Professional Documents
Culture Documents
1. Introduction
Considerable research has been devoted to documenting and understanding the
importance of just meeting or beating the consensus analyst forecast.1 For example,
several studies document evidence that managers adjust accruals in order to beat consensus analyst forecast targets (Ayers, Jiang, and Yeung 2006; Burgstahler and Eames
2006), which is consistent with the evidence that managers have incentives to do so
(Cheng and Wareld 2005; Matsumoto 2002; Matsunaga and Park 2001; McVay,
Nagar, and Tang 2006).2 However, it is not clear whether adjusting accruals to beat the
target increases or decreases earnings quality. It is also not clear how the board views
such behavior or how board member characteristics inuence managements tendency
to use accruals to beat the target. To provide evidence on these issues, I examine
whether the presence of nancial experts on the audit committee affects the use of
accruals to beat analyst forecast targets conditional on the impact of accruals on the
earnings quality.
Using accruals to beat the analyst forecast target could either increase or decrease the
quality of earnings, depending on the nature of managements private information.
Adjusting accruals to beat the target would increase the information value of earnings
when managers use accruals to signal private favorable information about future performance. On the other hand, when managers have private information that the rm will not
*Email: pei-hui.hsu@csueastbay.edu
2014 City University of Hong Kong and National Taiwan University
407
be able to sustain its performance in the future, adjusting accruals upward to beat the
target could provide an overly optimistic view of the rm and mislead shareholders.
Prior research suggests that board members monitor the nancial reporting process
and that higher quality boards increase earnings quality (Beasley 1996; Farber 2005).
This implies that the board should encourage or discourage accruals management to
beat analyst targets depending on whether the adjustment properly reects managements private information. Boards should allow rms to adjust accruals to beat the
analyst forecast if they expect future nancial performance to be good and object if
they expect poor future performance. This is consistent with boards protecting shareholders by monitoring the quality of nancial information presented by the rm.
However, this assumes that board members have sufcient knowledge and expertise
to evaluate managers private information, and there is evidence that the extent of
nancial expertise is not consistent across boards (Krishnan and Visvanathan 2008).
Regulators have expressed concerns regarding whether directors have sufcient knowledge to effectively monitor the nancial reporting process (Blue Ribbon Committee on
Improving the Effectiveness of Corporate Audit Cmmittees 1999; Sarbanes-Oxley Act
of 2002 [SOX] 2002). Academic research also supports this view by showing a positive
relation between appropriate expertise on the audit committee and the nancial reporting quality (Agrawal and Chadha 2005; Dhaliwal, Naiker, and Navissi 2010; Zhang,
Zhou, and Zhou 2007). As a result, in this paper, I argue that nancial expert director
is more likely to have sufcient knowledge to assess managers private information
about future performance.3
I rst expect an overall negative relationship between the likelihood that a rm will
adjust accruals to beat the analyst forecast and the intensity of nancial experts monitoring. Because audit committee members with nancial expertise are better able to
assess the quality of the information signaled by beating or missing the analyst forecast
target, managers in rms with more intensive monitoring by nancial experts would
only engage in accruals management when they have positive information about future
performance, while mangers in rms with less intensive monitoring by nancial experts
would always manage accruals to beat the target. I also expect to observe a differential
impact of nancial experts on the odds of using accruals to beat the target, contingent
upon future performance. When managers have negative information about future earnings, nancial experts on the audit committee are expected to discourage managers
from managing earnings upward to beat the analyst forecast because such adjustment
diminishes the information usefulness of earnings. Yet, when accruals adjustment
reveals managers private favorable information, nancial expert directors are expected
to allow managers to adjust accruals to beat the target.
Empirically, I identify 1460 rm-quarter observations that initially had earnings
before discretionary accruals below analyst forecasts by less than one cent. Of the 1460
observations, 1091 observations recognize enough amounts of positive discretionary
accruals and report earnings that are above analyst forecasts, while 369 observations do
not adjust earnings upward and therefore miss the target. I use next quarters earnings
surprises to capture rms future performance. Consistent with my expectation, I nd
that when there is at least one independent nancial expert on the audit committee,
rms are less likely to adjust earnings upward to beat the targets if rms next quarters
earnings fall below the analyst forecast.
While the prior results apply to nancial experts, I also investigate whether the
results are driven by accounting expertise on the audit committee. Consistent with the
argument that the effective audit committee members are those who have general
408
P.H. Hsu
409
410
P.H. Hsu
In addition, academic research generally supports the regulatory view that audit
committee nancial expertise is related to higher quality nancial reporting. Agrawal
and Chadha (2005) nd that the odds of restating nancial statements are signicantly
lower when the audit committee has an independent nancial expert on the audit committee. Farber (2005) nds that rms subject to an SEC enforcement action have fewer
nancial experts on their audit committees than a control group of similar rms.
Dhaliwal, Naiker, and Navissi (2010) document nancial expertise as positively associated with accruals quality. Zhang, Zhou, and Zhou (2007) nd that rms are more
likely to be identied with an internal control weakness if their audit committees have
less nancial expertise. As a result, in this paper, I focus on audit committee nancial
experts as dened by Section 407 of the SOX Act. I argue that these directors are better able to monitor nancial reporting and disclosure issues through their knowledge
base, educational background, or prior working experience.6
Since audit committee nancial expertise are better able to evaluate managers private information regarding future performance and to assess the quality of the information signaled by beating or missing the analyst forecast target. Managers in rms with
more intensive monitoring by nancial experts would only engage in accruals management when they have positive information about future performance, while mangers in
rms with less intensive monitoring by nancial experts would always manage accruals
to beat the target. This leads to an overall negative relationship between the unconditional likelihood that a rm will adjust accruals to beat the analyst forecast and the
intensity of nancial experts monitoring.
Hypothesis 1: The presence of an independent nancial expert on the audit committee is
negatively related to the odds of beating analyst forecasts by adjusting accruals.
411
412
P.H. Hsu
affected by the earnings surprise in quarter t, I use the latest consensus forecasts before
quarter ts earnings release to calculate the earnings surprises in quarter t + 1.
Section 407 of the SOX Act mandates rms to disclose whether there is at least one
nancial expert on the audit committee. Following Section 407, I create a dummy variable (FEDIR1) that equals one when there is at least one independent nancial expert
on the audit committee, and zero otherwise.7
3.4. Summary statistics
Table 1 provides descriptive statistics. Panel A shows that, relative to all Compustat
rms, my sample includes rms that are larger in size (log market value) and total
assets, and are more protable. Panel B shows the summary statistics for rms that
have earnings before discretionary accruals below analyst forecast by less than one
cent. As expected, the mean value of discretionary accruals (DAC) of rms that
beat the forecast is signicantly larger than the value of rms that miss the forecast.
On average, rms that beat the forecasts through accruals have signicantly less
negative adjusted forecast error (AdjFE) than rms that miss the forecast, implying
that it might be easier for those managers to adjust accruals to beat the target. The
actual forecast error (FE) of rms that beat the target is 0.001, which is signicantly larger than that of rms that miss the target. There is no signicant
difference in board composition between rms that beat the target and rms that
miss the target.
Panel C presents the distribution of independent nancial experts in my sample. Of
rms that beat the target by adjusting accruals, 10.26% do not have any independent
nancial expert on the audit committee; while of rms that choose to miss the target,
only 6.82% do not have any independent nancial expert on the audit committee. Panel
D shows the distribution of independent accounting experts on the audit committee. On
average, 85% of rms do not have any independent accounting expert on the audit
committee. This statistic is similar to prior studies. For example, Krishnan and
Visvanathan (2008) show that about 20% of rms have at least one accounting
nancial expert on the audit committee for the period 20002002.
4. Research design and empirical results
Hypothesis H1 predicts that the monitoring imposed by nancial experts reduces the
unconditional tendency to use accruals to beat the analyst forecast. To test this prediction, I estimate a logit model and predict a negative relation between the presence of a
nancial expert on the audit committee and the possibility of unconditional accruals
management to beat analyst forecast targets.
ProbJUSTBEATi;t;q 2 0; 1 logit a0 b FEDIRi;t;q c CONTROLi;t;q ni;t;q (1)
Following Matsumoto (2002), I include various control variables that might correlate with earnings surprises. I control for market value (SIZE), book-to-market ratio
(BTM), the number of outstanding shares (SHARES), the number of analysts
(NUM_ESTIMATE), dispersion of individual forecast (CV_AF), and a downward revision dummy (DOWN_REV). I also control for board characteristics. I include
BOARD_SIZE to capture the number of directors on the board and number of independent directors (BOARD_IND) to control for the independence of the board. The Blue
Descriptive statistics.
86,834
86,834
86,834
86,834
N
5.33***
5.13***
0.08***
0.010**
Mean
2.40
2.67
0.41
0.16
SD
Compustat sample
5.37
5.30
0.005
0.0004
Median
1091
1091
1091
1091
1091
1091
1091
1091
1091
1091
1091
1091
1091
Accruals measures
Discretionary accruals (DAC)
Nondiscretionary EPS (NDEPS)
Actual EPS (ACTUAL)
Consensus analyst forecast (MEDEST)
Forecast error adjusted for DAC (AdjFE)
Forecast error (FE)
Board characteristics
BOARD_SIZE
BOARD_IND
AC_SIZE
# of Financial expert
# of Independent nancial expert
# of Accounting expert
# of Independent accounting expert
11.18
7.42
3.68
1.62
1.55
0.14
0.13
0.003***
0.352**
0.356***
0.354**
0.001***
0.001***
Mean
4.02
2.86
0.88
1.97
1.08
0.38
0.36
0.008
0.306
0.306
0.306
0.002
0.007
SD
10.00
7.00
3.00
1.00
1.00
0.00
0.00
0.001
0.299
0.310
0.300
0.000
0.000
Median
Panel B: Firms with earnings before discretionary accruals below forecast by less than one cent
Beat the target
Table 1.
369
369
369
369
369
369
369
369
369
369
369
369
369
1460
1460
1460
1460
N
1.33
1.31
0.03
0.10
SD
11.34
7.52
3.78
1.61
1.56
0.13
0.13
0.000
0.302
0.300
0.310
0.007
0.009
Mean
4.14
3.00
0.99
1.09
1.07
0.36
0.36
0.002
0.303
0.302
0.302
0.303
0.004
SD
7.55
7.24
0.01
0.001
Mean
Test sample
(Continued)
10.00
7.00
4.00
1.00
1.00
0.00
0.00
0.000
0.240
0.240
0.250
0.000
0.010
Median
7.42
7.20
0.01
0.0003
Median
Table 1. (Continued).
1091
Total rms
100
%
10.26
50.97
38.77
369
#
27
204
138
1091
Total rms
100
%
85.62
13.01
1.37
369
#
316
49
4
100
%
85.63
13.27
1.10
100
%
6.82
55.28
37.40
Notes: The descriptive statistics are based on 1460 rm-quarter observations in the period 20042011 excluding rm that do not have quarterly earnings data for all quarters of
the given year. I delete all rms with Standard Industrial Classication (SIC) codes from 44005000 (utility industry), 60006999 (nancial services), and 90009999 (unclassible
industry). One thousand and ninety-one rm-quarter observations have earnings before discretionary accruals initially below forecasts by less than one cent but reports sufcient
positive discretionary accruals that allow earnings to beat analyst forecasts. Three hundred and sixty-nine rm-quarter observations have earnings before discretionary accruals initially below forecasts by less than one cent and report earnings that is below analyst forecasts. See Appendix 1 for variable measures.
***p < 0.01; **p < 0.05; *p < 0.1.
#
934
142
15
Panel D: Accounting Experts on Firms with earnings before discretionary accruals below forecast by less than one cent
Beat the target
#
112
556
423
Panel C: Financial experts on Firms with earnings before discretionary accruals below forecast by less than one cent
414
P.H. Hsu
415
Ribbon Committee suggests that audit committees should have at least three members,
implying that larger audit committees are more likely to have a wider knowledge base
on which to draw and are better able to perform their oversight duty. Thus, number of
audit committee members (AC_SIZE) is included to control for the size of the audit
committee.
Table 2 reports the results. As expected, there is a signicantly negative relation
between the measure of audit committee nancial expertise and the odds of unconditional accruals adjustments to beat the analyst forecast target. The coefcient on
FEDIR1 is 0.687 (with p value < 0.01), suggesting that the possibility of beating the
analyst forecast is lower for rm that has at least one independent nancial expert
director on the audit committee. Regarding the control variables, the coefcient on
NUM_ESTIMATE is signicantly positive, suggesting that rms are more likely to beat
the target by adjusting accruals when they have more analysts following. The signicant negative coefcient on BOARD_SIZE suggests that rms with larger boards are
less likely to use accruals to beat the target.
I then empirically test Hypothesis 2 by partitioning my sample based on rms
future nancial performance and estimating the following equation:
Prob(JUSTBEATi;t;q 2 0; 1 logit a0 b0 FEDIRi;t;q b1 BadNEWSi;t;q
FEDIRi;t;q b2 BadNEWSi;t;q c CONTROLi;t;q ni;t;q
(2)
In Equation (2), control variables are the same as Equation (1). 0 captures the
effect of the presence of at least one independent nancial expert on the odds of
Table 2. The effect of the presence of at least an independent nancial expert on the odds of
beating analyst forecast by accruals.
FEDIR1t
SIZEt
BTMt
SHARESt
NUM_ESTIMATEt
CV_AFt
DOWN_REVt
BOARD_SIZEt
BOARD_INDt
AC_SIZEt
Observations
Pseudo R2
Year & quarter xed effect
Industry xed effect
()
()
(+)
(+)
()
(+)
JUSTBEAT
2 statistics
0.687***
0.040
0.272
0.000
0.045***
0.359
0.273
0.446*
0.001
0.122
[5.069]
[0.323]
[1.316]
[0.365]
[3.604]
[1.163]
[1.160]
[1.759]
[0.0276]
[0.393]
1460
0.101
Yes
Yes
Notes: The dependent variable is JUSTBEAT, which equals one when a rm whose earnings before discretionary accruals is initially below forecasts by less than one cent but reports sufcient positive discretionary accruals that allow earnings to beat analyst forecasts, and zero, when a rm whose earnings before discretionary
accruals is initially below forecasts by less than one cent and reports earnings that is still below analyst forecasts. See Appendix 1 for variable measures. The regression is estimated with an intercept included but the
intercept is not reported. 2 statistics shown in brackets are based on robust standard errors cluster by year.
***p < 0.01; **p < 0.05; *p < 0.1.
416
P.H. Hsu
Table 3. The effect of the presence of at least an independent nancial expert on the odds of
beating target by accruals contingent on future earnings surprises.
Constant
FEDIR1t
BadNEWSt*FEDIR1t
BadNEWSt
SIZEt
BTMt
SHARESt
NUM_ESTIMATEt
CV_AFt
DOWN_REVt
BOARD_SIZEt
BOARD_INDt
AC_SIZEt
Observations
Pseudo R2
Year & quarter xed effect
Industry xed effect
Coefcient on 0 + 1
(+)
()
()
(+)
(+)
()
(+)
0
0
1
2
JUSTBEAT
2 statistics
0.870
0.280
0.808***
0.397***
0.026
0.290
0.000
0.047***
0.369
0.231
0.416
0.008
0.076
[1.360]
[4.351]
[2.784]
[0.201]
[1.388]
[0.417]
[3.989]
[1.128]
[1.004]
[1.506]
[0.362]
[0.237]
[0.695]
1460
0.105
Yes
Yes
1.088***
Notes: The dependent variable is JUSTBEAT. BadNEWS is a dummy variable that equals one if a rm fails
to beat analyst forecasts in the following quarter, and zero otherwise. See Appendix 1 for variable measures.
2 statistics shown in brackets are based on robust standard errors cluster by year.
***p < 0.01; **p < 0.05; *p < 0.1.
417
Table 4. The effect of the presence of at least an accounting expert on the odds of beating
analyst forecast by accruals contingent on future earnings surprises.
ACCDIR1t
BadNEWSt*ACCDIR1t
NonACC_FEDIR1t
BadNEWSt*NonACC_FEDIR1t
BadNEWSt
SIZEt
BTMt
SHARESt
NUM_ESTIMATEt
CV_AFt
DOWN_REVt
BOARD_SIZEt
BOARD_INDt
AC_SIZEt
Observations
Pseudo R2
Year & quarter xed effect
Industry xed effect
0
1
2
3
()
(+)
(+)
()
(+)
JUSTBEAT
2 statistics
0.152
0.687**
0.246
0.533**
0.293*
0.042
0.325***
0.001
0.045***
0.408
0.109
0.507
0.001
0.157
[0.654]
[2.012]
[1.590]
[2.260]
[1.750]
[0.276]
[2.666]
[0.487]
[3.043]
[0.996]
[0.363]
[1.538]
[0.0526]
[0.427]
1106
0.125
Yes
Yes
Notes: The dependent variable is JUSTBEAT. ACCDIR1 is dummy variable, which equals one when there is
at least one independent accounting expert on the audit committee, and zero otherwise. NonACC_FRDIR1 is
dummy variable, which equals one when there is at least one independent nonaccounting nancial expert on
the audit committee, and zero otherwise. See Appendix 1 for variable measures. The regression is estimated
with an intercept included but the intercept is not reported in 2 statistics shown in brackets based on robust
standard errors cluster by year.
***p < 0.01; **p < 0.05; *p < 0.1.
418
P.H. Hsu
Table 5. The effect of the presence of at least an independent nancial expert on the odds of
beating analyst forecast by accruals contingent on future earnings surprises: two-stage least
squares.
FEDIR1 Dummya
(1)
FEDIR1
^ t
FEDIR1
^ t *BadNEWSt
FEDIR1
BadNEWSt
SIZEt
BTMt
SHARESt
NUM_ESTIMATEt
CV_AFt
DOWN_REVt
BOARD_SIZEt
BOARD_INDt
AC_SIZEt
COMPANY_AGE
FCFt
R&Dt
SEGMENTt
CEO_TENUREt
LEVERAGEt
Observations
Pseudo R2
Year & quarter xed effect
Industry xed effect
Coefcients on 0 + 1
0.031**
0.063**
0.000
0.001
0.019
0.038
0.098
0.010*
0.079
0.034
0.012
0.006
0.002
0.001
0.033
2 statistics
[2.696]
[2.859]
[0.843]
[0.382]
[0.581]
[1.839]
[1.376]
[1.945]
[0.809]
[1.773]
[1.210]
[0.643]
[0.533]
[0.981]
[1.586]
JUSTBEAT
(2)
BEAT
2 statistics
2.981
1.174**
0.884**
0.166
0.596
0.000
0.039
0.088
0.131
1.040***
0.010
0.005
[1.398]
[2.101]
[1.994]
[0.675]
[1.620]
[0.004]
[1.627]
[0.200]
[0.544]
[2.991]
[0.174]
[0.010]
796
0.404
796
0.148
Yes
Yes
4.155***
Notes: Logistic regression of rst-stage model is estimated with FEDIR1 as the dependent variable (= 1 for
rms with at least one independent nancial expert, and zero otherwise). See Appendix 1 for variable measures.
All regressions are estimated with an intercept included but the intercept is not reported. 2 statistics shown in
brackets are based on robust standard errors cluster by year.
***p < 0.01; **p < 0.05; *p < 0.1.
a
419
6. Additional tests
6.1. Control for endogenous
One concern regarding the prior results is the possibility of a reverse causal relation.
This would occur if nancial experts with knowledge about the rms accounting systems systematically opt out of serving on boards of rms with low nancial reporting
quality. I address this endogeneity problem using two approaches. First, I estimate
simultaneous equations in the odds of beating targets by adjusting accruals and number
of nancial experts using 2SLS regressions. My rst-stage model captures the determinants of board structure, including monitoring and advising costs, information asymmetry, and business complexity (Coles, Daniel, and Naveen 2008).
Table 5 reports the parameter estimates. Column (1) presents the results on the
determinants of the presence of nancial expert director. FEDIR1 is positively related
to BOARD_IND. In addition, larger rms and rms with higher book-to-market ratio
are more likely to have at least one nancial expert on the audit committee. In the second stage, I replace FEDIR1 with the predicted value from the rst-stage model. I continue to nd a negative total effect of (predicted) FEDIR1 and the odds of beating
target by accruals when the rms have negative earnings surprise in quarter t + 1 (see
last row of Column (2) Table 5). In addition, the signicant negative coefcient on 1
suggests a signicant incremental effect of nancial experts on the odds of beating target for rms with negative earnings surprises.
While my results are qualitatively similar under 2SLS, this method could be subject
to specication error. Thus, following Hermalin and Weisbach (1991), I use lagged
(instead of contemporaneous) values of number of nancial experts in regressions of
the odds of beating analyst forecast targets, and all the results are unchanged.
6.2. Mangers exibility in using accruals to beat the target
Although managers have strong incentive to manage earnings to beat the target, prior
studies suggest that managers are not always able to do so (Barton and Simko 2002;
Hunt, Moyer, and Shevlin 1996). By assuming that rms with unmanaged earnings just
below forecasted target are more capable in adjusting accruals, my main results are
based on observations with earnings before discretionary accruals below analyst forecasts by less than one cent (Davis, Soo, and Trompeter 2009).9 Yet, one cent is an
arbitrage cut-off. In order to examine whether my results are sensitive to the choice of
cut-off, I construct samples based on rms with unmanaged earnings below analyst
forecasts by less than two cents, three cents, and four cents, respectively. Untabulated
results are robust when using alternative cut-offs.
Although above results show that my ndings are not sensitive to different cut-offs,
it is still possible that rms missing forecasts by less than couple cents are not able to
adjust accruals. In order to address such concern, I relax my assumption that managers
in rms with unmanaged earnings just below forecasted target are able to adjust accruals and construct a matched sample. Ideally, matching procedure yields control observations (rms do not use accruals to beat the target) that are identical to treatment
observations (rms with unmanaged earnings below the target but use accruals to beat
the target) in respect of the exibility in adjusting accruals. I conduct a two-step matching procedure. First, an observation can be selected as control observation when it has
discretionary accruals initially below forecast and reports earnings that is still below
analyst forecasts. I then match control observations with treatment observations based
420
P.H. Hsu
on year, prior-quarter NOAs, and industry (two-digit SIC code) (Barton and Simko
2002).10 Untabulated results from matched sample are consistent with my main
ndings.
6.3. Predictive ability of accruals
In order to make my results more generalizable, following prior study (Chang et al.,
2012), I examine whether independent audit committee nancial expert improves the
association between accruals component and future earnings. By arguing that rms with
more intensive nancial expert monitoring are more likely to use accruals to signal
future performance, the accrual component should better predict future earnings when
rms have nancial expert on the audit committee. Following Sloan (1996), I estimate
following Equation (3):
EARNi;t;q1 a b1 CFOi;t;q b2 ACCRi;t;q b3 CFOi;t;q FEDIRIi;t;q
b4 ACCRi;t;q FEDIRIi;t;q ni;t;q
(3)
The results of estimating Equation (3) are shown in Table 6. Consistent with prior
studies, the coefcients on 1 and 2 are signicantly positive (Sloan 1996). As
expected, I nd a signicantly positive 4, suggesting that the relation between the current periods accruals component and the following periods earnings is stronger when
the rm has at least one independent nancial expert on the audit committee.
6.4. Does board independence matter?
The key variable (FEDIR1) is dened as the presence of at least one independent
nancial expert on the audit committee. Since prior studies have documented a positive
relation between board independence and earning quality (Beasley 1996; Farber 2005),
it is not clear whether my prior ndings are driven by independent directors monitoring or by their nancial background. To address such concern, I examine whether independent non-nancial directors affect the odds of using accruals to beat analyst
Table 6. The effect of the presence of at least a nancial expert on the predictive ability of
accruals.
EARNt
CFOt
ACCRt
CFOt*FEDIR1t
ACCRUALt*FEDIR1t
Observations
Adjusted R2
Year & quarter xed effect
Industry xed effect
(+)
(+)
(+)
1
2
3
4
+ 1
0.582***
0.315***
0.053
0.159**
t-statistics
[12.81]
[11.55]
[1.071]
[2.800]
14,658
0.381
Yes
Yes
Notes: The dependent variable is EARNt + 1. See Appendix 1 for variable measures. The regression is
estimated with an intercept included, but the intercept is not reported. t-Statistics shown in brackets are based
on robust standard errors cluster by year.
***p < 0.01; **p < 0.05; *p < 0.1.
421
422
P.H. Hsu
restricts managers ability to use accruals, I focus on rms that have beginning NOAs
above the median of the corresponding industry-year-quarter NOAs or/and rms with
beginning length of operating cycle below the median of the corresponding industryyear-quarter operating cycle (Barton and Simko 2002; Zang 2011). I then re-estimate
Table 3 using RMPROD or/ RMDISX to substitute for discretionary accruals.
The results from using real earnings management as an alternative way to beat the target are similar to my main ndings. Untabulated results show a negative impact of nancial experts on the odds of using overproduction or/ cutting discretionary expenditures to
beat the target by when future performance is poor. In addition, I also nd that impact of
independent nancial experts on the odds of beating the targets by decreasing cost of
goods sold (inventory overproduction) is greater when future performance is poor.
7. Conclusion
Although prior evidence suggests that managers adjust accruals to beat the analyst forecast, the implication of accruals adjustment on earnings quality is not clear. When managers have negative information regarding future earnings performance, accruals
adjustment mislead shareholders and result in lower quality of nancial reporting. However, accruals signal strong future earnings growth and thereby improve the decision
usefulness of earnings. In this study, I examine whether the monitoring imposed by
nancial experts on the audit committee affect managers accruals adjustment to beat
analyst forecasts conditional upon future performance.
I argue that audit committee nancial experts who have sufcient nancial knowledge are better able to assess and understand managers private information and are
more capable of overseeing the nancial reporting process. Thus, when future performance is poor, they would intervene in accruals adjustment to beat the target when
future performance is poor. Consistent with my predictions, I nd that the presence of
at least one independent nancial expert on the audit committee is negatively related to
the use of accruals to beat analyst forecasts when the rm has next quarters earnings
below analyst forecasts. In addition, the negative effect of nancial experts on the odds
of beating the target by adjusting accruals is greater for rms with negative earnings
surprises than for rms with positive earnings surprises in quarter t + 1.
I also examine the impact of accounting experts on the quality of accruals and nd
that the presence of both accounting and nonaccounting expertise on the audit committee is associated with higher quality of accruals. This evidence is consistent with prior
studies and suggests that the SECs wide-ranging denition of nancial expertise may
not be a compromise to allay public criticism, but rather reects the need for broader
expertise on the board.
Notes
1.
2.
3.
Hereafter, I use beat the consensus analyst forecast to refer to just meet or beat the analyst forecast target.
Firms can also guide analyst forecasts downward to avoid missing targets (Matsumoto
2002). Expectation management is excluded from my analysis because it is a reporting strategy and does not affect earnings quality.
Following SEC, a nancial expert director has an understanding of generally accepted
accounting principles (GAAP); has the experience in preparing, auditing, analyzing, or evaluating nancial statements; and has the ability to access the accounting principle or has an
understanding of internal controls and procedures for nancial reporting (SEC 2003).
423
4.
Although managers also have incentive to lower earnings in order to miss the target, the
evidence is rare. The only explanation is that CEOs consider their stock options to more
likely miss targets. Missing targets can reward a CEO with option grants that are pegged to
a lower stock price (McAnally, Srivastava, and Weaver 2008).
5. Rather than being completely aligned with the interests of shareholders, boards interests
can be aligned with managers interests (i.e. the board consists of individuals assumed to be
beholden to the CEO). I address this concern by examining independent nancial experts in
my empirical tests. Independent directors who are concerned about their reputation are less
likely to align with managers (Byrd and Hickman 1992).
6. An audit committee nancial expert is a person who has an understanding of GAAP; experience in preparing, auditing, analyzing, or evaluating nancial statements; and the ability to
understand the accounting principle or to understand internal controls and procedures for
nancial reporting (SEC 2003). Find more information in SOX Section 407.
7. The focus on the independence of nancial experts is based on several reasons. First, prior
studies generally suggest that the interests of independent directors are more likely to be
aligned with shareholders due to concerns about their reputation (Byrd and Hickman 1992).
In addition, Section 301 of the SOX Act mandates the SEC to direct the national securities
exchange to require the listing company to have all independent audit committee members.
Even though not all reporting companies are listed on a national securities exchange or
association, Section 407 of the SOX Act explicitly requires a company to disclose whether
the nancial expert is independent of management because the SEC believes that investors
in these companies would be interested in knowing whether the audit committee nancial
expert is independent of management.
8. A nancial expert is classied as an accounting expert if the director has experience as a
public accountant, auditor, principal or chief nancial ofcer, controller, or principal or chief
accounting ofcer (Defond, Hann, and Hu 2005). Because RiskMetrics does not provide
information regarding nancial experts prole, the sample period for this test is restricted
from 2004 to 2008, the period covered by Corporate Library. This subsample consists of
1106 rm-quarter observations.
9. Prior research shows that the distributions of earnings surprises contain an unusually high
frequency of small positive surprises and an unusually low frequency of small negative surprises (Degeorge, Patel, and Zeckhauser 1999), implying that managers boost earnings
upward from just-miss-target to just-beat-target. Other studies also show that managers use
various accounting choices to achieve small positive earnings surprises (Burgstahler and
Eames 2006). Following these prior studies, Davis, Soo, and Trompeter (2009) directly
focus on rms with unmanaged earnings that just miss the forecast by less than one cent
and argue that such stricter metric is more likely to capture accruals management.
10. Prior studies also suggest that the length of operating cycle restricts managers exibility in
adjusting accruals (Zang 2011). I matched control observations with treatment observations
based on year, prior-quarter length of operating cycle, and industry. The results are consistent with the results when using NOA as matched factor.
References
Agrawal, A., and S. Chadha. 2005. Corporate Governance and Accounting Scandals. The
Journal of Law and Economics 48 (2): 371406.
Ayers, Benjamin C., John Jiang, and P. Eric Yeung. 2006. Discretionary Accruals and Earnings
Management: An Analysis of Pseudo Earnings Targets. The Accounting Review 81 (3):
617652.
Balsam, Steven, Eli Bartov, and Carol Marquardt. 2002. Accruals Management, Investor Sophistication, and Equity Valuation: Evidence from 10Q Filings. Journal of Accounting
Research 40 (4): 9871012.
Barton, Jan, and Paul J. Simko. 2002. The Balance Sheet as an Earnings Management
Constraint. The Accounting Review 77 (4): 127.
Beasley, Mark S. 1996. An Empirical Analysis of the Relation Between the Board of Director
Composition and Financial Statement Fraud. The Accounting Review 71 (4): 443465.
Blue Ribbon Committee on Improving the Effectiveness of Corporate Audit Committees. 1999.
Report and Recommendations of the Blue Ribbon Committee on Improving the Effectiveness
424
P.H. Hsu
of Corporate Audit Committees. New York: New York Stock Exchange and the National
Association of Securities Dealers.
Burgstahler, David, and Michael Eames. 2006. Management of Earnings and Analysts Forecasts
to Achieve Zero and Small Positive Earnings Surprises. Journal of Business Finance &
Accounting 33 (56): 633652.
Byrd, John W., and Kent A. Hickman. 1992. Do Outside Directors Monitor Managers?:
Evidence from Tender Offer Bids. Journal of Financial Economics 32 (2): 195221.
Chang, Hsihui, Sang-Hyun Suh, Edward M. Werner, and Jian Zhou. 2012. The Effect of SOX
on the Predictability of Future Cash Flows in Litigious and Non-Litigious Industries. AsiaPacic Journal of Accounting & Economics 19 (2): 210226.
Cheng, Qiang, and Terry D. Wareld. 2005. Equity Incentives and Earnings Management. The
Accounting Review 80 (2): 441476.
Cohen, Daniel A., and Paul Zarowin. 2010. Accrual-based and Real Earnings Management
Activities Around Seasoned Equity Offerings. Journal of Accounting and Economics 50 (1):
219.
Coles, Jeffrey L., Naveen D. Daniel, and Lalitha Naveen. 2008. Boards: Does One Size Fit
All? Journal of Financial Economics 87 (2): 329356.
Davis, Larry R., Billy S. Soo, and Gregory M. Trompeter. 2009. Auditor Tenure and the Ability
to Meet or Beat Earnings Forecasts. Contemporary Accounting Research 26 (2): 517548.
Dechow, Patricia M., Richard G. Sloan, and Amy P. Sweeney. 1995. Detecting Earnings Management. The Accounting Review 70 (2): 193225.
Defond, Mark L., Rebecca N. Hann, and Xuesong Hu. 2005. Does the Market Value Financial
Expertise on Audit Committees of Boards of Directors? Journal of Accounting Research 43
(2): 153193.
Degeorge, Francois, Jayendu Patel, and Richard Zeckhauser. 1999. Earnings Management to
Exceed Thresholds. The Journal of Business 72 (1): 133.
Dhaliwal, D. A. N., V. I. C. Naiker, and Farshid Navissi. 2010. The Association Between
Accruals Quality and the Characteristics of Accounting Experts and Mix of Expertise on
Audit Committees*. Contemporary Accounting Research 27 (3): 787827.
Farber, David B. 2005. Restoring Trust After Fraud: Does Corporate Governance Matter? The
Accounting Review 80 (2): 539561.
Hermalin, Benjamin E., and Michael S. Weisbach. 1991. The Effects of Board Composition and
Direct Incentives on Firm Performance. Financial Management 20 (4): 101112.
Hunt, Alister, Susan E. Moyer, and Terry Shevlin. 1996. Managing Interacting Accounting
Measures to Meet Multiple Objectives: A Study of LIFO Firms. Journal of Accounting and
Economics 21 (3): 339374.
Klein, April. 2002. Audit Committee, Board of Director Characteristics, and Earnings Management. Journal of Accounting and Economics 33 (3): 375400.
Krishnan, Gopal V., and Gnanakumar Visvanathan. 2008. Does the SOX Denition of an
Accounting Expert Matter? The Association Between Audit Committee Directors Accounting Expertise and Accounting Conservatism. Contemporary Accounting Research 25 (3):
827858.
Matsumoto, Dawn A. 2002. Managements Incentives to Avoid Negative Earnings Surprises.
The Accounting Review 77 (3): 483514.
Matsunaga, Steven R., and Chul W. Park. 2001. The Effect of Missing a Quarterly Earnings
Benchmark on the CEOs Annual Bonus. The Accounting Review 76 (3): 313332.
McAnally, Mary Lea, Anup Srivastava, and Connie D. Weaver. 2008. Executive Stock Options,
Missed Earnings Targets, and Earnings Management. The Accounting Review 83 (1):
185216.
McVay, Sarah, Venky Nagar, and Vicki Wei Tang. 2006. Trading Incentives to Meet the Analyst
Forecast. Review of Accounting Studies 11 (4): 575598.
Roychowdhury, Sugata. 2006. Earnings Management Through Real Activities Manipulation.
Journal of Accounting and Economics 42 (3): 335370.
Sarbanes-Oxley Act of 2002 [SOX]. 2002. Public Law No. 107-204, 116 Stat. 745, Sec. 1-1107.
SEC. 2003. Final Rule: Disclosure Required by Sections 301 and 407 of the Sarbanes-Oxley Act
of 2002. Washington, DC: SEC.
Sloan, Richard G. 1996. Do Stock Prices Fully Reect Information in Accruals and Cash Flows
About Future Earnings? The Accounting Review 71 (3): 289315.
425
Measures
% INDEP NON-FEDIR
Board measurements
BOARD_SIZE
BOARD_IND
AC_SIZE
AC_IND
FEDIR1
ACCDIR1
NonACC_FEDIR1
# INDEP NON-FEDIR
MEDEST
NDEPS
DAC
FE
AdjFE
Accruals measurements
ACTUAL
ACCR
Variables
426
P.H. Hsu
SEGMENT
SIZE
SHARES
R&D
OPERATING CYCLE
NUM ESTIMATE
NOA
EARN
DOWN_REV
LEVERAGE
COMPANY_AGE
CV_AF
Other variables
CEO_TENURE
CFO
Log (Number of years that current CEO serves in company) for rm i year t quarter q
Cash ows from continuing operations for rm i year t quarter q, measured as extraordinary items and discontinued
operations (Statement of Cash Flows) plus operating activities (XIDOCY + OANCFY), scaled by total assets (ATQ)
for quarter t 1
Log (Number of years company has been in business) for rm i year t quarter q
Coefcient on variation of the consensus forecast used to calculate earnings surprise for rm i year t quarter q, which is dened
as standard deviation scaled by the mean
Dummy variable equals one if long-term debt (DLTTQ) for rm i year t quarter q, scaled by total assets (ATQ) for quarter t 1,
is greater than the median value pooled over the sample period, and zero otherwise
Analyst downward revision for rm i year t quarter q, which equals one if at least one of the rms analysts revised his or her
forecast downward in the three months prior to the earnings announcement for quarter t, and zero otherwise
Earnings for rm i year t quarter q, which is measured as income before extraordinary Items (IBCY), scaled by total assets
(ATQ) for quarter t 1
Net operating assets for rm i year t, which is measured as shareholders equity (CEQQ) less cash and marketable securities
(CHEQ) plus total debt (DLCQ + DLTTQ), scaled by sales (SALEQ)
Number of analysts for rm i year t quarter q, which is measured as number of analysts whose forecasts are included in the
consensus forecast used to calculate earnings surprise
The operating cycle for rm i year t quarter q, which is computed as the days receivable plus the days inventory less the
days payable
Dummy variable equals one when the R&D expenditure (XRDQ) for rm i year t quarter q is less than industry median
for a given year, and zero otherwise
Log (Number of business segment and geographic segments) for rm i year t quarter q
Size of the rm for rm i year t quarter q, which is measured as natural logarithm of market value (PRCCQ*CSHOQ)
Number of common shares (CSHOQ) outstanding for rm i year t quarter q