal cost of
providing equity incentives may exceed their marginal benefit. The tradeoff between managerial risk
aversion and incentives may induce a negative relationship between firm risk and managerial incentives.
On the other hand, firm risk may be a proxy for growth opportunities which suggests a positive
relationship between firm risk and managerial incentives as discussed previously. Empirical evidence is
however mixed, with some studies finding a positive relationship (e.g., Demsetz and Lehn, 1985; and
Core and Guay, 1999) while others a negative relationship (e.g., Aggawal and Samwick, 1999; Palia,
2001; and Jin, 2002).

12
     The URL of John Graham’s website is: http://faculty.fuqua.duke.edu/~jgraham/taxform.html


                                                                                                          15
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



         Past firm performance is proxied by past stock returns (PAST_RET), calculated as the cumulative
stock return during the past 36 months. Some firms may make future stock and option grants contingent
on firm (accounting or stock) performance. Barber, Janakiraman, and Kang (1996) find that equity-based
compensation is larger following better firm performance. In comparison, Murphy (2003) finds that new-
economy firms make more new option grants after the 2000 NASDAQ crash. Thus, we use past
performance to control for the impact of past firm performance on compensation variables.
        CEO experience is proxied by CEO tenure (CEO_TENURE) and measured as the number of
years the CEO has been in the firm’s corner office. Managerial equity incentives may vary with
experience on the job. Milbourn (2003) argues that managerial experience is reflected in his reputation
which supports a positive relation between CEO tenure and equity incentives. On the other hand, Carter,
Lynch, and Tuna (2007) find a negative relationship between CEO tenure and equity incentives.
Entrenched CEOs may become more risk averse (e.g., the horizon problem) and require less equity
incentives. Therefore, we apply CEO_TENURE and its square value CEO_TENURE2 to capture the
nonlinear relationship between CEO tenure and compensation variables.
        Corporate governance is proxied by the governance index (GIM) constructed by Gompers, Ishii
and Metrick (2003). Core, Holthausen, and Larcker (1999) find that CEO compensation can be affected
by board characteristics and ownership structure. They argue that less efficient corporate governance
mechanism leads to greater CEO compensation. Proponents of the managerial power theory (e.g.,
Bebchuk, Fried and Walker, 2002; and Bebchuk and Fried, 2003) provide a similar argument for the
impact of corporate governance on managerial compensation.
        Core and Guay (1999) argue that firms use annual stock and option grants to make adjustment to
executives’ equity incentives in order to correct deviations from optimal incentive levels. Even if an
executive is provided with an optimal level of equity incentive at the beginning of the year, he may take
actions during year (such as exercising vested option and sell or purchase shares of the stock) to make his
incentives deviate from the optimal level. The size and mix of the CEO’s stock and option grants may
thus depend on deviations from the optimal incentive levels. Following Core and Guay (1999), we
estimate the deviation from the optimal incentive level for CEOs (INCT_RES) and use it to control for
incentive adjustments.13
        Finally, we add dummy variables to control for the impact of CEO turnovers. In the event of a
CEO turnover, the CEO on record for that year may be either the incoming or the departing CEO
depending on the timing of the turnover during the fiscal year. The incoming CEO may also have a


13
  See Core and Guay (1999) for further details of the estimation. The difference between our estimation and the
Core and Guay (1999)’s approach is that we use Jensen and Murphy (1990)’s measure of incentives and we also
control for firm fixed effects.


                                                                                                            16
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



different level of equity incentives compared to the departing CEO and the compensation for either CEO
may be just a fraction of what they are normally paid annually. We thus use a dummy variable to capture
the potential impact of CEO turnovers. In a robustness check, we also exclude firms with CEO turnovers
in 2001 or subsequent years. The results are not materially different.
         Table 1 presents summary statistics of compensation and control variables over the full sample
period. Of particular interest are the three incentive variables – the total equity incentives (TOT_PPS),
option incentives (OPT_PPS) and stock incentives (STK_PPS). The average total equity incentives for
CEOs in our sample is 2.16, suggesting that a $1,000 increase in firm value provides a $2.16 incentive for
the typical CEO. To provide this level of equity incentives, the CEO is paid an average of $4.6 million in
total compensation (TOT_COMP). It is also clear that most of the equity incentives are provided in the
form of option incentives. In more than 75% of the cases, CEO equity incentives are provided entirely by
option grants and none by restricted stock. These statistics highlight the importance of stock options in
providing equity incentives to CEOs.


5. The impact of option expensing on the use of equity incentives
         In this section, we examine the impact of mandatory option expensing on the corporate use of
equity incentives. We focus on two key research questions – 1) Do firms reduce their use of equity
incentives, especially option incentives, during the years leading up to mandatory option expensing? 2) Is
the reduction in equity incentives, if any, related to mandatory option expensing or merely coincidental?
We first document changes in the use of equity incentives over time and then explore the possible causes
for the changes.
5.1. Changes in equity incentives
         To illustrate changes in CEO equity incentives over time, we first analyze annual statistics of
CEO equity incentives over the period from 1993 to 2004. We calculate summary statistics separately for
each year during the sample period and report them in Table 2. As shown in Panel A of Table 2, the
median CEO’s total equity incentives (TOT_PPS) increases monotonically during the period 1993-2001,
rising from 0.27 in 1993 to 1.32 in 2001 (equivalent to a 21.8% increase per year), and then decreases
every year thereafter, declining from 1.32 in 2001 to 0.91 in 2004 (equivalent to a 11.7% decline per year).
The mean, first quartile and third quartile all exhibit similar inverted U-shaped patterns,14 suggesting a
sharp reversal in the use of equity incentives around 2001.




14
  The only minor difference is that the mean peaked in 2000 while the median (i.e., the second quartile) and the first
and third quartiles all peaked in 2001. This is likely due to outliers which tend to influence the mean but not the
quartiles.


                                                                                                                   17
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



        As total equity incentives are made up of option incentives and stock incentives, it is interesting
to find out which component is driving the sharp reversal in the use of equity incentives. Panels B and C
of Table 2 report the trend in option incentives (OPT_PPS) and stock incentives (STK_PPS), respectively.
While option incentives exhibit an even stronger inverted U-shaped pattern, stock incentives actually
experience a rising trend throughout the entire sample period. It thus appears that the sharp reversal in the
use of equity incentives is entirely driven by changes in option incentives. As shown in Panel B of Table
2, the median CEO option incentives increase every year from 1993 to 2001 with an average annual
increase of 24.6%. After peaking in 2001, it declines every year afterwards with an average annual drop
of 17.7%. In comparison, the median CEO does not receive any restricted stock grant in any of the 12-
year period. In this case, a better gauge for changes in CEO stock incentives can be gleaned from the
annual means. As shown in Panel C of Table 2, the mean CEO stock incentives exhibit an increasing
trend throughout the 12-year sample period, rising 13.2% a year during the 1993-2001 period and 16.3%
a year afterwards. Combining the annual statistics reported in all three panels, it is clear that most equity
incentives are provided by option grants while restricted stock grants account for only a small fraction of
the equity incentives received by CEOs. The sharp decline in equity incentives after 2001 is entirely
driven by changes in option incentives. If it were not for the 16.3% annual increase in stock incentives,
the decline in total equity incentives after 2001 would have been even greater. These results are consistent
with our theoretical predictions and the expected impact of mandatory option expensing.
        Although the inverted U-shaped patterns in total equity and option incentives are apparent from
the annual summary statistics in Table 2, further investigation is necessary since we have yet to control
for differences in firm and CEO characteristics. We need to find out whether the inverted U-shaped
patterns remain after the control variables are accounted for using multivariate regressions (19) or (20).
We also control for firm-fixed effects in the regression analysis since our focus is on the adjustments in
compensation contracts over time by each firm. By controlling for firm-fixed effects, we take into account
any unobserved firm-specific variables that are fixed for each firm over time but may influence
managerial contracting. The standard errors are corrected for clustering following Petersen (2006). In
order to reduce the impact of outliers on the coefficient estimation, we winsorize all continuous variables,
both dependent and independent, at the 1% and 99% levels.
        Table 3 reports the results of firm-fixed effect regressions of CEO equity incentives using
regressions based on Eqs. (19) and (20) over the period from 1993 to 2004. Model (1) is based on the
quadratic specification in Eq. (20) while Model (2) is based on the piecewise linear specification in Eq.
(19). In these regressions, we include various control variables that are known determinants of expected
equity incentives. Each regression is performed separately for total equity incentives, option incentives
and stock incentives. In all regression results reported in Table 3, coefficients for control variables have



                                                                                                          18
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



the expected sign and their statistical significance is generally consistent with prior research. For example,
the coefficients for firm size (FIRM_SIZE) and annual incentive adjustments (INCT_RES) are negative
and mostly significant at either the 1% or 5% level across all regressions. In comparison, the coefficients
for stock return volatility (STK_VOL), growth opportunity (GROWTH_OPP) and past stock returns
(PAST_RET) are all positive and mostly statistically significant. Interestingly, control variables for CEO
tenure, liquidity constraint, leverage, corporate tax rate and corporate governance do not explain much of
the variations in equity incentives and are generally statistically insignificant.
        More importantly, regression results in Table 3 provide further support for the sharp reversal in
the use of equity incentives, option incentives in particular, around 2001. Consider first the results from
the piecewise linear specification of the regression model. If there is no change in the time trend of equity
incentives at all between the two subperiods, the coefficients for TL1 and TL2 should be equal. As shown
in Table 3, the null hypothesis that TL1 = TL2 is rejected at the 1% level for total equity incentives
(TOT_PPS) and option incentives (OPT_PPS) and at the 5% level for stock incentives (STK_PPS). For
total equity incentives, the coefficient for the trend variable in the pre-2002 period (TL1) is 0.08 while the
corresponding coefficient in the post-2001 period (TL2) is –0.27, both are significant at the 1% level. It
suggests that the CEO’s total equity incentives are increasing before 2002, with an average increase of
0.08 per year, while the same incentives are decreasing after 2001, with an average decline of 0.27 per
year. The total downward adjustment in annual equity incentives (TL2 – TL1) over the two subperiods is
thus 0.35 (0.27 + 0.08) per year. Given the average (median) annual total equity incentives of 2.16 (0.87)
during the sample period (Table 1), the size of the downward shift is thus also economically significant,
representing a 16% (40%) decline in the mean (median) annual equity incentives for CEOs. For option
incentives, the trends in the two subperiods are similar, with a total downward shift of 0.36 per year. In
comparison, stock incentives experience an increasing trend in both subperiods, with an upward jump in
the annual growth rate in the second subperiod. These results thus continue to support a downward shift
in option incentives and an upward shift in stock incentives, even after we control for the determinants of
expected equity incentives.
        The quadratic specification of the regression model provides similar support for the downward
shift in total equity incentives, option incentives in particular. If there is no change in the use of equity
incentives between the two subperiods, the coefficient of the quadratic term (t2) should be zero. For total
equity incentives and option incentives, the results in Table 3 indicate that the coefficient for the linear
component (t) is positive while the coefficient for the quadratic component (t2) is negative. All
coefficients are significant at the 1% level, suggesting again an inverted U-shaped pattern. For stock
incentives, the coefficient for the linear component (t) is negative (but insignificant) while the coefficient
for the quadratic component (t2) is positive (and significant at the 5% level). This suggests an increasing



                                                                                                           19
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



trend in stock incentives, with a slightly rising rate of increase. Results from the quadratic specification of
the regression model are thus consistent with those from the piecewise linear specification.
        Although we focus on changes in CEO equity incentives, option incentives in particular, (as
measured by pay-performance sensitivity), other studies (e.g., Brown and Lee, 2007; Carter, Lynch and
Tuna, 2007) analyze changes in the dollar value of CEO compensation. For comparison purposes, we also
replicate the regression analysis of equity incentives (Table 3) for the dollar values of the CEO’s annual
total compensation (TOT_COMP), option compensation (OPT_COMP), restricted stock compensation
(STK_COMP) and cash pay (CASH_COMP), respectively, and report the results in Table 4. CEO total
compensation is the sum of option compensation, restricted stock compensation, cash pay and other
compensation. To minimize the impact of outliers, all compensation figures used in the regression
analysis are the logarithms of the reported figures in thousands of dollars. As shown in Table 4, CEO total
compensation exhibits a similar inverted U-shaped pattern we have documented for CEO equity
incentives. The null hypothesis that there is no change in CEO total compensation after 2001 is rejected at
the 1% level by the results from both regression specifications. Consequently, there is also evidence of a
sharp decline in CEO total compensation after 2001. This finding suggests that there is either a decline in
CEO’s reservation utility after 2001 or excess CEO pay above the minimum level required by the
reservation utility prior to 2002 or a combination of both.
        The results in Table 4 also suggest that the decline in CEO total compensation is primarily driven
by the decline in CEO option compensation. Of the three key components of CEO compensation, only
option compensation exhibits a sharp drop after 2001 while both stock compensation and cash pay
experience an upward shift around the same time. The changes in all three components are statistically
significant, with the change in option compensation the most significant and the change in cash pay the
least significant. Although both restricted stock grants and cash compensation experience an upward shift
after 2001, they are not nearly enough to make up for the large reductions in option compensation.
5.2. The relationship between option expensing and equity incentives
        We have documented a sharp reversal in the use of equity incentives, option incentives in
particular, around 2001. A rising trend before 2002 contrasts a declining one afterwards, coinciding with
changes in option expensing rules. Of course, the decline in CEO equity incentives in the post-2001
period could be caused by factors other than option expensing. To further investigate the relationship,
causal or otherwise, between option expensing and the use of equity incentives, we perform additional
empirical tests and compare the option expensing hypothesis with alternative explanations.
5.2.1. High incentive firms
        Not all firms are affected by option expensing in the same way. Some firms may have
underestimated the true cost of stock options and overused them prior to option expensing, due to the



                                                                                                            20
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



favorable accounting treatment (e.g., Hall and Murphy, 2003). Compared to other firms within the same
industry- and size-matched peer group, these “high” incentive firms provide their CEOs excessive levels
of option incentives. With the favorable accounting treatment of stock options repealed, all firms may
have to adjust their assessment of option costs sharply upwards and reduce their option use as a result.
The impact of option expensing is likely stronger for high incentive firms than for other firms. We thus
expect that the reduction in equity incentives, option incentives in particular, is larger for high incentive
firms than for other firms.
        To empirically test this hypothesis, we divide our sample firms into quartiles based on the levels
of equity incentives they provided to their CEOs prior to 2002. Holding the CEO’s perceived cost of
expensing $1 of options constant, firms in the top quartile are expected to make greater reductions in
equity incentives in the post-2001 period than firms in other quartiles. If the downward adjustments to
equity incentives are unrelated to option expensing, the changes are likely to be similar across the
quartiles.
        CEO portfolio incentives (PORT_PPS) at the end of 2001 are used to gauge the level of equity
incentives firms provide to their CEOs prior to option expensing. The year 2001 is appropriate as it is the
last year before our event year of 2002. For each firm in the sample, we determine CEO portfolio
incentives as the total equity incentives provided by his entire holdings of options, restricted and
unrestricted stock outstanding at the end of 2001. These equity incentives represent the CEO’s aggregate
incentives provided by stock and option grants received in 2001 and those received previously (and still
outstanding). For options granted in 2001, ExecuComp provides all required option terms (including
strike price and maturity) and the CEO’s equity incentives are calculated as previously described. For
options granted before 2001 but still outstanding at the end of 2001, we use Core and Guay’s (2002)
“one-year approximation” method to estimate the necessary option terms and then calculate their
incentives. For firms in the same industry (based on 2-digit SIC code) and firm-size (based on market
value of equity) decile, we rank the CEO’s total portfolio incentives at the end of 2001 and divide the
firms into four quartiles (Q1–Q4). Firms in the top quartile (Q4) provide the most equity incentives to
their CEOs relative to firms in other quartiles. These firms are classified to as high incentive firms.
        We make use of a dummy variable HIGH, which is equal to one for high incentive firms and zero
for other firms. We then include the interaction of this dummy variable with the two time trend variables
(t*HIGH and t2*HIGH in the quadratic model and TL1*HIGH and TL2*HIGH in the piecewise linear
model) to examine whether the use of equity incentives is different between high incentive firms and
other firms. While the interaction terms capture the difference between high incentive firms and other
firms, the usual trend variables (t and t2 in the quadratic model and TL1 and TL2 in the piecewise linear




                                                                                                          21
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



model) now measure the changes in CEO equity incentives for “low” incentive firms (firms in quartiles
Q1–Q3). The results are summarized in Table 5.
        Consider first the results from the piecewise linear specification of the regression model. The
coefficient of TL1*HIGH is positive and statistically significant for both total equity incentives and
option incentives, indicating that high incentive firms experience larger annual growth in these incentives
than other firms do prior to 2002. In comparison, the coefficient of TL2*HIGH is negative and significant
at the 1% level for both total equity incentives and option incentives, suggesting that the decline in these
incentives is also larger in high incentive firms than in other firms after 2001. Combining differences in
both subperiods, high incentive firms thus exhibit much larger reductions in both total equity incentives
and option incentives than other firms do. Between the two subperiods, the reduction in total equity
incentives (estimated by the coefficient of TL2 minus the coefficient of TL1) by low incentive firms is
0.25 per year. In comparison, the reduction in total equity incentives by high incentive firms is 0.46 larger
than by low incentive firms (estimated by the coefficient of TL2*HIGH minus the coefficient of
TL1*HIGH). The reduction in total equity incentives by high incentive firms is thus more than twice as
large as by low incentive firms. Not surprisingly, the reduction in option incentives by high incentive
firms is also more than twice as large as it is by low incentive firms. The null hypothesis that the
reduction in equity incentives is not different between high incentive firms and other firms (i.e.,
TL1*HIGH = TL2*HIGH) is strongly rejected at the 1% level. This is true for both total equity incentives
and option incentives. In contrast, none of the interaction terms between the dummy variable HIGH and
stock incentives (TL1*HIGH, TL2*HIGH, or TL2*HIGH – TL1*HIGH) is statistically significant,
suggesting that high incentive firms use similar stock incentives as other firms do. The differences in
equity incentives between high incentive firms and other firms are thus entirely driven by differences in
option incentives.
        Results from the quadratic specification of the regression model provide similar evidence that
high incentive firms make larger reductions in equity incentives than other firms do and the differences
are again entirely driven by changes in option incentives. In the regressions of total equity incentives and
option incentives, the coefficient of t*HIGH is positive and significant at the 1% level while the
coefficient of t2*HIGH is negative and significant at the 1% level. This is further evidence that high
incentive firms exhibit a stronger inverted U-shaped pattern in both total equity incentives and option
incentives than other firms do. In comparison, none of the corresponding coefficients in the regression of
stock incentives is statistically significant. This is again indication that high incentive firms use similar
stock incentives as other firms do.
        Combining the results from both specifications, there is consistent evidence that high incentive
firms reduce option incentives more than other firms do while there is little difference in the use of stock



                                                                                                          22
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



incentives across these firm groups. We also perform robustness checks subsequently and verify that the
results are not sensitive to our choice of the event year or definition of high incentive firms. These results
provide further evidence that the sharp reversal in the use of option incentives is related to option
expensing.
5.2.2. The Sarbanes-Oxley Act of 2002
         An alternative explanation for the downward shift in CEO option incentives in the post-2001
period is the passage of the Sarbanes-Oxley Act of 2002. As Jensen and Murphy (1990) argue, full and
timely disclosure of executive compensation, precisely the type of disclosure required by the Sarbanes-
Oxley Act, is likely to cause a decline in option incentives due to public outcry over excessive payouts.
The Sarbanes-Oxley Act is particularly relevant as it is passed in the middle of our event year of 2002
during which the likelihood of mandatory option expensing approaches certainty.
         Is the documented downward shift in CEO equity incentives related to option expensing or
caused by the passage of the Sarbanes-Oxley Act? To answer this question, we divide firms into two
subsamples with different levels of analyst coverage. As the level of analyst coverage rises, corporate
information including executive compensation is likely to be better monitored and scrutinized, making the
firm more transparent to the general public. As a result, firms covered by more analysts are less likely to
have problems with information asymmetry and are thus less likely to be affected by the passage of the
Sarbanes-Oxley Act than firms covered by fewer analysts. If the documented changes in option incentives
are related to the Sarbanes-Oxley Act, we expect the downward shift to be stronger for firms covered by
fewer analysts than firms covered by more analysts.
         Using analyst coverage information from the I/B/E/S database, we rank firms in our sample by
the number of analysts covering the firm (i.e., forecasting earnings for the firm) in 2001.15 The top half of
the sample forms our subsample of firms with good analyst coverage while the bottom half forms the
subsample of firms with poor analyst coverage. The mean (median) number of analysts is 13.9 (13.0) for
firms with good coverage and 2.8 (3.0) for firms with poor coverage. The difference in analyst coverage
is thus quite substantial between the two subsets of firms. We rerun regressions of option incentives for
the two subsamples separately and report the results in Table 6.
         As shown in Table 6, the inverted U-shaped pattern is clearly present for both types of firms with
good or poor analyst coverage. The coefficients for t and TL1 are positive, suggesting a rising trend in
option incentives before 2002, while the coefficients for t2 and TL2 are negative, suggesting a declining
trend in option incentives after 2001. The downward shift in option incentives is also statistically
significant at the 1% level. This is true for both types of firms with good or poor analyst coverage. The

15
  If a firm is not covered by the I/B/E/S database, we set the number of analysts to zero. Roughly 10% of our
sample firms fall into this category. The results are not materially affected if we drop these firms from our sample.


                                                                                                                    23
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



difference in analyst coverage thus cannot fully explain the decline in option incentives. Assuming
analyst coverage is a reasonable proxy for information asymmetry, we interpret these findings as support
for the option expensing hypothesis.
        Nevertheless, the inverted U-shaped pattern is stronger for firms covered by fewer analysts than
for firms covered by more analysts. Interpreting the results from the piecewise linear regression, for
example, the reduction in option incentives after 2001 is 0.25 (0.06 + 0.19) for firms with good coverage
and nearly twice as high at 0.49 (0.06 + 0.43) for firms with poor coverage. The difference in the
reduction of option incentives between firms with good and poor analyst coverage is statistically
significant at the 1% level. Results from the quadratic specification are similar, with the difference
significant at the 5% level. These differences are consistent with the impact of the Sarbanes-Oxley Act on
option incentives. We thus conclude that the passage of the Sarbanes-Oxley Act have contributed to the
downward shift in option incentives but does not appear to fully explain the changes we document.
5.2.3. The option backdating scandal
        Lie (2005) uncovers an option granting practice that appears to be widespread in U.S. firms. He
refers to it as the backdating of stock option awards whereby executives manipulate the option grant date
to reduce strike prices. They do this by choosing option grant dates within an allowable period to coincide
with low stock prices during that time period. As most stock options are granted at the money, the value
of the stock option award is inflated with backdating (holding the number of options granted fixed). In
August 2002, the Securities and Exchange Commission (SEC) requires firms to report all stock option
awards within two business days of the grants in response to changes mandated by the Sarbanes-Oxley
Act. With the new rules, executives’ ability of selecting a low stock price as the exercise price for their
stock options is greatly curbed.16
        Option backdating is relevant to the present study because the tightening of option reporting
requirements coincides with the change in option expensing rules. As the findings in Lie (2005) and
Heron and Lie (2006) suggest, option backdating is widespread prior to the passage of Sarbanes-Oxley
Act of 2002 but drops off dramatically afterwards. Because option backdating introduces an upward bias
in option values, the post-2001 downward shift in option incentives may be related to the backdating
problem prevalent before 2002. We thus need to investigate whether the downward shift in option
incentives is still present after we control for option backdating.
        Following Aboody and Kasznik (2000) and Lie (2005), we divide our sample firms into three
categories: scheduled firms (S), unscheduled firms (US), and unclassified firms (UC). Scheduled firms are
firms granting options on fixed calendar schedules and are the least likely to be involved in option

16
  Heron and Lie (2006) provide empirical evidence that the backdating practice is much weaker after the SEC
requires firm to report the related option grant information within two business days.


                                                                                                        24
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



backdating. Unscheduled firms that do not follow any schedule at all are the most likely to backdate
option grants. The remaining firms fall into the third category of unclassified firms that have a mixture of
both scheduled and unscheduled option grants. If option backdating is indeed responsible or partially
responsible for the documented downward shift in CEO option incentives in the post-2001 period, it
should affect unscheduled firms more than scheduled firms.
        Using Lie’s (2005) definition of scheduled/unscheduled grants,17 we identify 149 scheduled firms,
454 unscheduled firms, and 682 unclassified firms in our sample. We rerun firm-fixed effect regressions
of CEO option incentives (OPT_PPS) separately for the three groups of firms (S, US or UC), using both
piecewise linear and quadratic specifications. The results are summarized in Table 7. The inverted U-
shaped pattern is evident for all three categories of firms, suggesting a downward shift in CEO option
incentives for all three types of firms. This is evidence that option backdating does not fully explain the
reductions in the use of option incentives after 2001. Nevertheless, option backdating does impact the
magnitude of the downward shift in CEO option incentives. Take the results from the piecewise linear
specification of the regression model for example. The estimated annual downward shift in option
incentives is 0.32 (0.03 + 0.29) for scheduled firms but 63% larger at 0.52 (0.10 + 0.42) for unscheduled
firms. The difference is marginally significant at the 10% level. The results from the quadratic
specification of the regression model are similar. We thus conclude that the option backdating practice
contributes to the reduction in option incentives after 2001 but is unlikely the main culprit behind it.18
5.2.3. The 2000 stock market crash and technology firms
        The 2000 stock market crash and the subsequent bear market may have depressed stock prices in
the post-2001 period. The S&P 500 index rises 11.4% a year during the 1993-2001 period while the same
index increases only 1.8% a year during the 2002-2004 period. Such a large shift in market-wide stock
performance may have contributed to the downward shift in option incentives after 2001. To investigate
this possibility, we include returns on market indices, both value-weighted and equally-weighted CRSP
aggregate returns, as control variables in our regression analysis. Untabulated regression results show that
the results in Table 3 are not materially affected and the downward shift in option incentives is thus
unlikely influenced by the 2000 stock market crash.
        In addition, Murphy (2003) documents that compensation practices in new economy firms are
quite different from those in the old economy. New economy firms tend to use more equity compensation,

17
   Lie (2005) defines scheduled option awards only if the awards occur within one day of the one-year anniversary
of the prior year’s award date.
18
   We also apply the definition of scheduled option grants by Aboody and Kasznik (2000). They identify firms with
scheduled option awards if they award options within one week of the one-year anniversary of the prior year’s
award date. They identify firms with unscheduled option awards if the firms have no two awards separated by about
12 month. If firms have both scheduled and unscheduled awards, they categorize the corresponding firms into the
unclassified group. All qualitative inferences remain unchanged.


                                                                                                              25
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



especially option compensation. These firms are also front and center during the 2000 stock market crash
and are impacted the most by the subsequent bear market. Many new economy firms dramatically reduce
their use of options afterwards, with some even completely eliminating option compensation (e.g.,
Microsoft). Is it possible that the documented decline in the use of option incentives since 2001 are driven
by new economy firms?
        Following Murphy’s definition of new and old economy,19 we define 244 new economy firms and
682 old economy firms. Untabulated results show that the inverted U-shaped pattern in CEO option
incentives is present in both new and old economy firms. The downward shift in option incentives is thus
not driven by new economy firms. Nonetheless, we do find evidence that new economy firms reduce
option incentives more than comparable firms in the old economy. This is expected since option
incentives are used more extensively by new economy firms. The opposition to mandatory option
expensing by new economy firms is also more vocal, suggesting a greater perceived cost of option
expensing for these firms. These results are thus consistent with the option expensing hypothesis.
5.2.4. Voluntary expensing firms
        Seethamraju and Zach (2004) and Cater, Lynch, and Tuna (2007) find that firms that choose to
voluntarily expense their stock options reduce option grants to CEOs upon their expensing decision. In
this section, we further compare and analyze changes in CEO option incentives between voluntary
expensing firms and other firms. According to a Bear Stearns report, there are 483 firms that had adopted
or would adopt the option expensing provision by February 2004. Of these voluntary expensing firms,
183 are found in our sample. We then rerun our regression for CEO option incentives by adding a dummy
variable (EXP) for the 183 voluntary expensing firms. Untabulated results show that there is no difference
in the use of option incentives between voluntary expensing firms and other firms. Both types of firms
have similar inverted U-shaped patterns in CEO option incentives. The voluntary expensing dummy is not
statistically significant in either the piecewise linear or the quadratic specification. We thus conclude that
the decline in the use of option incentives is not driven by voluntary expensing firms.
        Interestingly, we do find evidence that voluntary expensing firms provide their CEOs larger
(smaller) option pay, in dollars estimated using the Black-Scholes-Merton model, than other firms before
2002 (after 2001). We perform regressions of the logarithm of CEO option pay (in constant 2004 dollars)
using both the piecewise linear and quadratic specifications. The voluntary expensing dummy is always
significant at the 1% level, with the coefficients of t*EXP positive and t2*EXP negative in the quadratic
specification and the coefficients of TL1*EXP positive and TL2*EXP negative in the piecewise linear

19
  Following Murphy (2003), new economy firms are defined as companies with primary SIC codes 3570, 3571,
3572, 3576, 3577, 3661, 3674, 4812, 4813, 5045, 5961, 7370, 7371, 7372, and 7373. Old economy firms are firms
with SIC codes less than 4000 and not otherwise categorized as new economy.



                                                                                                           26
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



specification. These results suggest that voluntary expensing firms may have adopted poorly designed
option plans before 2002 since they provide the same level of option incentives at greater costs than other
firms do. This problem seems to be resolved upon voluntary expensing, with these firms making
necessary adjustments to make their option plans as effective as other firms are.
5.2.5. CEO vs. other employees
        For option expensing purposes, all stock options affect reported income in the same way,
regardless of the recipients being the CEO or other employees. It thus can be argued that firms may cut
back option grants to non-CEO employees in the same way they do to their CEOs. This is consistent with
the expected impact of option expensing on the use of stock options. On the other hand, the same option
may not provide the same incentives to other employees as it does to the CEO since the CEO’s actions
can directly affect the value of the options while the actions of other employees generally have less
impact. As Kedia and Mozumdar (2002) and Oyer (2004) argue, the incentive purpose is a less important
reason for firms to make a broad-based option plan. Instead, firms are more likely to use stock options to
retain and compensate lower level workers. To the extent that stock options granted to the rank-and-file
are mainly used to retain employees rather than inducing them to make value-maximizing efforts, firms
may be constrained in their decision to cut back option grants to non-CEO employees.
        To test these competing hypotheses (i.e., option expensing vs. employee retention), we
investigate changes in option grants to all non-CEO employees. We find that the mean (median) level of
non-CEO option incentives is 23.75 (13.79) in 2001 but drops to 11.80 (7.83) by 2004. Both declines are
statistically significant at the 1% level. Unreported firm-fixed effect regressions provide further evidence
that the downward shift in non-CEO option incentives in the post-2001 period is significant and robust to
various control variables. In addition, high incentive firms experience a greater downward shift in non-
CEO option incentives than low incentive firms. These results support the option expensing hypothesis.
        To investigate the retention hypothesis, we examine whether the reductions in the use of options
are different between CEOs and non-CEO employees. We find that the allocation of stock options
between the CEO and non-CEO employees is essentially unchanged during the period from 2001 to 2004.
Non-CEO employees are awarded with an average (median) of 88% (91%) of the total stock options
granted in 2001. About the same proportion of stock options are granted to non-CEO employees in 2004.
These figures suggest that non-CEO option grants are cut in proportion to CEO option grants in the post-
2001 period. None of the results changes materially if we separate the top five executives from lower
level employees. These findings do not support the retention hypothesis.




                                                                                                         27
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



6. Further analysis
        To provide further support for our findings, we conduct additional robustness analysis in this
section. In particular, we want to investigate whether our inferences remain valid if we use alternative
event years, use alternative definitions of high incentive firms, use median regressions, exclude firms with
CEO turnovers, or exclude regulated industries.
6.1. Alternative classification of high incentive firms
        To measure the relative strength of equity incentives, we compare equity incentives provided to a
firm’s CEO with the corresponding incentives provided to CEOs in other firms within the same industry
and firm-size decile. High incentive firms are identified as those providing their CEOs with total portfolio
incentives (PORT_PPS) in the top quartile among firms in their peer group. This approach makes the
implicit assumption that industry and size are sufficient determinants of CEO equity incentives and
ignores other possible determinants that have already been identified in previous studies.
        To ensure the robustness of our findings, we redefine high incentive firms by including other
determinants of CEO equity incentives. In addition to industry and firm size, we also consider other
control variables we have discussed and used previously such as growth opportunities, stock return
volatility, leverage, CEO tenure and corporate governance (see Appendix 1 for the full list of control
variables). For the year of 2001, we regress CEO portfolio incentives against the full set of control
variables across all firms in the sample. We then calculate residuals from this cross-sectional regression as
the difference between the actual level of portfolio incentives and the predicted level. Following Core and
Guay (1999), these residuals (INCT_RES) are treated as proxies for excess CEO equity incentives at the
end of 2001, relative to other firms in the sample. CEOs with larger residuals have stronger equity
incentives than CEOs with smaller residuals. We reclassify high incentive firms as firms in the top
quartile of residuals (INCT_RES) and the remaining firms as low incentive firms. This new classification
takes into account other control variables as well as industry and firm size.
        With the new definition of high incentive firms, we rerun firm-fixed effect regressions of CEO
option incentives (OPT_PPS) using both piecewise linear and quadratic specifications and summarize the
results in Table 8. The reclassification of high incentive firms only marginally changes the regression
estimates and none of our inferences is materially affected. The inverted U-shaped pattern is still evident,
with a positive trend in the pre-2002 period followed by a negative trend in the post-2001 period. More
importantly, the downward shift in CEO option incentives in the post-2001 period is much stronger for
high incentive firms than for low incentive firms as the coefficients for the interaction terms indicate.
Interpreting the results from the piecewise linear specification, the downward shift in CEO equity
incentives for high incentive firms is 0.31 (0.08 + 0.23) larger than the corresponding shift for low
incentive firms, statistically significant at the 1% level. The magnitude of the difference is only



                                                                                                          28
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



marginally lower than it is reported in Table 5 using the previous definition of high incentive firms (0.32
= 0.06 + 0.26). Consequently, our inferences are robust to alternative classifications of high incentive
firms.
6.2. Median regressions
         Another concern is the potential impact of outliers on our regression results analyzed so far.
Although we do winsorize all continuous variables at the 1% and 99% levels, it is still prudent to use
alternative estimation methods that are less likely to be influenced by outliers. Following prior research
(e.g., Jensen and Murphy, 1990), we perform median regressions to investigate the robustness of our
findings. Unlike OLS regressions, median regressions minimize the sum of absolute deviations instead of
the sum of squared deviations and are thus less likely to be affected by outliers.
         We rerun regressions of option incentives using median regressions for both piecewise linear and
quadratic specifications and summarize the results in Table 9. The downward shift in CEO option
incentives is still evident in the regression results with a positive trend in the pre-2002 period and a
negative trend after 2001. As expected, the magnitude of the downward shift is smaller in the median
regressions than in the OLS regressions. Take the results from the piecewise linear specification for
example. The downward shift in option incentives is 0.10 (0.06 + 0.04) from the median regression
compared to 0.36 (0.06 + 0.30) from the corresponding OLS regression (in Table 3). Nevertheless, both
changes are statistically significant at the 1% level. The results from the quadratic specification are
similar. We thus conclude that our inferences are not materially affected by outliers.
6.3. The choice of 2002 as the event year
         With 2002 as our event year, we define high and low incentive firms using CEO portfolio
incentives (PORT_PPS) at the end of 2001. As firm-specific events (such as CEO turnovers) may cause
unusual deviations in CEO equity incentives from the industry norm in any given year, it is prudent to
consider alternative choices of the event year.
         We first consider using either 2001 or 2000 as the event year instead of 2002. High and low
incentives are reclassified using the new event year. We also consider another classification of high and
low incentive firms based on the average CEO portfolio incentives over the three-year period from 1999
to 2001. Although the event year remains to be 2002, the averaging should remove much of the impact
from firm-specific events. We then rank the average CEO portfolio incentives against other CEOs in
firms within the same industry and firm-size decile. High incentive firms are identified as those with the
three-year average CEO portfolio incentives in the top quartile of the size- and industry-matched group.
Untabulated results indicate that the difference in the downward shift in CEO option incentives between
high and low incentive firms is still significant at the 1% level in all three alternative classifications of




                                                                                                          29
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



high and low incentive firms. All our inferences regarding the impact of mandatory option expensing on
CEO option incentives remain qualitatively unchanged.
6.4. CEO turnovers
        In all our regressions so far, we use dummy variables to control for CEO turnovers. In the event
of a CEO turnover, the CEO on record for that year may be either the incoming or the departing CEO
depending on the timing of the turnover during the fiscal year. The incoming CEO may also have a
different level of equity incentives compared to the departing CEO and the compensation for either CEO
may be just a fraction of what they are normally paid annually. As a result, the CEO on record that year
may have lower levels of compensation than in other years. To further investigate the effect of CEO
turnovers, we exclude all firms that have CEO turnovers after 2001. By dropping these firms, we
eliminate any potential downward shifts in CEO equity incentives in the post-2001 period due to CEO
turnovers. We do not exclude firms that experience CEO turnovers before 2002. The inclusion of these
firms is likely to reduce equity incentive in the pre-2002 period, which is biased against the option-
expensing hypothesis and our model predictions. A total of 1,022 firms remain after removing firms with
CEO turnovers after 2001, a loss of roughly 36% of the sample firms. We repeat our analysis in Tables 3-
5 for this subsample of firms and find no material change in our inferences.
6.5. Regulated industries
        Finally, firms in regulated industries may have different executive compensation practices
compared to firms in unregulated industries. In particular, firms in regulated industries may respond
differently to mandatory option expensing relative to other firms. For example, Aboody, Barth, and
Kasznik (2004b) report that, financial services industry (i.e., banks and insurance services firms)
constitutes 34.8% of the firms that voluntarily choose to recognize option expenses in 2002 and early
2003, in contrast to only 8.9% of the control firms. The impact of option expensing may be different for
these firms. We thus divide our sample firms into two subsamples of firms in regulated and unregulated
industries. In order to ensure that our results are not driven by firms in regulated industries, we rerun the
firm-fixed effect regressions after removing firms in the regulated industries. Following common practice
in the literature, we exclude firms with the 2-digit SIC of 49 (electric utility) and 60-69 (financials). This
step reduces our sample to 1,294 firms. Untabulated results show that the coefficients of interest, those
associated with trend variables and interaction terms, retain the same signs as before. While the
magnitudes of the estimated coefficients are affected marginally, their statistical significance is not
materially affected. Thus, our results are robust to the exclusion of firms in regulated industries.




                                                                                                           30
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



7. Conclusions
     In this paper, we have examined the impact of mandatory option expensing on executive
compensation. Although moving the cost of option grants from footnotes to the income statement does
not necessarily reveal any incremental information, we argue that executives’ perceived personal loss
from option expensing may lead to a downward adjustment to option incentives. Using CEO
compensation data from 1993 to 2004, we indeed find that firms reduce the use of option incentives but
maintain or slightly increase the use of stock incentives after 2001.
     Our contribution to the literature is three fold. First, we document a sharp reversal in the use of
option incentives, with a rising trend prior to 2002 contrasting a declining trend afterwards, and an
increasing trend in stock incentives throughout the entire period. Secondly, we provide strong evidence
that the reduction in option incentives is related to mandatory option expensing rather than merely
coincidental. We find that (1) high incentive firms make larger reductions in option incentives than other
firms do; (2) firms make similar reductions to option grants made to the CEO, other top executives and
lower level employees; (3) voluntary expensing firms make similar reductions in option incentives as
other firms do; (4) other regulatory, business and market events that coincide with the change in option
expensing rules such as the Sarbanes-Oxley Act of 2002, the option backdating scandal, and the 2000
stock market crash contribute to but do not fully explain the documented decline in option incentives.
Finally, we provide a theoretical explanation for the documented decline in equity incentives, option
incentives in particular, using a principal-agent model. Although option expensing may not have any
direct impact on a firm’s business or financial operations, it may alter the optimal contract between the
firm and its CEO, potentially leading to real economic consequences for the firm.
     Interestingly, the reductions in option incentives are not made up by matching increases in stock
incentives. While CEO option incentives experience a downward shift of 0.36 per year, the corresponding
stock incentives only rise 0.02 per year in comparison (as reported in Table 3 after taking into account
various control variables). In addition, the gap is even larger for high incentive firms as they make larger
reductions in option incentives than other firms do but make similar increases in stock incentives as other
firms do. Our results thus suggest that mandatory option expensing leads to large reductions in the use of
option incentives but only marginal increases in the use of stock incentives. How does the net reduction in
equity incentives affect firm valuation and performance? In addition, there is also evidence that firms
reduce CEO total compensation after 2001 (Table 4) which is primarily driven by the reductions in option
compensation. Are CEOs making downward adjustment to their reservation utilities after 2001 or forced
to give up some or all of their excess pay? These intriguing questions are left for future research.




                                                                                                         31
Do you want know more? http://www.isknow.com
  Topic: http://www.isknow.com/compensation



                                                              Appendix 1

Variable                                                   Definition                                                   Source
                                          Compensation Variable
OPT_PPS    Stock option grants as a proportion of total shares outstanding times the Black-Scholes-Merton hedge     ExecuComp
           ratio
STK_PPS    Restricted stock grants as a proportion of total shares outstanding                                      ExecuComp
TOT_PPS    The sum of option incentives and stock incentives (OPT_PPS + STK_PPS)                                    ExecuComp
TOT_COMP   The value of total annual compensation, in millions of constant 2004 dollars                             ExecuComp
OPT_COMP   The value of annual option grants estimated using the modified Black-Scholes-Merton model, in            ExecuComp
           millions of constant 2004 dollars
STK_COMP   The value of annual restricted stock grants, in millions of constant 2004 dollars                        ExecuComp
CASH_COMP  Salary and bonus, in millions of constant 2004 dollars                                                   ExecuComp
                                             Control Variable
FIRM_SIZE  The logarithm of the market value of equity ($MM)                                                        Compustat
STK_VOL    The standard deviation of trailing 36 monthly stock returns                                              CRSP
CEO_TENURE The number of years the CEO has been presiding over the current firm                                     ExecuComp
GROWTH_OPP (Book value of debt + market value of equity) / total assets                                             Compustat
LEVERAGE   Total debts / total assets                                                                               Compustat
CASH_SHORT The three-year average of (common dividends + preferred dividends + cash flow used in investing          Compustat
           activities – cash flow from operations) / total assets
PAST_RET   The cumulative stock returns in the past 36 months                                                       CRSP
GIM        The Gompers, Ishii and Metrick (2003) governance index                                                   IRRC
MTR        The marginal tax rate based on income before interest expense, simulated by John Graham                  John Graham,
                                                                                                                    Duke U.
INCT_RES         Excess equity incentives (actual minus predicted equity incentives) estimated using the Core and   ExecuComp,
                 Guay (1999) method                                                                                 CRSP, Compustat




                                                                                                                                 32
  Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



References


Aboody, D., 1996. Market valuation of employee stock options. Journal of Accounting and Economics 22,
   357-391.
Aboody, D., Kasznik, R., 2000. CEO stock option awards and the timing of corporate voluntary
   disclosures. Journal of Accounting and Economics 29, 73-100.
Aboody, D., Barth, M.E., Kasznik, R., 2004a. SFAS No.123 stock-based compensation expense and
   equity market values. Accounting Review 79, 251-275.
Aboody, D., Barth, M.E., Kasznik, R., 2004b. Firms’ voluntary recognition of stock-based compensation
   expense. Journal of Accounting Research 42, 123-149.
Aggarwal, R.K., Samwick, A.A., 1999. The other side of the trade-off: the impact of risk on executive
   compensation. Journal of Political Economy 107, 65-105.
Baber, W., Janakiraman, S., Kang, S., 1996. Investment opportunities and the structure of executive
   compensation. Journal of Accounting and Economics 21, 297-318.
Baker, G.P., Hall, B.J., 2004. CEO incentives and firm size. Journal of Labor Economics 22, 767-798.
Balsam, S., Bartov, E., Yin, J., 2004. Disclosure versus recognition of option expense: and empirical
   investigation of SFAS No. 148 and stock returns. Working paper, Temple University, New York
   University and Rutgers University.
Bebchuk, L.A., Fried, J.M., 2003. Executive compensation as an agency problem. Journal of Economic
   Perspectives 17, 71-92.
Bebchuk, L.A., Fried, J.M., Walker, D.I., 2002. Managerial power and rent extraction in the design of
   executive compensation. University of Chicago Law Review 69, 751-846.
Bizjak, J., Brickley, J., Coles, J., 1993. Stock-based incentive compensation and investment behavior.
   Journal of Accounting and Economics 16, 349–372.
Brown, L.D., Lee, Y.-J., 2007. The impact of SFAS 123R on changes in option-based compensation.
   Working paper, Georgia State University.
Carter, M.E, Lynch, L.J., Tuna, I., 2007. The role of accounting in the design of CEO equity
   compensation. Accounting Review 82, 327-357.
Core, J.E., Guay, W.R., 1999. The use of equity grants to manage optimal equity incentive levels. Journal
   of Accounting and Economics 28, 151-184.
Core, J.E., Guay, W.R., 2001. Stock option plans for non-executive employees. Journal of Financial
   Economics 61, 253-287.
Core, J.E., Guay, W.R., 2002. Estimating the value of employee stock option portfolios and their
   sensitivities to price and volatility. Journal of Accounting Research 40, 613-630.



                                                                                                       33
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Core, J.E., Guay, W.R., Larcker, D.F., 2003. Executive equity compensation and incentives: a survey.
   Economic policy review, 27-50.
Core, J.E., Holthausen, R.W., Larcker, D.F., 1999. Corporate governance, chief executive officer
   compensation, and firm performance. Journal of Financial Economics 51, 371-406.
Daniel, N.D., Kale, J.R., Naveen, L., 2003. Do option expensing announcements convey information to
   the stock market? Working paper, Georgia State University.
Dechow, P.M., Hutton, A.P., Sloan, R.G., 1996. Economic consequences of accounting for stock-based
   compensation. Journal of Accounting Research 34, 1-20.
Demsetz, H., Lehn, K., 1985. The structure of corporate ownership: causes and consequences. Journal of
   Political Economy 93, 1155-1177.
Elayan, F., Pukthuanthong, K., Roll, R., 2005. Investors like firms that expense employee stock options
   and they dislike firms that fail to expense. Journal of Investment Management 3, 1-24.
Gaver, J., Gaver, K., 1993. Additional evidence on the association between the investment opportunity set
   and corporate financing, dividend, and compensation policies. Journal of Accounting and Economics
   16, 125–160.
Gibbons, R., Murphy, K.J., 1992. Optimal incentive contracts in the presence of career concerns: theory
   and evidence. Journal of Political Economy 100, 468-505.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of
   Economics 118, 107-155.
Graham, J., Harvey, C., Rajgopal, S., 2005. The economic implications of financial reporting. Journal of
   Accounting and Economics (forthcoming).
Hall, B.J., Liebman, J.B., 1998. Are CEOs really paid like bureaucrats? Quarterly Journal of Economics
   113, 653-691.
Hall, B.J., Murphy, K.J., 2003. The trouble with stock options. Journal of Economic Perspectives 17, 49-
   70.
Heron, R., Lie, E., 2006. Does backdating explain the stock price pattern around executive stock option
   grants? Forthcoming Journal of Financial Economics.
Holmstrom, B., 1979. Moral hazard and observability. Bell Journal of Economics 10, 74-91.
Ittner, C.D., Lambert, R.A., Larcker, D.F., 2003. The structure and performance consequences of equity
   grants to employees of new economy firms. Journal of Accounting and Economics 34, 89-127.
Jensen, M., Murphy, K., 1990. Performance pay and top-management incentives. Journal of Political
   Economy 98, 225-264.
Jin, L., 2002, CEO compensation, diversification, and incentives, Journal of Financial Economics 66, 29-
   63.



                                                                                                      34
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



John, T.A., John, K., 1993. Top-management compensation and capital structure. Journal of Finance 48,
   949-974.
Kedia, S., Mozumdar, A., 2002. Performance impact of employee stock options. Working paper, Harvard
   University and Boston University.
Lie, E., 2005. On the timing of CEO stock option awards. Management Science 51, 802-812.
Matsunaga, S.R., 1995. The effects of financial reporting costs on the use of employee stock options.
   Accounting Review 70, 1-26.
McConnell, J., Servaes, H., 1990. Additional evidence on equity ownership and corporate value. Journal
   of Financial Economics 27, 595-612.
Milbourn, T.T., 2003. CEO reputation and stock-based compensation. Journal of Financial Economics 68,
   233-262.
Mirrlees, J., 1976. The optimal structure of incentives and authority within an organization. Bell Journal
   of Economics 7, 105-131.
Morck, R., Shleifer, A, Vishny, R., 1988. Management ownership and market valuation: an empirical
   analysis. Journal of Financial Economics 20, 293-315.
Murphy, K., 1999. Executive compensation. In: Ashenfelter, O., Card, D. (Eds), Handbook of Labor
   Economics 3, 2485-2563.
Murphy, K., 2003. Stock-based pay in new economy firms. Journal of Accounting and Economics 34,
   129-147.
Oyer, P., 2004. Why do firms use incentives that have not incentive effects? Journal of Finance 59, 1619-
   1649.
Oyer, P., Schaefer, S., 2005. Accounting, governance, and broad-based stock option grants. Working
   paper, Stanford University and Northwestern University.
Palia, D., 2001. The endogeneity of managerial compensation in firm valuation: a solution. Review of
   Financial Studies 14, 735-764.
Petersen, M., 2006. Estimating standard errors in finance panel data sets: comparing approaches. Working
   paper, Northwestern University.
Schaefer, S., 1998. The dependence of pay-performance sensitivity on the size of the firm. Review of
   Economics and Statistics 80, 436-443.
Seethamraju, C., Zach, T., 2004. Expensing stock options: the role of publicity. Working paper,
   Washington University in St. Louis.
Simth, C.W., Watts, R., 1992. The investment opportunity set and corporate financing, dividend, and
   compensation policies. Journal of Financial Economics 32, 263-292.




                                                                                                       35
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Yermack, D., 1995. Do corporations award CEO stock options effectively? Journal of Financial
  Economics 39, 237-269




                                                                                         36
Do you want know more? http://www.isknow.com
 Topic: http://www.isknow.com/compensation



 Table 1
 Summary statistics

 This table provides summary statistics of compensation and control variables for firms in our sample over
 the 1993–2004 period. Definitions of these variables are in Appendix 1. The full sample consists of 1,609
 firms.

     Variables               N           Mean        1st quartile    Median      3rd quartile     St. dev
TOT_PPS                    15153        2.1610         0.1345        0.8731         2.3873        4.4475
OPT_PPS                    15153        1.8293         0.0000        0.7391         2.1774        3.0349
STK_PPS                    15153        0.1484         0.0000        0.0000         0.0000        0.4835
TOT_COMP                   15083        4.6071         1.2077        2.4078         5.1817        6.2271
OPT_COMP                   15083        2.2440         0.0020        0.7050         2.3065        4.2922
STK_COMP                   15152        0.3869         0.0000        0.0000         0.0000        1.2634
CASH_COMP                  15153        1.4180         0.6293        1.0411         1.7500        1.2781
FIRM_SIZE                  16349        7.2993         6.1811        7.1680         8.3206        1.5949
STK_VOL                    14598        0.1210         0.0759        0.1044         0.1498        0.0634
CEO_TENURE                 14020        8.2523         3.0000        6.0000        11.0000        7.3325
GROWTH_OPP                 16349        2.1505         1.1677        1.5330        2.3326         2.1189
LEVERAGE                   16286        0.2254         0.0629         0.2112       0.3407         0.1888
CASH_SHORT                 15355        -0.1755       -0.2479        -0.1691       -0.1024        0.1480
PAST_RET                   14598        0.2934        -0.0607        0.3103         0.6911        0.7534
GIM                        12290        9.2430         7.0000        9.0000        11.0000        2.6901
MTR                        11575        0.3217         0.3497        0.3500         0.3501        0.0819




                                                                                                       37
 Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Table 2
Yearly summary statistics of annual CEO compensation

This table reports yearly summary statistics for CEO equity incentives from 1993 to 2004. Option
incentives (OPT_PPS) and stock incentives (STK_PPS) measure incentives provided by the CEO’s option
grants and restricted stock grants, respectively. Total equity incentive (TOT_PPS) is the sum of option
incentives and stock incentives. Definitions of these incentive variables are in Appendix 1.

Panel A: Total equity incentives (TOT_PPS)

          Year        Mean          1st quartile      Median         3rd quartile    St.dev
          1993        1.1261          0.0000           0.2739          0.9617        3.4934
          1994        1.5612          0.0000           0.4211          1.5863        4.0535
          1995        1.3907          0.0000           0.4844          1.4780        3.0310
          1996        1.7227          0.0322           0.6126          1.7492        3.3474
          1997        2.0945          0.0930           0.7813          2.1922        4.4485
          1998        2.4111          0.1757           0.9016          2.5846        5.5372
          1999        2.5724          0.2138           1.0529          3.0389        4.3904
          2000        2.7766          0.2908           1.2420          3.2019        5.0627
          2001        2.6698          0.3151           1.3247          3.2481        4.4206
          2002        2.4038          0.2808           1.1825          2.8107        4.4273
          2003        2.1126          0.2266           1.0243          2.3895        4.3380
          2004        1.8524          0.2166           0.9119          2.0491        4.7750




                                                                                                    38
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Panel B: Option incentives (OPT_PPS)

          Year       Mean             1st quartile   Median   3rd quartile   St.dev
         1993        0.9468             0.0000       0.2035      0.8123      2.1540
         1994        1.3507             0.0000       0.3436      1.4760      2.6577
         1995        1.2526             0.0000       0.3969      1.3454      2.5303
         1996        1.5368             0.0000       0.5436      1.5511      2.8369
         1997        1.8282             0.0000       0.6954      2.0319      3.1634
         1998        1.9928             0.0739       0.7794      2.4494      3.3288
         1999        2.2838             0.1342       0.9218      2.9038      3.4272
         2000        2.3678             0.1701       1.1015      2.9784      3.5494
         2001        2.3296             0.2009       1.1813      2.9698      3.3511
         2002        1.9696             0.1459       0.9547      2.4949      2.9237
         2003        1.7174             0.0000       0.8251      2.0655      2.7417
         2004        1.3771             0.0000       0.6593      1.6440      2.3029

Panel C: Stock incentives (STK_PPS)

         Year        Mean         1st quartile       Median   3rd quartile   St.dev
         1993       0.0622             0.0000        0.0000     0.0000       0.2726
         1994       0.0865             0.0000        0.0000     0.0000       0.3411
         1995       0.0850             0.0000        0.0000     0.0000       0.3441
         1996       0.1009             0.0000        0.0000     0.0000       0.3940
         1997       0.1010             0.0000        0.0000     0.0000       0.3753
         1998       0.1177             0.0000        0.0000     0.0000       0.4304
         1999       0.1251             0.0000        0.0000     0.0000       0.4495
         2000       0.1385             0.0000        0.0000     0.0000       0.4778
         2001       0.1673             0.0000        0.0000     0.0000       0.5323
         2002       0.2139             0.0000        0.0000     0.0000       0.6304
         2003       0.2033             0.0000        0.0000     0.1091       0.5410
         2004       0.2632             0.0000        0.0000     0.2459       0.5990




                                                                                      39
Do you want know more? http://www.isknow.com
      Topic: http://www.isknow.com/compensation



      Table 3
      Changes in CEO equity incentives from 1993 to 2004

      This table reports results from firm-fixed effect regressions of the CEO’s total equity incentives
      (TOT_PPS), option incentives (OPT_PPS) and stock incentives (STK_PPS) in the sample period.
      Whenever a variable has a missing value, we record it as zero and use a dummy variable to control for the
      effect of missing values. Estimated coefficients of the intercept and dummy variables for CEO turnovers
      and missing values are suppressed. Clustered standard errors are reported in brackets. ***, **, and *
      denote significance at the 1%, 5%, and 10% levels, respectively.

                                         Quadratic model (1)                  Piecewise linear model (2)
Independent variable           TOT_PPS       OPT_PPS       STK_PPS        TOT_PPS    OPT_PPS        STK_PPS
                               0.3667***     0.3307***        -0.0012
t
                                [0.0513]      [0.0482]       [0.0084]
                              -0.0266*** -0.0256***         0.0013**
t2
                                [0.0031]      [0.0029]       [0.0005]
                                                                         0.0778***      0.0578**      0.0117***
TL1
                                                                           [0.0245]      [0.0228]      [0.0039]
                                                                         -0.2748***    -0.2954***     0.0312***
TL2
                                                                           [0.0329]      [0.0319]      [0.0066]
Test: TL1=TL2 (p-value)                                                     0.0000        0.0000          0.0151
                              -0.4803***    -0.3421***     -0.0548***    -0.4818***    -0.3446***     -0.0545***
FIRM_SIZE
                                [0.0836]      [0.0785]       [0.0132]      [0.0839]      [0.0789]       [0.0132]
                               4.0259***     3.1480***         0.0487     3.0231**      2.0421**          0.1235
STK_VOL
                                [1.1029]      [0.9487]       [0.1770]      [1.1889]      [1.0168]       [0.1937]
                                  0.0197        0.0106         0.0031       0.0193        0.0102          0.0031
CEO_TENURE
                                [0.0201]      [0.0187]       [0.0032]      [0.0201]      [0.0187]       [0.0032]
                                 -0.0009       -0.0007        -0.0001       -0.0009       -0.0007        -0.0001
CEO_TENURE2
                                [0.0006]      [0.0006]       [0.0001]      [0.0006]      [0.0006]       [0.0001]
                                 0.0906*      0.0775*          0.0016     0.1024**      0.0878**          0.0012
GROWTH_OPP
                                [0.0469]      [0.0424]       [0.0037]      [0.0467]      [0.0422]       [0.0037]
                                  0.1455        0.1724        -0.0136       0.1893        0.2027         -0.0135
LEVERAGE
                                [0.4067]      [0.3937]       [0.0591]      [0.4084]      [0.3947]       [0.0591]
                                 -0.1874        0.0448          -0.1        -0.1537        0.096         -0.1052
CASH_SHORT
                                [0.4658]      [0.4167]       [0.0719]      [0.4697]      [0.4199]       [0.0721]
                                0.1552**      0.1191*        0.0166*      0.1559**      0.1202**        0.0165*
PAST_RET
                                [0.0665]      [0.0608]       [0.0093]      [0.0666]      [0.0609]       [0.0093]
                                 -0.3915       -0.1661         -0.129       -0.4047       -0.1827        -0.1276
MTR
                                [0.5923]      [0.5112]       [0.0833]      [0.5918]      [0.5105]       [0.0833]
                                 -0.0233       -0.0179        -0.0031       -0.0285       -0.0223        -0.0029
GIM
                                [0.0293]      [0.0266]       [0.0051]      [0.0293]      [0.0266]       [0.0051]
                               -0.0042**    -0.0039***        -0.0002     -0.0043**    -0.0040***        -0.0002
INCT_RES
                                [0.0017]      [0.0015]       [0.0002]      [0.0017]      [0.0015]       [0.0002]
N                                 12059         12059          12059         12059         12059          12059
Adj. R2                           0.3972        0.3964         0.3766       0.3967        0.3962          0.3767



                                                                                                            40
      Do you want know more? http://www.isknow.com
          Topic: http://www.isknow.com/compensation



          Table 4
          Changes in dollar values of CEO pay from 1993 to 2004

          This table reports results from firm-fixed effect regressions of dollar values (in logarithm) of the CEO’s
          total pay (TOT_COMP), option grants (OPT_COMP), restricted stock grants (STK_COMP), and cash pay
          (CASH_COMP). All pay figures are in 2004 constant dollars. Whenever a variable has a missing value,
          we record it as zero and use a dummy variable to control for the effect of missing values. Estimated
          coefficients of the intercept and dummy variables for CEO turnovers and missing values are suppressed.
          Clustered standard errors are reported in brackets. ***, **, and * denote significance at the 1%, 5%, and
          10% levels, respectively.

                                  Quadratic model (1)                                 Piecewise linear model (2)
Independent            TOT_       OPT_         STK_    CASH_               TOT_          OPT_          STK_         CASH_
variable               COMP       COMP        COMP     COMP                COMP          COMP         COMP          COMP
                    0.1326*** 0.4896*** -0.1503*** 0.0325***
t
                      [0.0125]   [0.0562]    [0.0465] [0.0093]
                    -0.0068*** -0.0337*** 0.0173***    0.0006
t2
                      [0.0008]   [0.0035]    [0.0029] [0.0006]
                                                                          0.0623*** 0.1587***    0.0147            0.0356***
TL1
                                                                           [0.0059]   [0.0263]  [0.0222]            [0.0046]
                                                                         -0.0406*** -0.3946*** 0.3130***           0.0557***
TL2
                                                                           [0.0086]   [0.0395]  [0.0359]            [0.0066]
Test: TL1=TL2
                                                                            0.0000       0.0000       0.0000            0.0187
(p-value)
                    0.3678*** 0.6244*** 0.1659** 0.1677*** 0.3664*** 0.6142*** 0.1725** 0.1686***
FIRM_SIZE
                      [0.0214]   [0.0831]   [0.0714]  [0.0186]   [0.0213]   [0.0834]   [0.0714]   [0.0186]
                    1.2144***      1.2067 -2.3162*** -0.3782* 0.8359***      -1.1266    -0.9766    -0.2383
STK_VOL
                      [0.2457]   [0.9524]   [0.8135]  [0.2040]   [0.2633]   [1.0291]   [0.8751]   [0.2235]
                    0.0240***      0.0333   0.0334*     0.0019  0.0239***      0.033   0.0335*      0.0019
CEO_TENURE
                      [0.0053]   [0.0233]   [0.0199]  [0.0044]   [0.0053]   [0.0233]   [0.0198]   [0.0043]
                    -0.0007*** -0.0021***    -0.0009   -0.0001 -0.0007*** -0.0021***   -0.0009    -0.0001
CEO_TENURE2
                      [0.0002]   [0.0007]   [0.0006]  [0.0002]   [0.0002]   [0.0007]   [0.0006]   [0.0002]
                    -0.0364*** -0.0586* -0.0699*** -0.0292*** -0.0342***      -0.051 -0.0729*** -0.0287***
GROWTH_OPP
                      [0.0107]   [0.0324]   [0.0216]  [0.0068]   [0.0107]   [0.0322]   [0.0215]   [0.0068]
                       -0.0143     -0.212    0.0559    -0.0938    -0.0127    -0.2428     0.0844    -0.0853
LEVERAGE
                      [0.0850]   [0.3584]   [0.3009]  [0.0626]   [0.0849]   [0.3572]   [0.2996]   [0.0627]
                    -0.5457*** -1.2996***    -0.1661  -0.1706* -0.5210*** -1.1158***   -0.2791   -0.1868**
CASH_SHORT
                      [0.1067]   [0.4264]   [0.3722]  [0.0887]   [0.1067]   [0.4262]   [0.3736]   [0.0894]
                    0.0663***     -0.0796 0.1353*** 0.0957*** 0.0669***      -0.0752  0.1327*** 0.0953***
PAST_RET
                      [0.0159]   [0.0583]   [0.0471]  [0.0122]   [0.0159]   [0.0583]   [0.0472]   [0.0122]
                       0.1579      0.0206    -0.5044 0.2467***     0.1508    -0.0277    -0.4763  0.2503***
MTR
                      [0.1146]   [0.4337]   [0.3660]  [0.0833]   [0.1141]   [0.4308]   [0.3653]   [0.0833]
                       0.0032      0.0181   0.0520*  0.0180***     0.0023     0.0151   0.0530*   0.0177***
GIM
                      [0.0071]   [0.0330]   [0.0285]  [0.0060]   [0.0071]   [0.0330]   [0.0284]   [0.0059]
                       -0.0006  -0.0032*     -0.0004 -0.0007**    -0.0007  -0.0036**    -0.0002  -0.0007**
INCT_RES
                      [0.0004]   [0.0017]   [0.0011]  [0.0004]   [0.0004]   [0.0017]   [0.0011]   [0.0004]
N                       12008      12008      12059     12059      12008      12008      12059      12059
Adj. R2                0.7292      0.4455    0.4798     0.7695     0.7295     0.4477     0.4808     0.7697



                                                                                                                   41
          Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Table 5
Changes in CEO equity incentives in high incentive firms

This table reports results from firm-fixed effect regressions of the CEO’s total equity incentives
(TOT_PPS), option incentives (OPT_PPS) and stock incentives (STK_PPS) in the sample period. HIGH
is a dummy variable, which equals one for high incentive firms (whose CEO’s portfolio incentives in
2001 are in the top quartile of the peer group of industry and size matched firms) and zero for other firms.
Whenever a variable has a missing value, we record it as zero and use a dummy variable to control for the
effect of missing values. Estimated coefficients of the intercept and dummy variables for CEO turnovers
and missing values are suppressed. Clustered standard errors are reported in brackets. ***, **, and *
denote significance at the 1%, 5%, and 10% levels, respectively.




                                                                                                         42
Do you want know more? http://www.isknow.com
      Topic: http://www.isknow.com/compensation




                                       Quadratic model (1)                Piecewise linear model (2)
Independent variable         TOT_PPS       OPT_PPS        STK_PPS     TOT_PPS    OPT_PPS        STK_PPS
                             0.2691***     0.2647***        -0.0056
t
                              [0.0501]      [0.0485]       [0.0088]
                            -0.0192*** -0.0205*** 0.0016***
t2
                              [0.0031]      [0.0030]       [0.0006]
                             0.4336***     0.2933***         0.0199
t*HIGH
                              [0.0978]      [0.0881]       [0.0145]
                            -0.0327*** -0.0224***           -0.0016
t2*HIGH
                              [0.0069]      [0.0064]       [0.0011]
                                                                       0.0575**      0.0443*    0.0112***
TL1
                                                                        [0.0240]     [0.0228]    [0.0040]
                                                                      -0.1882***   -0.2330***   0.0351***
TL2
                                                                        [0.0345]     [0.0334]    [0.0071]
Test: TL1=TL2 (p-value)                                                   0000        0.0000      0.0057
                                                                       0.0889**      0.0597*      0.0024
TL1*HIGH
                                                                        [0.0354]     [0.0306]    [0.0051]
                                                                      -0.3678***   -0.2643***     -0.0161
TL2*HIGH
                                                                        [0.0777]     [0.0726]    [0.0129]
Test: TL1*HIGH =
                                                                        0.0000      0.0003       0.2346
TL2*HIGH (p-value)
                            -0.4814***    -0.3423***    -0.0546***    -0.4802***   -0.3430***   -0.0542***
FIRM_SIZE
                              [0.0837]      [0.0788]      [0.0132]      [0.0842]     [0.0793]     [0.0132]
                             3.9346***     3.0867***        0.0447     2.9297**      1.9765*         0.12
STK_VOL
                              [1.0991]      [0.9471]      [0.1766]      [1.1850]     [1.0144]     [0.1934]
                                0.0152        0.0072        0.0027        0.0143       0.0063      0.0027
CEO_TENURE
                              [0.0203]      [0.0188]      [0.0033]      [0.0204]     [0.0189]     [0.0033]
                               -0.0008       -0.0006       -0.0001       -0.0007      -0.0005      -0.0001
CEO_TENURE2
                              [0.0006]      [0.0006]      [0.0001]      [0.0006]     [0.0006]     [0.0001]
                             0.0920**       0.0784*         0.0017     0.1034**     0.0885**       0.0013
GROWTH_OPP
                              [0.0467]      [0.0423]      [0.0037]      [0.0465]     [0.0420]     [0.0037]
                                0.134         0.165         -0.014        0.1726       0.1912       -0.014
LEVERAGE
                              [0.4058]      [0.3933]      [0.0593]      [0.4065]     [0.3938]     [0.0591]
                               -0.1663        0.0613       -0.0982       -0.1304       0.1147      -0.1035
CASH_SHORT
                              [0.4631]      [0.4153]      [0.0719]      [0.4665]     [0.4183]     [0.0721]
                             0.1533**       0.1177*       0.0165*      0.1533**      0.1182*      0.0164*
PAST_RET
                              [0.0663]      [0.0606]      [0.0093]      [0.0665]     [0.0608]     [0.0093]
                               -0.3808       -0.1602       -0.1289       -0.3866      -0.1713      -0.1273
MTR
                              [0.5892]      [0.5092]      [0.0833]      [0.5895]     [0.5090]     [0.0833]
                               -0.0266       -0.0202       -0.0033       -0.0317      -0.0247      -0.0031
GIM
                              [0.0293]      [0.0267]      [0.0051]      [0.0293]     [0.0266]     [0.0051]
                            -0.0046***    -0.0041***       -0.0002    -0.0049***   -0.0044***      -0.0002
INCT_RES
                              [0.0017]      [0.0015]      [0.0002]      [0.0017]     [0.0015]     [0.0002]
N                               12059         12059         12059         12059        12059        12059
Adj. R2                         0.3987        0.3972        0.3767        0.3982       0.3971      0.3767



                                                                                                   43
      Do you want know more? http://www.isknow.com
      Topic: http://www.isknow.com/compensation



      Table 6
      The impact of Sarbanes-Oxley Act of 2002 on CEO option incentives (OPT_PPS)

      This table reports separate regression results of CEO option incentives for firms with good analyst
      coverage and for those with poor analyst coverage. A firm has good (poor) analyst coverage if the number
      of analysts making earnings forecasts for the firm in 2001 is above (below) the median. Whenever a
      variable has a missing value, we record it as zero and use a dummy variable to control for the effect of
      missing values. Estimated coefficients of the intercept and dummy variables for CEO turnovers and
      missing values are suppressed. Clustered standard errors are reported in brackets. ***, **, and * denote
      significance at the 1%, 5%, and 10% levels, respectively.

                                           Quadratic model (1)                 Piecewise linear model (2)
Independent variable                Poor coverage     Good coverage         Poor coverage      Good coverage
                                     0.4102***           0.2612***
t
                                       [0.0718]           [0.0635]
                                     -0.0335***         -0.0188***
t2
                                       [0.0044]           [0.0038]
Test: Poor-Good=0 (p-value)                      0.0115
                                                                                0.0594*               0.0563*
TL1
                                                                                [0.0334]              [0.0326]
                                                                              -0.4266***            -0.1861***
TL2
                                                                                [0.0496]              [0.0407]
Test: Poor-Good=0 (p-value)                                                                0.0034
                                       -0.1997           -0.3903***             -0.1858             -0.4016***
FIRM_SIZE
                                      [0.1243]             [0.1079]            [0.1252]               [0.1090]
                                     3.7587***             2.6531**              2.3016                 1.9488
STK_VOL
                                      [1.4004]             [1.2803]            [1.4706]               [1.3970]
                                       0.0085                0.0093              0.0086                 0.0095
CEO_TENURE
                                      [0.0274]             [0.0244]            [0.0274]               [0.0244]
                                       -0.0006              -0.0006             -0.0006                -0.0006
CEO_TENURE2
                                      [0.0009]             [0.0007]            [0.0009]               [0.0007]
                                       0.1201              0.0645*               0.1325              0.0745**
GROWTH_OPP
                                      [0.0976]             [0.0358]            [0.0970]               [0.0354]
                                        -0.61              1.2674**             -0.6002              1.3208**
LEVERAGE
                                      [0.5548]             [0.5229]            [0.5531]               [0.5301]
                                       0.3194               -0.3311              0.3922                -0.3053
CASH_SHORT
                                      [0.6090]             [0.5590]            [0.6149]               [0.5598]
                                       0.0957                0.1058              0.0981                 0.1083
PAST_RET
                                      [0.0936]             [0.0857]            [0.0937]               [0.0855]
                                       -0.3147               0.0275             -0.3422                 0.0172
MTR
                                      [0.7550]             [0.6568]            [0.7521]               [0.6567]
                                       -0.0134              -0.0164             -0.0194                -0.0201
GIM
                                      [0.0388]             [0.0360]            [0.0386]               [0.0360]
                                       -0.0031            -0.0054**            -0.0032*             -0.0056***
INCT_RES
                                      [0.0019]             [0.0021]            [0.0020]               [0.0021]
N                                       5569                  6490                5569                   6490
Adj. R2                                0.3829                0.4069              0.3834                 0.4063




                                                                                                           44
      Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Table 7
The impact of option backdating on option incentives (OPT_PPS)

This table reports results from separate firm-fixed effect regressions of CEO option incentives for
subsamples of firms. Sample firms are divided into three categories: firms with scheduled (S),
unscheduled (US) and unclassified (UC) option grants. Firms granting options on a fixed calendar
schedule are unlikely to be involved in backdating while those following no schedule at all are the most
likely to backdate option grants. The remaining firms have both scheduled and unscheduled option grants.
The firm-fixed effect regressions are run separately for the three groups of firms. Whenever a variable has
a missing value, we record it as zero and use a dummy variable to control for the effect of missing values.
Estimated coefficients of the intercept and dummy variables for CEO turnovers and missing values are
suppressed. Clustered standard errors are reported in brackets. ***, **, and * denote significance at the
1%, 5%, and 10% levels, respectively.




                                                                                                        45
Do you want know more? http://www.isknow.com
      Topic: http://www.isknow.com/compensation




                                    Quadratic model (1)                   Piecewise linear model (2)
Independent variable             S            US          UC            S             US           UC
                            0.2428**      0.4392***    0.3865***
t
                             [0.0969]      [0.1088]     [0.0723]
                           -0.0209*** -0.0335***      -0.0292***
t2
                             [0.0062]      [0.0063]     [0.0045]
Test: S-US=0 (p-value)               0.1541
                                                                       0.0274        0.1012*     0.0761**
TL1
                                                                     [0.0404]        [0.0566]     [0.0312]
                                                                   -0.2867*** -0.4230***        -0.3255***
TL2
                                                                     [0.0651]        [0.0713]     [0.0496]
Test: S-US=0 (p-value)                                                        0.0810
                              -0.1509    -0.4467**    -0.3166***        -0.16       -0.4462**   -0.3227***
FIRM_SIZE
                             [0.1428]     [0.1844]      [0.0986]     [0.1416]        [0.1851]     [0.0991]
                               1.7312       2.6472     4.2019***       0.3815          0.7105     2.8859*
STK_VOL
                             [2.3066]     [1.8055]      [1.4218]     [2.4409]        [1.9663]     [1.5216]
                            0.1177***      -0.0059        -0.002    0.1164***         -0.0088      -0.0029
CEO_TENURE
                             [0.0395]     [0.0422]      [0.0277]     [0.0397]        [0.0423]     [0.0278]
                           -0.0027***      -0.0008       -0.0002   -0.0026***         -0.0007      -0.0002
CEO_TENURE2
                             [0.0010]     [0.0012]      [0.0009]     [0.0010]        [0.0012]     [0.0010]
                               0.0559       0.0987        0.0474       0.0578          0.1099        0.061
GROWTH_OPP
                             [0.1020]     [0.0836]      [0.0695]     [0.1022]        [0.0824]     [0.0690]
                              -1.2576       0.9212       -0.4475      -1.2042          0.9796      -0.4419
LEVERAGE
                             [1.0246]     [0.7778]      [0.5222]     [1.0195]        [0.7804]     [0.5239]
                               1.5408       0.7054       -0.6679       1.6168          0.8898      -0.6282
CASH_SHORT
                             [1.2113]     [0.6900]      [0.6725]     [1.2073]        [0.7016]     [0.6749]
                              -0.1937       0.011      0.3031***      -0.1647         -0.0068    0.3050***
PAST_RET
                             [0.1369]     [0.1149]      [0.0914]     [0.1352]        [0.1147]     [0.0918]
                              -0.2247       0.8125       -1.0372       -0.232          0.7429      -1.0601
MTR
                             [1.0252]     [0.9478]      [0.7990]     [1.0089]        [0.9525]     [0.7936]
                               0.0382      -0.0364       -0.0283       0.0323         -0.0369      -0.0342
GIM
                             [0.0942]     [0.0538]      [0.0401]     [0.0955]        [0.0533]     [0.0402]
                              -0.0051    -0.0067**     -0.0056**      -0.0052      -0.0071***    -0.0058**
INCT_RES
                             [0.0040]     [0.0026]      [0.0027]     [0.0041]        [0.0027]     [0.0027]
N                               1232         3352          5352         1232            3352         5352
Adj. R2                        0.4261       0.3982        0.3612       0.4266          0.3994       0.3609




                                                                                                    46
      Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Table 8
Alternative definitions of high incentive firms: option incentives (OPT_PPS)

This table reports results from firm-fixed effect regressions of CEO option incentives from 1993 to 2004
using an alternative definition of high incentive firms. CEO portfolio incentives (PORT_PPS) at the end
of 2001 are first compared with the predicted portfolio incentives which are estimated using a cross-
sectional regression with various control variables. A firm is deemed a high incentive firm if its CEO’s
excess portfolio incentives (actual minus predicted) is in the top quartile. Whenever a variable has a
missing value, we record it as zero and use a dummy variable to control for the effect of missing values.
Estimated coefficients of the intercept and dummy variables for CEO turnovers and missing values are
suppressed. Clustered standard errors are reported in brackets. ***, **, and * denote significance at the
1%, 5%, and 10% levels, respectively.




                                                                                                      47
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation




 Independent variable                       Quadratic model (1)   Piecewise linear model (2)
                                               0.2612***
 t
                                                 [0.0512]
                                               -0.0204***
 t2
                                                 [0.0031]
                                               0.3102***
 t*HIGH
                                                 [0.0820]
                                               -0.0220***
 t2*HIGH
                                                 [0.0061]
                                                                           0.0426*
 TL1
                                                                           [0.0235]
                                                                         -0.2395***
 TL2
                                                                           [0.0331]
 Test: TL1=TL2 (p-value)                                                    0.0000
                                                                          0.0784***
 TL1*HIGH
                                                                           [0.0295]
                                                                         -0.2266***
 TL2*HIGH
                                                                           [0.0755]
 Test: TL1*HIGH = TL2*HIGH
 (p-value)                                                                   0.0012
                                                -0.3395***               -0.3415***
 FIRM_SIZE
                                                  [0.0787]                 [0.0792]
                                                2.8875***                  1.7646*
 STK_VOL
                                                  [0.9576]                 [1.0272]
                                                     0.01                    0.0097
 CEO_TENURE
                                                  [0.0188]                 [0.0188]
                                                   -0.0007                  -0.0007
 CEO_TENURE2
                                                  [0.0006]                 [0.0006]
                                                  0.0742*                 0.0848**
 GROWTH_OPP
                                                  [0.0426]                 [0.0423]
                                                   0.2666                    0.2973
 LEVERAGE
                                                  [0.3921]                 [0.3931]
                                                   0.0961                    0.1583
 CASH_SHORT
                                                  [0.4196]                 [0.4217]
                                                  0.1197*                 0.1221**
 PAST_RET
                                                  [0.0616]                 [0.0618]
                                                   -0.1916                  -0.2092
 MTR
                                                  [0.5159]                 [0.5133]
                                                   -0.0146                  -0.0191
 GIM
                                                  [0.0267]                 [0.0267]
                                                -0.0039***               -0.0042***
 INCT_RES
                                                  [0.0015]                 [0.0015]
 N                                                  11680                    11680
 Adj. R2                                           0.3863                    0.3861




                                                                                               48
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation



Table 9
Median regressions of option incentives (OPT_PPS)

This table reports the results from median regressions of CEO option incentives. Median regressions are
less likely to be impacted by outliers as they minimize the sum of absolute deviations rather than the sum
of squared deviations. HIGH is a dummy variable, which equals one for high incentive firms (whose
CEO’s portfolio incentives in 2001 are in the top quartile of the peer group of industry and size matched
firms) and zero for other firms. Whenever a variable has a missing value, we record it as zero and use a
dummy variable to control for the effect of missing values. Estimated coefficients of the intercept and
dummy variables for CEO turnovers and missing values are suppressed. Clustered standard errors are
reported in brackets. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.




                                                                                                       49
Do you want know more? http://www.isknow.com
Topic: http://www.isknow.com/compensation




Independent variable               Quadratic model (1)       Piecewise linear model (2)
                              0.1436***         0.1308***
t
                               [0.0173]          [0.0184]
                             -0.0079***        -0.0068***
t2
                               [0.0012]          [0.0012]
                                                0.0450***
t*HIGH
                                                 [0.0140]
                                               -0.0042***
t2*HIGH
                                                 [0.0014]
                                                             0.0566***        0.0545***
TL1
                                                              [0.0069]         [0.0063]
                                                            -0.0403***        -0.0308**
TL2
                                                              [0.0136]         [0.0137]
Test: TL1=TL2 (p-value)                                        0.0000           0.0000
                                                                              0.0125***
TL1*HIGH
                                                                               [0.0045]
                                                                              -0.0564**
TL2*HIGH
                                                                               [0.0258]
Test: TL1*HIGH =
                                                                               0.0170
TL2*HIGH (p-value)
                             -0.1858***        -0.1854***   -0.1882***       -0.1878***
FIRM_SIZE
                               [0.0080]          [0.0083]     [0.0090]         [0.0081]
                              6.4103***        6.4844***     6.3100***       6.3679***
STK_VOL
                               [0.2220]          [0.2315]     [0.2569]         [0.2311]
                               0.0089*            0.0069       0.0089           0.0079
CEO_TENURE
                               [0.0053]          [0.0055]     [0.0060]         [0.0054]
                             -0.0009***        -0.0008***   -0.0008***       -0.0008***
CEO_TENURE2
                               [0.0002]          [0.0002]     [0.0002]         [0.0002]
                                 0.0113           0.0075       0.0118           0.0099
GROWTH_OPP
                               [0.0078]          [0.0081]     [0.0088]         [0.0079]
                               0.1236*          0.1382**     0.1538**         0.1606**
LEVERAGE
                               [0.0673]          [0.0700]     [0.0756]         [0.0679]
                                -0.0108           -0.0252      -0.018           -0.0229
CASH_SHORT
                               [0.0918]          [0.0955]     [0.1034]         [0.0929]
                              0.1925***        0.1922***     0.1983***       0.2012***
PAST_RET
                               [0.0168]          [0.0175]     [0.0189]         [0.0170]
                                -0.1529           -0.1735      -0.158           -0.1951
MTR
                               [0.1481]          [0.1540]     [0.1668]         [0.1499]
                              0.0189***        0.0188***     0.0178***       0.0176***
GIM
                               [0.0043]          [0.0044]     [0.0048]         [0.0043]
                             -0.0028***        -0.0031***   -0.0029***       -0.0034***
INCT_RES
                               [0.0005]          [0.0005]     [0.0005]         [0.0005]
N                                12059             12059       12059             12059
Adj. R2                          0.0953           0.0955       0.0952           0.0953




                                                                                          50
Do you want know more? http://www.isknow.com

				
My first website Thank You! My first website Thank You! http://www.isknow.com
About Do you want know more? http://www.isknow.com/