Table4 tests conservatism using Basu’s regression. Under the definition, the
coefficient of R measures the timeliness of earnings with respect to good news. In
contrast, the coefficient of R*D measures the “incremental” timeliness of earnings
regarding bad news. Therefore, β3 is predicted to be positive and significant. In this
implies that unrealized gains is postponed to future periods.
Table 4
Test of Basu’s Model: Fiscal Year Returns
Model: EPSBPt1 = β0 + β1R + β2D + β3R*D
17
Table5 tests the relation between proportion of equity-based compensation and
accounting conservatism:
EPSBPt
1= β
0+ β
1R + β
2D + β
3R*D +ECOMP*(β
4+ β
5R + β
6D + β
7R*D ) +
Size*(β
8+ β
9R + β
10D + β
11R*D ) + Leverage*(β
12+ β
13R + β
14D + β
15R*D )
+ Growth* (β
16+ β
17R + β
18D + β
19R*D ) +ε
In the regression, β1 measures earnings timeliness with respect to good news and β3 measures the asymmetric timeliness with respect to bad news. β5, β9, β13, and β17
measure the association of β1 with ECOMP, Leverage, Growth, and Size, respectively.
β7, β11, β15, and β19 measure the association of β3 with ECOMP, Leverage, Growth,
and Size, respectively. The coefficient of ECOMP*R*D (β7), measuring the relation
between proportion of executives’ equity-based compensation and accounting
conservatism with regard to bad news, is predicted to be negative.
Within the sample period 1993-2012, Table5 represents that coefficient on R*D
is significantly positive (0.425), and the coefficient on ECOMP*R*D is significantly
negative (-0.0525). These results suggest that as ratio of executives’ equity-based
compensation to total compensation declines, earnings become more asymmetrically
timely in recognizing bad news.
18
Turning to control variables, they are in accordance with the theoretical
prediction. The coefficient of Leverage*R*D is positive and significant as expected,
indicating that firms with greater leverage are more asymmetrically timely in
recognizing bad news. The coefficient of Growth*R*D and Size*R*D are both
significantly negative which is consistent with extant empirical evidence. There is
relatively more public information for larger firms (Banz 1981). Firm size is
positively correlated with the relative amount of public information which reduces
information asymmetry between investors generating a negative association with
conservatism.
19 Table 5
Test of Hypothesis: There is a negative relation between the proportion of equity-based compensation to executives and the level of conservatism.
Model: EPSBPt1 = β0 + β1R + β2D + β3R*D + ECOMP*(β4 + β5R + β6D + β7R*D ) + Size*(β8 + β9R + β10D + β11R*D ) + Leverage*(β12 + β13R + β14D + β15R*D ) + Growth*(β16 + β17R + β18D + β19R*D )
t statistics in parentheses
* p<.1, ** p<0.05, *** p<0.01, p-values are one-tailed when the sign of the coefficient is predicted, two-tailed otherwise.
Independent Variables Predict Sign 1993-2012
R + -0.0211
20
To provide more specific relation between equity-based compensation and
accounting conservatism, I divide the whole sample period into four parts: 1993-2001,
2003-2005, 2006-2008, and 2009-2012 owing to following reasons:
There has been a dramatic change in both the level and composition of executive
pay in the U.S. during the last two decades. The level of compensation has increased
substantially in 1990s, and the increase was attributed to the grant of stock options.
During the early 1990s, stock options became a single largest component of
compensation (Murphy 1999). I set the first period from 1993-2001 due to a small
decrease in stock options granted to executives since 2002. Corporate accounting
scandals including those at Enron, WorldCom and other companies, have been linked
(-0.26)
21
to escalation in option grants. Stock options have been criticized on giving managers a
strong incentive to risk chasing investments and misleading shareholders about the
true condition of their companies (MaDick 2003). To avoid bias analysis, it is
necessary to exclude year 2002 from the sample.
Next period is 2003-2005. As of fiscal year 2006, executive compensation is
reported under new filling requirements. Companies with fiscal year end after Dec
2005 have to adjust to new reporting requirement FAS123(R). This statement focuses
primarily on accounting for share-based payment transactions exchanging employee
services. Equity-based compensation has to be expensed and be reflected in the
financial statements based on fair value of the awards. Prior to FASB 123(R),
companies could expense options using the intrinsic value method and often recorded
no associated expense on their Income Statement. Before 2006 the variable collected
from Execucomp called “OPTION_AWARDS_BLK_VALUE” and “RSTKGRNT”
are used. The OPTION_AWARDS_FV and STOCK_AWARDS_FV columns are
essentially comparable between the old and new reporting formats after 2006. They
both represent the value of options/restricted stock that were awarded during the
indicated fiscal year. The one difference is that under the old format, the Black
Scholes values for options were calculated by S&P (since companies were not
required to report them); under the new format, Black Scholes values are reported by
22
the company. Hall and Murphy (2002) suggest that models of firms’ choices of equity
compensation methods should include the accounting considerations. Carter et al.
(2007) support the assertion that accounting affects the design of executive
compensation. On the basis of previous studies, year 2006 might be a watershed for
executives’ structure of compensation.
The third and fourth periods are 2006-2008 and 2009-2012. Subprime mortgage
crisis in 2007 and the collapse of Lehman Brothers and problems of Merrill Lynch,
AIG, Freddie Mac, Fannie Mae… in late 2008 almost brought down the world’s
financial system. The financial crisis is considered by many economists the worst
financial crisis since the Great Depression of the 1930s, which might affect the
analysis results. Therefore, I divide the year after 2006 into those two parts.
Appendix C reports the descriptive statistics for the variables used in this study
of different periods.
23 Table 6
Test of Hypothesis: There is a negative relation between the proportion of equity-based compensation to executives and the level of conservatism.
Model: EPSBPt1 = β0 + β1R + β2D + β3R*D + ECOMP*(β4 + β5R + β6D + β7R*D ) + Size*(β8 + β9R + β10D + β11R*D ) + Leverage*(β12 + β13R + β14D + β15R*D ) + Growth*(β16 + β17R + β18D + β19R*D )
t statistics in parentheses
* p<.1, ** p<0.05, *** p<0.01, p-values are one-tailed when the sign of the coefficient is predicted, two-tailed otherwise.
Independent Variables
Predict Sign
1993-2001 2003-2005 2006-2008 2009-2012
R + -0.0615** -0.0916* -0.106 0.0762
Leverage -0.00798 -0.00536 -0.0362 -0.0415**
(-0.84) (-0.30) (-1.31) (-2.42)
Leverage*R -0.0142 -0.0968*** 0.0439 -0.0767**
24
Growth -0.00158** -0.000521 0.00181 0.00307**
(-2.57) (-0.41) (0.99) (2.30)
Growth*R -0.000505 0.00443* -0.00477 0.00567*
(-0.59) (1.91) (-1.01) (1.71)
Growth*D 0.00175 0.00342 -0.00743*** 0.00324
(1.59) (1.21) (-2.84) (1.29)
25
The coefficient of ECOMP*R*D in period 1993-2001 is negative but not
significant. Thanks to Jensen & Murphy (1990) demonstrating that the compensation
of top executives is virtually independent of their performance, peoples’
understanding of the link between CEO compensation and company performance has
improved substantially. They argue that pay-performance sensitivity in managerial
compensation contracts is too low to provide executives with incentives to act in the
interests of shareholders. The most powerful link between shareholder wealth and
executive wealth is direct stock ownership by the CEO. However, CEO holdings as a
percentage of corporate value compare to prior decades have declined. Figure1 shows
that the mean of ECOMP has grown in the 1990s and reached 40% (the sample mean
of ECOMP is 38%) in 1998. Nevertheless, the empirical result is insignificant. I think
the first reason might be that equity-based compensation for US executives has just
become increasing popular in 1990s. Second, companies generally do not treat
options as an expense, either at time of grant or exercise, on company financial
statements at that time. In 1992, SEC’s new disclosure rules, comprising with firms,
would only report “numbers” of option grants. It suggests that shareholders may be
difficult to understand manager’s compensation policy.
26
0.1.2.3.4.5
mean of RMCOMP
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Figure 1 The mean of ECOMP
Period 2003-2005 and 2006-2008 are consistent with my hypothesis that the
coefficient of ECOMP*R*D is significantly negative at 0.05 levels. These results
suggest that the asymmetric timeliness of earnings declines with managerial
proportion of executives’ equity-based compensation. Executives’ equity-based
compensation plays a role in accounting conservatism policy. In 1995, SFAS 123
encourages firms to calculate stock-based compensation expense based on the fair
value of options granted, but permits entities to continue using APB 25. Following the
financial reporting scandals of firms such as Enron and WorldCom, dozens of firms
began to announce their intention to recognize SFAS 123 expense voluntarily. Firms
that are more active in the capital markets are more likely to reap benefits from such a
signal (Aboody et al. 2004). Executives’ compensation structure will change through
the time and accounting considerations. Prior to implementation of FAS123(R), firms
make decisions based on the perceived costs rather than the economic costs, they
27
grant more options than they would. It is important to note that firms appear to
substitute away from stock options towards other forms of performance-based pay as
opposed to salary because of FAS123(R) (Hayes et al 2012). Murphy (2012) also
illustrates that the use of restricted shares gradually substitute for stock options to top
executives. According to Hall and Murphy (2003), incentives are maximized through
granting nontradable restricted stock rather than options, and it also affects managerial
incentives to engage in risky investments. Consequently, the trends can support period
2003-2005 and 2006-2008 are consistent with my hypothesis that the coefficient of
ECOMP*R*D is significantly negative.
The coefficient of ECOMP*R*D in period 2009-2012 is negative but not
significant. Although options accounted for only 20 percent of total pay, while
restricted stock had ballooned to 34 percent by 2010 (Murphy 2012), catastrophic
corporate failure in late 2008 draws shareholders’ attention on options which
incentivize managers’ excessive risk seeking behavior. On the other hand, benefits of
options will backfire in bear market (Hall and Murphy 2003). According to CenFIS
reports, the financial crisis in September and October 2008 was accompanied by
stunning decreases in stock prices. The S&P 500 fell 48 percent in a little over six
months. Executives and employees at many companies hold worthless options in
response to financial crisis. The effects mutually may cause the relation between the
28
proportion of equity-based compensation to executives and the level of conservatism
in significant.
Table7 reports the results which replace ECOMP with logT5EBC to assess my
results and provide additional insights. The findings are similar to those in Table6.
EPSBPt
1= β
0+ β
1R + β
2D + β
3R*D + logT5EBC* (β
4+ β
5R + β
6D + β
7R*D )
+ Size*(β
8+ β
9R + β
10D + β
11R*D ) + Leverage*(β
12+ β
13R + β
14D + β
15R*D ) + Growth*(β
16+ β
17R + β
18D + β
19R*D ) +ε
where
logT5EBC is the log of equity-based compensation, including stock option and
restricted stock grants, to top five executives. Valuation is based on the grant date fair
value reported by company or using standard and poor’s Black-Scholes methodology
before 2006.
29 Table 7
Test of Hypothesis: There is a negative relation between the proportion of equity-based compensation to executives and the level of conservatism.
Model: EPSBPt1 = β0 + β1R + β2D + β3R*D + logT5EBC*(β4 + β5R + β6D + β7R*D ) + Size*(β8 + β9R + β10D + β11R*D ) + Leverage*(β12 + β13R + β14D + β15R*D ) + Growth*(β16 + β17R + β18D + β19R*D )
t statistics in parentheses
* p<.1, ** p<0.05, *** p<0.01, p-values are one-tailed when the sign of the coefficient is predicted, two-tailed otherwise.
Independent Variables
Predict Sign
1993-2001 2003-2005 2006-2008 2009-2012
R + -0.0599** -0.104* -0.116 0.0795
logT5EBC -0.00326** -0.00173 -0.000297 -0.000236
(-2.48) (-0.64) (-0.10) (-0.08)
logT5EBC*R -0.00564** 0.00998** -0.00216 -0.00582
(-2.42) (1.98) (-0.28) (-1.02)
logT5EBC*D -0.00194 -0.00726 -0.00673 0.00748
(-0.83) (-1.20) (-1.49) (1.34)
logT5EBC*R*D - 0.00868 -0.0680*** -0.0358*** -0.00291
(1.18) (-2.99) (-2.66) (-0.18)
Leverage -0.00684 -0.00425 -0.0373 -0.0437**
(-0.72) (-0.24) (-1.35) (-2.55)
Leverage*R -0.0124 -0.0915*** 0.0406 -0.0721**
30
Growth -0.00205*** -0.00103 0.00160 0.00317**
(-3.35) (-0.82) (0.87) (2.39)
Growth*R -0.000886 0.00451** -0.00473 0.00431
(-1.04) (1.96) (-1.00) (1.33)
Growth*D 0.00153 0.00434 -0.00735*** 0.00291
(1.41) (1.54) (-2.81) (1.16)
31