-
Electronic copy available at:
http://ssrn.com/abstract=2427530
1
Stock Liquidity and Earnings Management*
Jing Fang+
School of Accounting and Finance
The Hong Kong Polytechnic University
* I am grateful to Alex Edmans and Vivian W. Fang for their
insightful comments on the previous version
of this paper, to members of my dissertation committee (C.S.
Agnes Cheng (chair), Dana Hollie, Joseph
Legoria, and Ji-Chai Lin), to Joshua Ronen for stimulating ideas
that underlie several additional tests of
this paper, and to seminar participants at the Louisiana State
University and the AAA 2012 annual
meeting for their helpful comments. + Email:
[email protected], School of Accounting and Finance, the
Hong Kong Polytechnic
University, Hong Kong.
-
Electronic copy available at:
http://ssrn.com/abstract=2427530
2
Stock Liquidity and Earnings Management
Abstract
We find a negative relation between stock liquidity and earnings
management. Our
finding is insensitive to specific stock liquidity and earnings
management measures used and
remains strong after controlling for a comprehensive list of
possible covariates including firm
fixed effects. We use the 2001 decimalization as a
quasi-experiment to run a difference-in-
differences analysis (DiD) and find that firms experiencing
greater stock liquidity improvement
encounter significantly greater drop in earnings management
after the decimalization than their
matched pairs that are otherwise similar. Our DiD finding
suggests a causal negative effect of
stock liquidity on earnings management. Using the matched sample
for the DiD analysis we also
find differential changes in analyst followings and efforts,
institutional ownership, and stock
price efficiency surrounding the decimalization, which is
consistent with our hypothesized
mechanisms through which stock liquidity deters earnings
management.
JEL classification: G14, M12, M41, M52.
Keywords: Stock Liquidity, Earnings Management, Difference in
Differences, Decimalization.
-
3
1. Introduction
In this study, we examine the effect of stock liquidity on
accounting-based earnings
management.1 We hypothesize a negative between stock liquidity
and earnings management.
One of the most cited motives for managers to manage reported
earnings is that managers
attempt to maneuver investors expectations about their firms
economic prospects through
earnings management and move their firms stock prices in their
preferred directions at their
desired magnitudes. Stock liquidity encourages private
information production by investors by
increasing the marginal value of information, shift shareholder
base toward large, sophisticated
investors by inducing the formation of blockholdings, and spur
arbitrage by reducing costs and
risks related to arbitrage. Therefore, we reason that stock
liquidity makes it difficult for managers
to move their firms stock prices through earnings management.
Moreoever, earnings
management is value-destroying because it consumes valuable
organizational resources,
especially top executives limited time. Stock liquidity ensures
that stock prices could timely and
faithfully reflect the value-destroying consequence of earnings
management. Therefore, we
reason that stock liquidity also makes it less beneficial for
managers to engage in earnings
management by timely and faithfully revealing the
value-destroying consequence of earnings
management in stock prices.
We find a negative relation between stock liquidity and earnings
management. Our
finding is robust to specific stock liquidity and earnings
management measures used. In the main
test, we find a negative relation between the absolute value of
discretionary accruals and the
stock liquidity measure proposed in Corwin and Schultz (2012).
Using the stock liquidity
measure proposed in Hasbrouck (2009), our finding remains
unchanged. Moreover, we find that
stock liquidity is also negatively related to the likelihood of
a SEC Accounting and Auditing
Enforcement Release (AAER) and the likelihood of an accounting
restatement. More importantly,
we show that the negative relation between stock liquidity and
the absolute value of discretionary
accruals remains strong after controlling for a long list of
possible covariates and firm - / year - /
1 In all following paragraphs, we mean accounting-based earnings
management by earnings management unless
stated otherwise.
-
4
industry- fixed effects, suggesting that our finding may not be
confounded by omitted correlated
variables.
To establish causality, we adopt the difference-in-differences
(DiD) approach. We use the
decimalization surrounding 2001 as a quasi-experiment to examine
how the exogenous variation
in stock liquidity generated by the decimalization affects
changes in earnings management.
Specifically, we match firms with changes in stock liquidity in
the top one-third (namely the
treatment firms) with firms with changes in stock liquidity in
the bottom one-third (namely the
control firms) by using a one-to-one nearest neighbor propensity
score matching without
replacement. Our matching ensures that the treatment firms and
the control firms are similar
along a host of characteristics right before the
decimalization.
Using the DiD approach, we find that the treatment firms
experience a significant drop in
the absolute value of discretionary accruals while the control
firms experience no significant
change in the absolute value of discretionary accruals. In
addition, we find no significant changes
in the absolute value of discretionary accruals for both the
treatment firms and the control firms
in the year immediately prior to the decimalization. We also
find no significant difference
between the treatment firms and the control firms regarding the
change in the absolute value of
discretionary accruals in the year immediately prior to the
decimalization. Overall, our finding
from the DiD analysis suggests that stock liquidity has a causal
negative effect on earnings
management.
We use the matched sample constructed for the DiD analysis to
validate underlying
mechanisms through which stock liquidity dampens managers
incentives to manage reported
earnings. We use financial analysts information production
efforts as a rough indicator of
overall information production efforts by market participants.
We find that the control firms
experience significant decrease in analyst followings after the
decimalization while the treatment
firms experience no significant change in analyst followings. We
also find that the treatment
firms experience significant increase in analyst efforts while
the control firms experience no
significant change in analyst efforts.
With respect to the effect of stock liquidity on shareholder
base composition, we find that
the treatment firms experience significantly greater increase in
institutional holdings owned by
-
5
transient investors that are found to be better informed than
other types of institutional investors
and by dedicated investors that arguably have greater incentives
to deter earnings management.
Nevertheless, we find that both the treatment firms and the
control firms experience similar
significant decreases in institutional holdings owned by
quasi-indexing investors that arguably
have no incentives to engage in active information production or
monitoring.
Regarding the effect of stock liquidity on arbitrage, we find
that both the treatment firms
and the control firms experience significant increases in short
interests while increases in short
interests are not significantly different between the treatment
firms and the control firms. We
argue that it may not be surprising to observe no significant
difference between the treatment
firms and the control firms regarding increases in short
interests because the threat of ex-post
arbitrage deters ex-ante earnings management and less earnings
management leaves fewer
opportunities for short arbitrage. Nevertheless, we use a broad
sample and document strong
cross-section evidence about the positive relation between stock
liquidity and short interests.
To examine the effect of stock liquidity on stock price
efficiency, we measure stock price
efficiency as the extent to which stock prices deviate from the
fundamental value of traded
stocks. We find that the treatment firms experience significant
improvement in stock price
efficiency after the decimalization while the control firms
experience either no significant change
or significant deterioration in stock price efficiency depending
on the fundamental value measure
used. Overall, using the matched sample constructed for the DiD
analysis we find that changes in
analyst followings and efforts, in institutional holdings owned
by different types of investors, in
short interests, and in stock price efficiency are generally in
line with our hypothesized
mechanisms.
In addition, using two different approaches to estimating the
fundamental value of traded
stocks we document a significant positive relation between
discretionary accruals and equity
value errors (i.e., the difference between market value and
estimated fundamental value) for
firms with stock liquidity below the sample median while we find
no significant relation for
firms with stock liquidity above the sample median. Our finding
is consistent with our reasoning
that stock liquidity makes it difficult for managers to move
stock prices in their preferred
directions at preferred magnitudes through earnings
management.
-
6
Furthermore, we find that the negative relation between stock
liquidity and the absolute value
of discretionary accruals is stronger for firms whose CEOs
equity portfolio delta is above the
sample median than for firms whose CEOs equity portfolio delta
is below the sample median.
We measure a CEOs equity portfolio delta as the sensitivity of
the CEOs equity holdings to a 1%
change in stock prices. To ensure comparability across periods
and firms we scale a CEOs
equity portfolio delta by the CEO total cash compensation (i.e.,
the ratio of the CEOs equity
portfolio delta to the sum of the CEOs equity portfolio delta
and total cash compensation).
Earnings management is costlier to managers whose equity
portfolio delta is greater when stock
prices could timely reflect the value-destroying consequence of
earnings management. Therefore,
our finding is consistent with our reasoning that stock
liquidity makes it less beneficial for
managers to engage in earnings management by timely and
faithfully showing the value-
destroying consequence of earnings management in stock
prices.
Our study adds to the growing literature that examines the real
effects of stock markets in
general and the role of efficient stock prices in disciplining
managers in particular.2 It has long
been recognized that if stock prices could timely and faithfully
reflect the consequence of
managers actions, managers would have greater incentives to take
desirable actions and avoid
undesirable ones (Fama 1980; Fishman and Hagerty 1989). Our
study provides evidence
that stock liquidity, the most important aspect of market
microstructure with respect to its effect
on stock price efficiency (Chordia et al. 2008; OHara 2003),
deters earnings management. In
our best knowledge, our study is the first empirical work that
documents a causal negative effect
of stock liquidity on earnings management.
Our paper also adds to the growing literature that links stock
liquidity to firm performance.
For instance, Fang et al. (2009) find a positive relation
between stock liquidity and firm
performance. Studies following Fang et al. (2009) attempt to
identify potential governance
channels through which stock liquidity contributes to firm
value. For instance, Edmans et al.
(2013) find that liquidity facilitates governance through
intervention and, to a greater degree,
through trading (also see Bharath et al. 2013). Our study
suggests a new channel through which
2 Interested readers can refer to Bond et al. (2012) for their
excellent review of research work that examines the real
effects of stock markets.
-
7
stock liquidity contributes to firm value. That is, stock
liquidity deters earnings management that
consumes valuable organizational resources, especially top
managers limited time and attention.
Our study complements existing accounting research that examines
the relation between
accounting standards and choices and market efficiency. Existing
accounting research focuses on
how accounting standards and choices affect stock market
efficiency. Our finding suggests that
stock liquidity and ensuing changes also affect managers
decisions about accounting choices in
general and earnings management in particular. Our finding about
the causal effect of stock
liquidity on earnings management suggests a more complete view
about the relation between
stock market efficiency and accounting choices including
earnings management. Our study
demonstrates the importance of taking into account factors
affecting the price discovery process
in understanding accounting phenomena and in accounting research
design (also see Lee 2001).
Our study suggests a stock liquidity-based explanation to the
over-time variation in aggregate
earnings management. We find that in line with our cross-section
evidence the over-time
variation in aggregate stock liquidity well accounts for the
over-time variation in aggregate
earnings management. Interestingly, we find that during 2000
2005 (six years in total) decline
in aggregate absolute value of discretionary accruals occurred
simultaneously with improvement
in aggregate stock liquidity ( = -0.94). Prior studies (e.g.,
Cohen et al. 2008) attribute the
decline in aggregate absolute value of discretionary accruals
during 2002-2005 to the passage of
SOX. Our findings suggest that in addition to SOX and other
concurrent events improvement in
stock liquidity and ensuing changes in information production
activities, shareholder base
composition, and arbitrage may also serve as a fundamental
factor that drives the decline in the
aggregate absolute value of discretionary accruals during
2000-2005.3
The rest of the paper is organized as follows. In section 2, we
develop the hypothesis. Section
3 describes data sources, samples, empirical measures for stock
liquidity and earnings
management, and descriptive statistics for the sample used in
the main test of our hypothesis.
Section 4 presents results from the main test. Section 5
presents results from three robustness
3 Besides reductions in the minimal tick size, other concurrent
changes also affect trading costs and, therefore, stock
liquidity (see Chordia et al. 2011). For instance, institutional
commissions decline over time; advancement in
technology makes it easier for institutions to execute automated
algorithmic trading and online brokerage accounts
make trading easier for retail investors.
-
8
tests, the test that controls for more possible covariates and
firm fixed effects, the DiD test, and
other further analyses. Section 6 concludes.
2. Hypothesis Development
Anecdotal cases, survey of executives, and findings of archival
studies suggest that
managers engage in earnings management (see Dechow et al. 2011;
Graham et al. 2005; Healy
and Palepu 2003). Various motives underlie managers earnings
management decisions such as
avoidance of debt covenant violations, and maximization of their
compensations (Fields et al.
2001; Healy and Wahlen 1999). Among the most often cited motives
is that managers manage
reported earnings in an attempt to maneuver market participants
expectation of their firms
economic fundamentals and move their firms stock prices in
desired directions.
Findings of prior studies suggest that, at least to some extent,
managers succeed in
maneuvering investors expectation of their firms economic
fundamentals through earnings
management. For instance, Bartov et al. (2002) find that firms
that manage reported earnings to
meet or beat analysts earnings expectations (MBE) command a
valuation premium compared
with firms that do not manage reported earnings and fail to MBE.
Findings with implications
similar to Bartov et al. (2002) are documented in Barth et al.
(1999), Kasznik and McNichols
(2002), and Skinner and Sloan (2002).
We argue that stock liquidity affects the extent to which
managers succeed in moving
stock prices toward their preferred directions at their desired
magnitudes through earnings
management as a result of its effect on the value of
information, the composition of shareholder
base, and information-based arbitrage. Stock liquidity reduces
trading costs and assists informed
market participants in disguising their private information.
Thus, stock liquidity enables market
participants to profit from private information (Holmstrm and
Tirole 1993; Kyle and Vila 1991;
Maug 1998). As a result, stock liquidity motivates private
information acquisition (Grossman
and Stiglitz 1980; Holmstrm and Tirole 1993). We reason that
when stock liquidity is high and,
consequently, market participants spend great resources in
information production including
studying the value implication of reported earnings, it becomes
difficult for managers to mislead
market participants through earnings management.
-
9
Both analytical predictions and empirical findings suggest that
stock liquidity encourages
the formation of blockholdings (Brav et al. 2010; Edmans 2009;
Edmans et al. 2013; Maug 1998;
Kyle and Vila 1991). Blockholders arguably possess superior
information about firms
fundamental values. First, blockholders have incentives to
become informed as a result of the
large amount that blockholders can sell upon negative
information (Edmans 2009). Second,
because quality information acquisition incurs fixed costs such
as hiring well-trained analysts
shareholders will only acquire information on large ownership
stakes (Boehmer and Kelley
2009). Moreover, blockholders possess better capabilities of
conducting high-quality
fundamental analysis as a result of their scale and resources
(Bushee and Goodman 2007).
Furthermore, blockholders have better access to management
because of their large equity
holdings (Bushee and Goodman 2007). Existing research has
accumulated considerate evidence
that confirms blockholders information superiority over the
general public. 4 For instance,
Collins et al. (2003) find that the accrual component of
reported earnings is less mispriced in
firms with greater institutional ownership. We reason that when
stock liquidity shifts the
shareholder base towards sophisticated, large investors it
becomes difficult for managers to
maneuver market participants expectation of their firms economic
fundamentals through
earnings management.
Stock liquidity stimulates information-based arbitrage.
Arbitrage is costly and risky (Lee
2001; OHara 2003; Shleifer and Vishny 1997). Stock liquidity
reduces trading costs and allows
arbitrageurs to quickly alter their holding positions at prices
that do not fully reveal their private
information. Therefore, stock liquidity makes information-based
arbitrage lucrative. Stock
liquidity also mitigates risks related to arbitrage by easing
trading between investors and
facilitating establishment and reestablishment of arbitrage
positions. Arbitrageurs are generally
well-informed, possibly motivated by the great risk and cost to
be overcome by them. For
instance, Karpoff and Lou (2010) find that abnormal short
interest increases steadily in the
nineteen months before financial misrepresentation is publicly
revealed, suggesting that short
arbitrageurs can detect firms engaging in earnings management.
We reason that when stock
liquidity is high and, consequently, information-based arbitrage
is active it becomes difficult for
4 Interested readers can refer to Bushee and Goodman (2007) and
Edmans (2009) for the list of references.
-
10
managers to move stock prices in their preferred directions at
their desired magnitudes through
earnings management.
Earnings management consumes valuable organizational resources
such as managerial
attention and cognition (Goldman and Slezak 2006; Peng and Rell
2008). Peng and Rell (2008)
argue that the principal cost of earnings management may be not
money spent by the company,
but the waste of managers' limited time that could be spent in
pursuit of long-run value (p.285).
Peng and Rells (2008) argument may be of great relevancy in
todays increasingly brutal
competitive environments wherein managerial attention and
cognition are becoming increasingly
scarce strategic resources (Gavetti 2005; Yadav et al. 2007).
When managers waste valuable
organization resources including their limited time on earnings
management the fundamental
value of their firms would be lower.
Stock liquidity ensures that stock prices faithfully and timely
reflect the fundamental
value of underlying traded stocks as a result of its effect on
the value of information, the
composition of shareholder base, and information-based
arbitrage. For instance, Holmstrm and
Tirole (1993) analytically show that in response to liquidity
improvement market participants
spend more efforts in private information acquisition and
meanwhile trade more aggressively on
their private information to maximize profits, which leads to
more efficient prices about traded
assets (also see Edmans 2009; Grossman and Stiglitz 1980).
Empirical evidence generally
confirms the positive effect of stock liquidity on stock price
efficiency (e.g., Chordia et al. 2008,
2011).5 Therefore, we argue that when stock liquidity is high,
stock prices will timely reflect the
value-destroying consequence of earnings management. We reason
that managers will find it less
beneficial to engage in earnings management when stock liquidity
is high and, consequently,
stock prices timely reflect the value-destroying consequence of
earnings management.
To sum up, stock liquidity makes it difficult for managers to
move stock prices in their
preferred directions at their desired magnitudes through
earnings management by motivating
private information production, shifting shareholder base
towards sophisticated, large investors,
and stimulating informed arbitrage; stock liquidity makes it
less desirable for managers to
5 In this study, we find that firms experiencing dramatic
improvement in stock liquidity resulting from the
decimalization encountered significant enhancement of price
efficiency as gauged by the extent to which their stock
prices deviate from underlying fundamental values.
-
11
engage in earnings management by ensuring that stock prices
timely reflect the value-destroying
consequence of earnings management. Therefore, we reason that
when stock liquidity is high
managers will engage less in earnings management, which leads to
our below hypothesis:
Hypothesis 1 (H1): ceteris paribus, the greater stock liquidity
the less earnings management.
3. Data, Sample, Variable Measurement, and Descriptive
Statistics
3.1 Data and Sample
In the main test of H1, we obtain accounting- and auditor-
related data from
COMPUSTAT, stock-related data from CRSP, and institutional
ownership data from Thomson
CDA/Spectrum Institutional 13f Holdings. To carry out other
tests, we obtain restatement data
from AUDITANALYTICS, AAER data from the Center of Financial
Reporting and
Management, analysts-related data from I/B/E/S, executives- and
compensation- related data
from EXECUCOMP, short interest data from COMPUSTAT, GDP data
from Bureau of
Economic Analysis, and institutional investor classification
data from Dr. Bushees website.
In the main test of H1 we use the absolute value of
discretionary accruals as the proxy for
earnings management. We exclude financial (SIC 6000-6999) and
utilities (SIC 4900-4999)
firms from the sample since discretionary accruals estimation is
problematic for these firms
(DeFond and Subramanyam 1998). To maximize statistical power and
generalizability, we only
require that a firm-year observation has no missing values for
variables used in a test to be
included in the test. As a result, different tests involve
different sample compositions.
3.2 Variable Measurement
3.2.1 Measuring Financial Misreporting
Following existing studies (e.g., Armstrong et al. 2013; Jiang
et al. 2012; Zang 2012) we use
the absolute value of discretionary accruals as the proxy for
earnings management in the main
test of H1. Discretionary accruals are the difference between
total accruals and normal accruals.
We adopt the model proposed in Dechow et al. (1995) to estimate
normal accruals. The model is
as follows:
-
12
where, for firm i and fiscal year t, TAC is the earnings before
extraordinary items and
discontinued operations (COMPUSTAT: ibc) minus the operating
cash flows from continuing
operations taken from the statement of cash flows (COMPUSTAT:
oancf xidoc) (see Hribar
and Collins 2002); A is total assets (COMPUSTAT: at); S is net
sale (COMPUSTAT: sale); REC
is the accounts receivable (COMPUSTAT: rect); PPE is the gross
value of property, plant, and
equipment (COMPUSTAT: ppegt); standards for change from fiscal
year t-1 to fiscal year t.
For each year, we estimate the regression equation (1) for every
industry classified by two-
digit SIC codes. Our estimation approach controls for
industry-wide variations in economic
conditions that affect total accruals while allowing the
coefficients to vary across time. We
require that the minimal number of observations for each
industry-year combination is fifteen.
Discretionary accruals are the estimated residuals of regression
equation (1).
3.2.2 Measuring Stock Liquidity
We adopt the high-low stock liquidity measure proposed in Corwin
and Schultz (2012). The
high-low stock liquidity measure possesses desirable attributes.
First, the high-low stock
liquidity measure has intuitive theoretical foundation. Corwin
and Schultz (2012) base
development of the high-low stock liquidity measure on two
uncontroversial empirical
regularities. Namely, daily high prices are always
buyer-initiated while daily low prices are
always seller-initiated. The ratio of high-to-low prices
reflects both the fundamental volatility
and the bid-ask spread of the stock. The component of the
high-to-low price ratio attributed to
the fundamental volatility increases proportionately with the
trading interval while the
component attributed to the bid-ask spread stays relatively
constant over a short period.
Second, Corwin and Schultz (2012) show that the high-low stock
liquidity measure
outperforms other low-frequency measures in capturing
cross-sections of both spread levels and
month-to-month changes in spreads. When a low-frequency stock
liquidity measure is used in
cross-section regressions, it is desirable for the low-frequency
stock liquidity measure to have
high cross-section correlations with stock liquidity measures
computed from high-frequency
intraday transaction data. In addition, the high-low stock
liquidity measure is much less
computationally demanding than stock liquidity measures
estimated from intraday transaction
data. Because of the large size of samples used in this study,
computational feasibility requires
-
13
us to use low-frequency stock liquidity measures. By
construction, Corwin and Schultzs (2012)
high-low stock liquidity measure actually captures stock
illiquidity. We use the natural log of the
inverse of the high-low estimate as the stock liquidity measure.
In Appendix 2 we provide brief
background for Corwin and Schultzs (2012) high-low stock
liquidity as well as Hasbroucks
(2009) stock liquidity measure.
3.3 Descriptive Statistics
Table 1 reports descriptive statistics for the sample used in
the main test of H1. We refer
to prior studies (e.g., Armstrong et al. 2013; Zang 2012) to set
up the regression model for the
main test of H1. Specifically, we control for auditor
characteristics, various firm characteristics,
and real activities-based earnings management that are either
related to or associated with
earnings management. In addition, we also control for year and
industry fixed effects. Appendix
1 provides definitions for all these control variables. Table 1
Panel A shows that regarding
statistical distributions variables used in the main test of H1
are comparable to those used in prior
studies (e.g., Armstrong et al. 2013; Zang 2012).6
Table 1 Panel B reports Pearson and Spearman correlations
between variables used in the
main test of H1. We are cautious about drawing inferences from
correlations between variables.
However, we want to point out that in line with the prediction
of H1 stock liquidity and absolute
value of discretionary accruals are significantly negatively
correlated. Moreover, the stock
liquidity measure (i.e., LIQ_HL) varies with firm
characteristics as expected. For instance,
LIQ_HL is higher for larger, more mature firms while its lower
for firms with more volatile
business operations as measured by the standard deviations of
cash flows from operations and
sales.
4. Empirical Results
To test H1, we run OLS regression to estimate below model:
6 For the sake of saving space, we only report summary
statistics and correlations for variables used in the main test
of H1. Summary statistics and correlations for variables used in
all other tests are provided at request.
-
14
where ADAt is the absolute value of discretionary accruals as a
percentage of total assets,
estimated by using the model proposed in Dechow et al. (1995);
LIQ_HLt is the natural log of the
inverse of the high-low estimate of bid-ask spread proposed in
Corwin and Schultz (2012),
computed over a period of 252 trading days that in the last
month of fiscal year t. All other
variables are as defined in Appendix 1. H1 predicts that 1 <
0.
Table 2 presents OLS estimates of Equation (2). Standard errors
are adjusted for
heteroscedasticity and clustered at the firm level. Consistent
with the prediction of H1, the
coefficient on LIQ_HLt is negative (1 = -0.293, t = -4.04). The
economic magnitude of the effect
is also significant. We would observe a 6.2% drop in the
absolute value of discretionary accruals
among firms with stock liquidity one standard deviation below
the sample mean if stock liquidity
of these firms is increased to one standard deviation above the
sample mean.7
5. Robustness Tests and Additional Analyses
5.1 Robustness Tests
We run a battery of robust tests of H1. Table 3 reports results
from three of these
robustness tests. Table 3 Panel A reports results from the
robustness test that adopts the stock
liquidity measure proposed in Hasbrouck (2009). Hasbrouck (2009)
shows that when measured
on an annual basis his estimate is highly correlated with the
effective cost measure estimated
from intraday transactions. In Appendix 2, we provide brief
background for Hasbroucks (2009)
stock liquidity measure. Table 3 Panel A shows that our
inference about H1 remains unchanged
when we use the stock liquidity measure proposed in Hasbrouck
(2009).
To ensure that our inference about H1 applies to earnings
management in general and is
robust to specific earnings management measures used, we follow
existing studies to use the
likelihood of a restatement or a SEC AAER as the proxy for
earnings management in robustness
tests. Table 3 Panel B presents results from the robustness test
that examines the relation between
stock liquidity and the likelihood of a SEC Accounting and
Auditing Enforcement Release
(AAER). We refer to Armstrong et al. (2013) and Zang (2012) for
the choice of covariates. All
variables are as defined in Appendix 1. In Table 3 Panel B the
dependent variable is
7 We document a much stronger effect after controlling for more
covariates and firm fixed effects (see Table 4).
-
15
Prob({AAERi,t=1}). AAERi,t is an indicator variable that equals
one if the SEC published a AAER
that identified accounting fraud or misrepresentation at firm i
in fiscal year t and zero if
otherwise. We restrict our analysis to observations with fiscal
years before 2008 to avoid
potential selection bias. Results presented in Table 3 Panel B
are obtained by using Probit
regression. Consistent with the prediction of H1, we find that
the coefficient on LIQ_HLt is
negative (t = -2.35).
Table 3 Panel C presents results from the robustness test that
examines the relation
between stock liquidity and the likelihood of an accounting
restatement. We refer to Armstrong
et al. (2013) and Zang (2012) for the choice of covariates. All
variables are as defined in
Appendix 1. In Table 3 Panel C, the dependent variable is
Prob({RESi,t=1}). RESi,t equals one if
any of firm is financial results (quarterly, annual, or
otherwise) in fiscal year t are subsequently
restated and equals zero if otherwise. There are generally time
lags between restatement dates
and restated financial results. As turned out in the data, more
than 90% of restatements are made
within four years after the financial results. To avoid
potential selection bias, we limit our
analysis to observations with fiscal years before 2009.
Results presented in Table 3 Panel C are obtained by using
Probit regression. As shown in
Model 1, the relation between stock liquidity and the likelihood
of an accounting restatement is
negative but statistically insignificant. In Model 2, we explore
whether there is a nonlinear
relation between stock liquidity and the likelihood of an
accounting restatement. In Model 2,
instead of using a continuous measure of stock liquidity, we
create a dummy variable (i.e.,
D_LIQ_HL) that equals one if LIQ_HL is in the top quintile of
the sample and zero if otherwise.
As shown in Model 2, the coefficient on D_LIQ_HL is negative
(-0.089) and significant (t = -
2.41), suggesting that we would observe significant drop in the
incidence of accounting
restatements when firms move to the top stock liquidity
quintile.
In unreported results, we adopt alternative regression models of
normal accruals proposed in
Dechow et al. (2003) and Jones (1991) to estimate discretionary
accruals and find that our
inference about H1 is robust to the way in which discretionary
accruals are estimated. Moreover,
instead of using two-digit SIC codes to classify the industry
membership of firm-year
-
16
observations we apply the latest Fama-French industry
classification scheme (49 industries in
total) to estimate discretionary accruals. We find that using a
different industry classification
scheme does not alter our inference regarding H1. To sum up,
results from these robustness tests
further corroborate the finding documented in the main test of
H1, suggesting that the negative
relation between stock liquidity and earnings management is not
specific to stock liquidity and
earnings management measures used.
5.2 Endogeneity: Omitted Correlated Variables and Reverse
Causality
Endogeneity problems are ubiquitous in archival studies (Roberts
and Whited 2011).
Endogeneity problems mainly stem from either omitted correlated
variables or reverse causality.
In our research setting, there could be reasons for stock
liquidity and earnings management to be
jointly determined. Also, there might be reverse causality
between stock liquidity and earnings
management. For instance, Lang et al. (2012) find that greater
stock liquidity is associated with
smaller discretionary accruals. Therefore, it is possible that
our finding about the effect of stock
liquidity on earnings management might be driven by potential
reverse causality between stock
liquidity and earnings management.
5.2.1 Controlling for Firm Fixed Effects and More Covariates
To address the potential endogeneity problem stemming from
omitted correlated
variables, we control for firm fixed effects and additional
possible covariates identified from
related studies (e.g., Badertscher 2011; Bergstresser and
Philippon 2006). These additional
covariates are as defined in Appendix 1. Firm fixed effects
methods solve joint determination
problems in which an unobserved time-invariant variable
simultaneously determines both stock
liquidity and earnings management. Controlling for firm fixed
effects is also equivalent to
looking only at the relation between within-firm changes in
stock liquidity and within-firm
changes in earnings management.
Table 4 presents estimates from the tests that controls for firm
fixed effects and additional
possible covariates. There is still evidence of a negative
relation between stock liquidity and the
absolute value of discretionary accruals. The estimate of the
coefficient on LIQ_HL is -0.644
with a significant t-statistic of 2.47.
-
17
The results from the test controlling for firm fixed effects go
a long way toward
dismissing omitted variables explanations as sources of
endogeneity. Because only the effects of
within-rm changes in the absolute value of discretionary
accruals and in stock liquidity are
taken into account, rm-specic omitted variables cannot readily
explain the observed relation
between stock liquidity and the absolute value of discretionary
accruals. On top of the long list of
possible covariates, rm xed effects take care of most
time-invariant unobserved variables. We
conclude that omitted variables are highly unlikely to explain
the negative relation between stock
liquidity and the absolute value of discretionary accruals even
though we could not completely
rule out the possibility.
5.2.2 Difference-in-differences analysis using the 2001
decimalization
To address the potential endogeneity problem stemming from
reverse causality, we use
the difference-in-differences (hereafter, DiD) approach to
determine the effect of a change in
stock liquidity caused by an exogenous event on earnings.
Specifically, this methodology
compares changes in ADA of a sample of treatment firms with
changes in stock liquidity in the
top one-third to changes in ADA of a sample of control firms
with changes in stock liquidity in
the bottom one-third but that are otherwise comparable before
the exogenous event.
The DiD methodology possess several desirable features. First,
the DiD methodology
rules out omitted trends that are correlated with stock
liquidity and ADA in both the treatment
firms and the control firms. Second, the DiD approach helps
establish the direction of causality
as the experiment is conducted surrounding an exogenous change
in stock liquidity. Third, as
with the firm fixed effects the DiD approach controls for
time-invariant unobserved differences
between the treatment firms and the control firms.
Following prior studies (e.g., Fang et al. Forthcoming) we use
the 2001 decimalization of
the minimum tick size as the opportunity for quasi experiment.
Prior studies document
significant improvement in stock liquidity as a result of the
decimalization (Bessembinder 2003;
Chordia et al. 2005). For the empirical relation tested in this
study, the decimalization is a good
candidate for quasi experiment because decimalization directly
affects stock liquidity but
unlikely affect earnings management directly. At the same time,
changes in stock liquidity
surrounding the decimalization exhibit wide variation in the
cross-section of stocks. More
-
18
importantly, we would not expect the change in future earnings
management to affect the change
in stock liquidity brought about by the decimalization. Hence,
an examination of the change in
earnings management following the change in stock liquidity
resulting from the decimalization
provides a quasi- experiment for our test.
We construct a treatment group and a control group of firms
using propensity score
matching. Specifically, we begin with all firms with non-missing
matching variables including
ADA in the pre-decimalization year (t-1) and the
post-decimalization year (t+1), with t indicating
the year during which the decimalization occurred. On the basis
of LIQ_HLt-1 to t+1, we sort
2,576 sample firms into three equal groups and retain only the
top group representing firms that
experience the greatest increase in stock liquidity surrounding
the decimalization (namely the
treatment group) and the bottom group representing firms that
experience the least improvement
in stock liquidity surrounding the decimalization (namely the
control group).
To apply the propensity score matching, we first estimate a
Probit model based on the
1,717 sample firms in the top and the bottom groups. The
dependent variable is one if the firm-
year belongs to the treatment group and zero otherwise. The
Probit model includes all control
variables from the main test of H1 and ADA measured in the year
immediately preceding the
decimalization. We also control for industry fixed effects in
the Probit model. These variables
are included to help satisfy the parallel trends assumption as
the DiD estimator should not be
driven by the differences in any industry or firm
characteristics. Table 5 Panel A presents
parameter estimates from the Probit model used to estimate the
propensity scores. The results
show that the specification captures a significant amount of
variation in the choice variable, as
indicated by a pseudo R2 of 19.5% and a p-value from the 2 test
of the overall model fitness well
below 0.001.
We then use the predicted probabilities (i.e., propensity
scores) to perform a nearest-
neighbor propensity score matching. Specifically, each firm in
the top group (i.e., the treatment
firms) is matched to a firm from the bottom group with the
closest propensity score (i.e., the
control firms) without replacement. We end up with 473 unique
pairs of matched firms.
-
19
The validity of the DiD estimate critically relies on the
parallel trends assumption.
Following prior studies (e.g., Fang et al. Forthcoming) we
conduct a number of diagnostic tests
to demonstrate that we do not violate the parallel trends
assumption. In the first test, we rerun the
Probit model by using the matched sample. The Probit estimates
are presented in the Post-
match column of Table 5 Panel A. None of the coefficients on
independent variables is
statistically significant including the coefficient on
pre-decimalization ADA. Also, the Post-
match coefficient estimates are much smaller than the Pre-match
ones, suggesting that the
Post-match results are not simply an artifact of a decline in
the degree of freedom due to the
drop in sample size. In addition, pseudo R2 drops drastically
from 19.5% prior to the matching to
1.1% post the matching. And a 2 test for the overall model
fitness shows that we cannot reject
the null hypothesis that all of the coefficient estimates are
zero (with a p-value of 0.9998).
In the second diagnostic test, we examine the difference between
the propensity scores of
the treatment firms and those of their matched control firms.
Table 5 Panel B demonstrates that
the difference is rather trivial. Finally, we report the
univariate comparisons between the
treatment firms and the control firms with respect to their
pre-decimalization characteristics and
corresponding t-statistics in Panel C. As shown, we observe no
statistically significant
differences between the treatment firms and the control firms
with respect to their characteristics
in the pre-decimalization regime. In particular, the two groups
of firms have similar levels of
pre-decimalization stock liquidity (i.e., LIQ_HL) and ADA, even
though their stock liquidity is
affected by the decimalization differently. Overall, our
diagnostic tests suggest that the
propensity score matching process has removed material
observable differences (other than the
difference in the change in stock liquidity surrounding the
decimalization), which increases the
likelihood that changes in ADA are caused only by the exogenous
change in stock liquidity
resulting from the decimalization.
Table 5 Panel D presents the DiD estimator. Column (1) reports
the average change in
ADA for the control firms and Column (2) reports the average
change in ADA for the treatment
firms. Change in ADA (i.e., CHG_ADAt-1 to t+1) is the difference
between ADA immediately
before the decimalization and ADA immediately after the
decimalization. In columns (3) and (4),
we report the DiD estimator and the corresponding two-tailed
t-statistics.
-
20
Several interesting findings emerge. First, ADA of both the
treatment firms and the
control firms drops after the decimalization, which is
consistent with H1 that stock liquidity is
negatively related to earnings management on average. Second,
the drop in ADA is much greater
for the treatment firms than for the control firms as the DiD
estimator is statistically significant at
the 1% level. Actually the drop in ADA is statistically
significant only for the treatment group.
Following Roberts and Whiteds (2011) suggestion, we repeat the
DiD analysis in pre-
decimalization years to further verify the internal validity of
our DiD finding. Specifically, we
examine whether significant difference exists between the
treatment firms and the control firms
regarding change in ADA from year t-2 to year t-1. Table 5 Panel
E presents results from this
falsification test. As shown, we observe no significant change
in ADA for both the treatment
firms and the control firms and no significant difference
between the treatment firms and the
control firms regarding change in ADA. Findings from this
falsification test further suggest that
our DiD finding is more likely due to the differential impact of
the decimalization on the stock
liquidity of the treatment firms and the control firms, as
opposed to some alternative forces.
5.3 Possible Mechanisms
We run several tests to examine whether the hypothesized
mechanisms through which
stock liquidity deters managers from engaging in earnings
management change surrounding the
decimalization as predicted. It is definitely challenging to
provide definitive proof of underlying
mechanisms through which stock liquidity deters earnings
management. Therefore, our tests here
are only suggestive.
5.3.1 Stock Liquidity and Private Information Production
We reason that one of the mechanisms through which stock
liquidity deters earnings
management is that stock liquidity motivates market participants
to engage in information
production by enhancing the value of information. Essentially,
we cannot observe the overall
production of information about a firm by all market
participants. However, we can infer efforts
expended by financial analysts in producing information about a
firm (Barth et al. 2001).
Financial analysts serve as critical information intermediaries
in the capital market through their
private acquisition and processing of information about firms
economic prospects. We argue
-
21
that the amounts of information production efforts expended by
financial analysts might serve as
a rough indicator of overall information production efforts made
by market participants.
Following Barth et al. (2001) we gauge financial analysts
information production efforts
by the number of analysts following a firm (i.e., ANA_COV) and
analysts efforts (i.e.,
ANA_EFF) where ANA_EFF is defined as
, Ni,t : number of analysts following firm i
in year t and nj: number of firms followed by analyst j in year
t. We then examine the effect of
stock liquidity on financial analysts information production
efforts as measured by ANA_COV
and ANA_EFF in the DiD framework using the matched sample
constructed in Section 4.2.2.
Table 6 Panel A presents the DiD estimator. The treatment firms
experience no
significant change in analyst followings while the control firms
experience significant drop in
analyst followings. And the changes in analyst followings are
statistically significantly different
between the control firms and the treatment firms. Regarding
ANA_EFF, the treatment firms
experienced statistically significant increase after the
decimalization while the control firms
experienced no material change in ANA_EFF. And changes in
ANA_EFF are marginally
significantly different between the control firms and the
treatment firms. Overall, our evidence
suggests that information production by financial analysts in
particular (and by market
participants in general) in response to change in stock
liquidity may be an underlying mechanism
through which stock liquidity deters earnings management.
5.3.2 Stock Liquidity and Institutional Ownership
We reason that another mechanism through which stock liquidity
deters earnings
management is that stock liquidity encourages the formation of
blockholders and thus shifts the
shareholder base towards sophisticated, large institutional
investors. To examine the effect of
stock liquidity on the shareholder base composition, we compute
changes in institutional
ownership (i.e., INST_OWN) in the DiD framework using the
matched sample constructed in
Section 4.2.2. Table 6 Panel B presents the DiD estimator. As
shown, the treatment firms
experience, both statistically and economically, significant
increase in institutional ownership
after the decimalization while the control firms experience no
material change in institutional
-
22
ownership. Regarding the change in institutional ownership the
difference between the treatment
firms and the control firms is statistically significant.
Institutional investors are not homogenous and differ in various
dimensions such as
investment horizon (Bushee 1998; Yan and Zhang 2009). We follow
the institutional investor
classification scheme created by Bushee (1998, 2001) and explore
changes in institutional
holdings owned by transient investors (i.e., INST_OWN_TRA),
quasi-indexers (i.e.,
INST_OWN_QIX), and dedicated investors (i.e., INST_OWN_DED),
respectively. We argue that
these three types of investors have differential impacts on
managers incentives to engage in
earnings management.
Findings of prior studies (e.g., Yan and Zhang 2009) suggest
that transient institutional
investors are adept at private information acquisition and, as a
result, are well informed.
Therefore, we argue that increase in institutional holdings
owned by transient investors (i.e.,
INST_OWN_TRA) makes it more difficult for managers to move stock
prices through earnings
management.
Findings of prior studies (e.g., Chen et al. 2007) suggest that
dedicated institutional
shareholders tend to actively monitor managers for wrongdoings
due to their dedicated
ownership position. Earnings management is value-destroying
because it consumes valuable
organizational resources, especially managers limited time.
Also, dedicated institutional
shareholders would not benefit much from temporary changes in
stock prices driven by earnings
management as a result of their dedicated ownership position.
Moreover, Karpoff et al. (2008)
find that once detected accounting frauds can impose substantial
penalties on firms. We argue
that dedicated institutional shareholders would bear a big
portion of penalties imposed on firms
in the form of dramatic decline in the value of their
shareholdings once accounting frauds are
detected. Therefore, it is expected that due to their dedicated
ownership position dedicated
institutional shareholders have greater incentives to monitor
managers and prevent earnings
management in the first place than the other two types of
institutional shareholders. Therefore, it
is reasonable to posit that increase in institutional holdings
owned by dedicated investors (i.e.,
INST_OWN_DED) leads to less earnings management.
-
23
Quasi-indexing institutional investors use passive indexing
strategies and thus have few
or no incentives to monitor specific invested firms or spend
resources in acquiring information
about specific invested firms. Therefore, changes in
institutional holdings owned by quasi-
indexers (i.e., INST_OWN_QIX) may not have direct impact on
earnings management. However,
we argue that decrease in institutional holdings owned by
quasi-indexers may increase
institutional holdings owned by transient investors and
dedicated investors as a result of the
crowding-out effect and, therefore, indirectly leads to less
earnings management.
Table 6 Panel B also presents the DiD estimators for
institutional holdings owned by
transient investors, quasi-indexers, and dedicated investors.
Both the treatment firms and the
control firms experience significant increase in institutional
holdings owned by transient
investors after the decimalization while the increase is
significantly greater for the treatment
firms than for the control firms. As shown, only the treatment
firms experience significant
increase in institutional holdings owned by dedicated investors.
Regarding the change in
institutional holdings owned by dedicated investors, the
difference between the treatment firms
and the control firms is statistically significant. Both the
treatment firms and the control firms
experience significant decrease in institutional holdings owned
by quasi-indexing investors after
the decimalization, and the decreases experienced by the
treatment firms and the control firms
are not statistically significant. Interestingly, Table 6 Panel
B reveals that for the control firms
decrease in institutional holdings owned by quasi-indexing
investors cancels out the increase in
institutional holdings owned by transient investors, which
accounts for why we observe no
significant change in overall institutional ownership after the
decimalization for the control firms.
In summary, we show that after the decimalization firms with
greater improvement in
stock liquidity experience greater increase in institutional
holdings owned by transient investors
that are well informed and by dedicated investors that have
great incentives to deter managers
from engaging in earnings management. Our evidence suggests that
changes in institutional
holdings owned by different types of institutional investors in
response to changes in stock
liquidity may explain why firms with greater improvement in
stock liquidity after the
decimalization experience greater decrease in the absolute value
of discretionary accruals.
5.3.3 Stock Liquidity and Arbitrage
-
24
Stock liquidity stimulates arbitrage by reducing costs and risks
faced by arbitrageurs. We
argue that as a result of its effect on arbitrage stock
liquidity discourages earnings management.
If managers succeeded in maneuvering market participants
expectations about their firms
economic prospects through earnings management,
incoming-increasing discretionary accruals
arguably would cause overvaluation while income-decreasing
discretionary accruals arguably
would cause undervaluation. To arbitrage on undervalued stocks
requires taking a long position
while arbitrage on overvalued stocks requires taking a short
position. In practice, taking a short
position is costlier and riskier than taking a long position
(see Hirshleifer et al. 2011). Therefore,
short arbitrageurs arguably may have greater incentives to get
informed as a result of great risk
and cost to be overcome by them (Boehmer et al. 2008). Here, we
examine the effect of stock
liquidity on short interests in the DiD framework using the
matched sample constructed in
Section 4.2.2.
Table 6 Panel C reports the DiD estimator. As shown, on average,
both the treatment
firms and the control firms experience significant increases in
short interests after the
decimalization. However, regarding the increases in short
interests the difference between the
treatment firms and the control firms is not statistically
significant. We argue that it is not
surprising to find no significant difference between the
treatment firms and the control firms
because stock liquidity prevents overvaluation in the first
place as a result of its effect on
earnings management and, as a result, leaves fewer opportunities
for short arbitrage. For example,
the threat of ex-post active short arbitrage deters ex-ante
income-increasing earnings
management in the first place. It may be revealing to examine
the cross-section relation between
stock liquidity and short interests.
Table 7 reports results from the Tobit regression that takes
short interests as a function of
stock liquidity and other covariates. We refer to Dechow et al.
(2001) to set up the regression
model. As shown, in the cross-section stock liquidity is
positively related to short interests. In
summary, our evidence is generally consistent with a story that
stock liquidity discourages
earnings management in the first place as a result of its
positive effect on arbitrage.
5.3.4 Stock Liquidity and Stock Price Efficiency
-
25
We reason that stock liquidity and ensuing stock price
efficiency deter earnings
management by ensuring that stock prices could timely reflect
the value-destroying consequence
of earnings management. To examine the effect of stock liquidity
on stock price efficiency, we
investigate changes in stock price efficiency in the DiD
framework using the matched sample
constructed in Section 4.2.2. We measure stock price
inefficiency essentially as the extent to
which stock prices deviate from the fundamental value of traded
stocks. Specifically, we use two
measures: the absolute value of (MV / FV minus one) (i.e.,
INEFF_FL) where MV is the market
value of equity measured at the fiscal year end and FV is the
fundamental value of equity
estimated by using the analyst-based valuation method proposed
in Frankel and Lee (1998), and
the absolute value of the difference between LMV and LFV (i.e.,
INEFF_RKRV) where LMV is
the natural log of market value of equity measured at the fiscal
year end and LFV is natural log
of the fundamental value of equity estimated by using the
accounting-based valuation method
proposed in Rhodes-Kropf et al. (2005).
Table 6 Panel D reports the DiD estimators. Using INEFF_RKRV as
the measure for
stock price inefficiency, we find that the treatment firms
experience significant improvement in
stock price efficiency after the decimalization while the
control firms experience significant
deterioration in sock price efficiency, and that the difference
between the treatment firms and the
control firms is statistically significant regarding the change
in INEFF_RKRV. Using INEFF_FL
as the measure for stock price inefficiency, we find that only
the treatment firms experience
marginal improvement in stock price efficiency after the
decimalization, and that the difference
between the treatment firms and the control firms is
statistically insignificant regarding the
change in INEFF_FL. We conjecture that small sample size and
thus low statistical power may
explain why we cannot find strong evidence using INEFF_FL. Only
60 pairs out of 473 pairs
have no missing values for INEFF_FL.8 Overall, our finding
suggests that the effect of stock
liquidity on stock price efficiency may serve as a mechanism
through which stock liquidity
affects earnings management.
8 Presuming no change in stock price efficiency for firms that
miss values for INEFF_FL, we rerun the DiD test. We
find that change in INEFF_FL is -0.167 (t = -2.87, p = -0.004)
for the treatment firms and is -0.025 (t = -0.82, p =
0.415) for the control firms, and that regarding the change in
INEFF_FL the difference between the treatment firms
and the control firms is 0.143 (t = 2.17, p = 0.030).
-
26
5.4 Stock Liquidity and the Relation between Discretionary
Accruals and Equity Valuation
Errors
We reason that stock liquidity makes it difficult for managers
to move stock prices in
their preferred directions at their preferred magnitudes through
earnings management as a result
of its effect on information production, shareholder base
composition, and arbitrage. The direct
empirical implication of our reasoning is that we should observe
weaker relation between
discretionary accruals and equity valuation errors when stock
liquidity is higher. We argue that
finding evidence consistent with the implication will lend
further support to the empirical
validity of our reasoning.
It is challenging to estimate the fundamental value of traded
stocks. To ensure that our
finding is not sensitive to specific measures, we use two
different measures. The first measure is
MV / FV minus one (i.e., EQ_VAL_ERR_FL) where MV is the market
value of equity measured
at the end of the fourth month after fiscal year end and FV is
the fundamental value of equity
estimated by using the analyst-based valuation method proposed
in Frankel and Lee (1998). The
second measure is the difference between LMV and LFV (i.e.,
EQ_VAL_ERR_RKRV ) where
LMV is the natural log of the market value of equity measured at
the end of the fourth month
after fiscal year end and LFV is the natural log of the
fundamental value of equity estimated by
using the accounting-based valuation method proposed in
Rhodes-Kropf et al. (2005).
Table 8 reports results for the analysis that examines the
effect of stock liquidity on the
relation between discretionary accruals and equity valuation
errors. Referring to previous studies
(i.e., Jiang et al. 2005; Zhang 2006) we control for various
firm characteristics that are related to
or associated with valuation uncertainty faced by investors. As
shown, regardless of the
measures used, we observe significant positive relation between
discretionary accruals (i.e., DA)
and equity valuation errors for firms with their stock liquidity
below the sample median while we
find no significant relation between discretionary accruals
(i.e., DA) and equity valuation errors
for firms with their stock liquidity above the sample median.
Our finding lends further support to
the empirical validity of our reasoning that stock liquidity
makes it difficult for managers to
-
27
move stock prices in their preferred directions at their
preferred magnitudes through earnings
management.
5.5 CEOs Equity Incentive and the Relation between Stock
Liquidity and Earnings
Management
We reason that one mechanism through which stock liquidity
deters earnings
management is that stock liquidity causes stock prices to timely
and faithfully reflect the value-
destroying consequence of earnings management. The direct
implication of our reasoning is that
the negative relation between stock liquidity and earnings
management will be stronger when
managers have a greater stake in their firms stock prices.
Following existing studies, we examine the effect of CEOs stake
in their firms stock
prices on the relation between stock liquidity and earnings
management. Specifically, we
measure a CEOs stake in his firms stock price as the sensitivity
of the CEOs equity portfolio
(i.e., stocks, restricted stocks, and unexercised exercisable /
un-exercisable stock options) to a 1%
change in the stock price (see Core and Guay, 2002). Following
Bergstresser and Philippon
(2006), we use the ratio of the CEOs equity portfolio delta to
the sum of the CEOs equity
portfolio delta and the CEOs total cash compensation measured at
the end of fiscal year t-1 (i.e.,
EQ_INC_CEO) to ensure comparability across periods and firms.9
We then refer to the sample
median to sort the observations into two groups and run the OLS
regression separately for each
group.
Table 9 reports the results. As shown, consistent with the
implication of our reasoning the
negative relation between stock liquidity and the absolute value
of discretionary accruals is
stronger when CEOs have greater stake in their firms stock
prices as measured by
EQ_INC_CEO.10
We conclude that our finding here lends further support to the
empirical
validity of our reasoning.
5.6 Trends of Stock Liquidity and Earnings Management
9 We obtain qualitatively the same results by scaling the CEOs
equity portfolio delta by the CEOs total annual
compensation. 10
We adopt the sample used in the test for results reported in
Table 4 to run a similar test that controls for more
covariates and firm fixed effects, and obtain essentially the
same results.
-
28
Stock liquidity varies over time (see Chordia et al. 2008).
Given the strong cross-section
evidence about the negative relation between stock liquidity and
earnings management, we
wonder whether earnings management and stock liquidity
negatively co-vary over time in the
aggregate. We draw Figure 1 to examine the time-series patterns
of aggregate stock liquidity and
the absolute value of discretionary accruals. To draw Figure 1
we just require that a firm-year
observation has no missing values for stock liquidity (i.e.,
LIQ_HL) and the absolute value of
discretionary accruals (i.e., ADA). We first assign percentage
ranks to each firm-year observation
according to its stock liquidity and absolute value of
discretionary accruals. Then we compute
equally-weighted percentage ranks for stock liquidity and the
absolute value of discretionary
accruals each year.
Figure 1A reveals that consistent with the findings documented
in Cohen et al. (2008)
there is an overall trend of increase in the absolute value of
discretionary accruals during 1990-
2000 while there is an overall trend of decrease during
2001-2005. Figure 1A also reveals that
consistent with our cross-section evidence the absolute value of
discretionary accruals and stock
liquidity negatively co-vary over time in the aggregate. More
importantly, variation in aggregate
stock liquidity well accounts for variation in aggregate
absolute value of discretionary accruals
( = -0.74). In a period of financial turmoil when investors face
great uncertainty, investors may
not trade for information reasons such as flight-to-quality on
the one hand, and on the other
hand stock liquidity may dry out (Caballero and Krishnamurthy
2008; Ns et al. 2011). We
conjecture that if excluding 1999-2001 and 2008 2009 when huge
uncertainty permeates the
stock market we should observe a stronger relation between stock
liquidity and the absolute
value of discretionary accruals in the aggregate. Figure 1B
shows that consistent with our
conjecture variation in aggregate stock liquidity accounts for
more variation in aggregate
absolute value of discretionary accruals ( = -0.81) when
1999-2001 and 2008 2009 are
excluded.
Cohen et al. (2008) attribute the decline in aggregate absolute
value of discretionary
accruals during 2002-2005 to the passage of SOX. However, Figure
1A show that during 2000
2005 (six years in total) decline in aggregate absolute value of
discretionary accruals occurred
-
29
simultaneously with improvement in aggregate stock liquidity ( =
-0.94).11 In summary, our
findings suggest that besides the passage of SOX improvement in
stock liquidity and ensuing
changes in information production activities, shareholder base
compositions, and arbitrage
activities may also drive the decline in aggregate absolute
value of discretionary accruals during
2001-2005.
6. Conclusion
This study examines the effect of stock liquidity on earnings
management. We find a
negative relation between stock liquidity and earnings
management. Our finding is insensitive to
specific stock liquidity and earnings management measures used,
and remains strong after
controlling for a long list of possible covariates including
firm fixed effects. We use the
decimalization of the minimum tick size as a quasi-experiment to
carry out a difference-in-
differences (DiD) analysis, and find that firms with greater
stock liquidity improvement (the
treatment firms) experience a significant drop in the absolute
value of discretionary accruals
while firms with less stock liquidity improvement (i.e., the
control firms) experience no
significant drop. However, before the decimalization the
treatment firms and the control firms
are similar along a host of characteristics. Our results from
the DiD analysis suggest that the
negative effect of stock liquidity on earnings management is
causal.
Our finding has important policy implications since stock
liquidity can be altered by
policies and regulations. Our study suggests an unexplored
channel through which stock liquidity
contributes to firm value and social welfare. That is, at the
firm level stock liquidity contributes
to firm value by deterring earnings management that consumes
valuable organizational resources
such as top executives limited attention and cognition that
could be used in pursuit of long-run
value; at the society level, as a result of its effect on
earnings management, stock liquidity
enhances trust between managers and investors and, therefore,
contributes to the functioning of
the capital market.
11
Figure 1A also shows that aggregate absolute value of
discretionary accruals actually started to decline in 2001
(one year before the passage of SOX) when market-wide stock
liquidity starts to improve dramatically (see Chordia
et al. 2011).
-
30
References
Armstrong, C. S., D. F. Larcker, G. Ormazabal, and D. J. Taylor.
2013. The relation between
equity incentives and misreporting: The role of risk-taking
incentives. Journal of Financial
Economics 109: 327-50.
Badertscher, B. A. 2011. Overvaluation and the choice of
alternative earnings management
mechanisms. The Accounting Review 86: 1491-518.
Barth, M. E., J. A. Elliott, and M. W. Finn. 1999. Market
rewards associated with patterns of
increasing earnings. Journal of Accounting Research 37:
387-413.
Barth, M. E., R. Kasznik, and M. F. McNichols. 2001. Analyst
coverage and intangible assets.
Journal of Accounting Research 39: 1-34.
Bartov, E., D. Givoly, and C. Hayn. 2002. The rewards to meeting
or beating earnings
expectations. Journal of Accounting and Economics 33:
173-204.
Bergstresser, D., and T. Philippon. 2006. CEO incentives and
earnings management. Journal of
Financial Economics 80: 511-29.
Bessembinder, H. 2003. Trade execution costs and market quality
after decimalization. Journal
of Financial and Quantitative Analysis 38: 747-77.
Bharath, S. T., S. Jayaraman, and V. Nagar. 2013. Exit as
governance: An empirical analysis.
The Journal of Finance 68: 2515-47.
Boehmer, E., C. M. Jones, and X. Zhang. 2008. Which shorts are
informed? The Journal of
Finance 63: 491-527.
Boehmer, E., and E. K. Kelley. 2009. Institutional investors and
the informational efficiency of
prices. Review of Financial Studies 22: 3563-94.
Bond, P., A. Edmans, and I. Goldstein. 2012. The real effects of
financial markets. Annual
Review of Financial Economics 4: 339-60.
Brav, A., W. Jiang, and H. Kim. 2010. Hedge fund activism: A
review. Foundations and Trends
in Finance 4: 1-66.
Bushee, B. J. 1998. Investors on myopic R&D investment
behavior. The Accounting Review.
Bushee, B. J., and T. H. Goodman. 2007. Which Institutional
investors trade based on private
information about earnings and returns? Journal of Accounting
Research 45: 289-321.
Chen, X., J. Harford, and K. Li. 2007. Monitoring: Which
institutions matter? Journal of
Financial Economics 86: 279-305.
Chordia, T., R. Roll, and A. Subrahmanyam. 2008. Liquidity and
market efficiency. Journal of
Financial Economics 87: 249-68.
. 2011. Recent trends in trading activity and market quality.
Journal of Financial Economics 101: 243-63.
Chordia, T., A. Sarkar, and A. Subrahmanyam. 2005. An empirical
analysis of stock and bond
market liquidity. Review of Financial Studies 18: 85-129.
-
31
Cohen, D. A., A. Dey, and T. Z. Lys. 2008. Real and
accrual-based earnings management in the
pre- and post-Sarbanes-Oxley periods. Accounting Review 83:
757-87.
Collins, D. W., G. Gong, and P. Hribar. 2003. Investor
sophistication and the mispricing of
accruals. Review of Accounting Studies 8: 251-76.
Core, J., and W. Guay. 2002. Estimating the value of employee
stock option portfolios and their
sensitivities to price and volatility. Journal of Accounting
Research 40: 613-30.
Corwin, S. A., and P. Schultz. 2012. A simple way to estimate
bidask spreads from daily high and low prices. The Journal of
Finance 67: 719-60.
Dechow, P. M., W. Ge, C. R. Larson, and R. G. Sloan. 2011.
Predicting material accounting
misstatements. Contemporary Accounting Research 28: 17-82.
Dechow, P. M., A. P. Hutton, L. Meulbroek, and R. G. Sloan.
2001. Short-sellers, fundamental
analysis, and stock returns. Journal of Financial Economics 61:
77-106.
Dechow, P. M., S. A. Richardson, and I. Tuna. 2003. Why are
earnings kinky? An examination
of the earnings management explanation. Review of Accounting
Studies 8: 355-84.
Dechow, P. M., R. G. Sloan, and A. P. Sweeney. 1995. Detecting
earnings management. The
Accounting Review 70: 193-225.
DeFond, M. L., and K. Subramanyam. 1998. Auditor changes and
discretionary accruals. Journal
of Accounting and Economics 25: 35-67.
Edmans, A. 2009. Blockholder trading, market efficiency, and
managerial myopia. The Journal
of Finance 64: 2481-513.
Edmans, A., V. Fang, and E. Zur. 2013. The effect of liquidity
on governance. Review of
Financial Studies 26: 1443-82.
Fama, E. F. 1980. Agency problems and the theory of the firm.
Journal of Political Economy 88:
288-307.
Fang, V., X. Tian, and S. Tice. Forthcoming. Does stock
liquidity enhance or impede firm
innovation? Journal of Finance.
Fang, V. W., T. H. Noe, and S. Tice. 2009. Stock market
liquidity and firm value. Journal of
Financial Economics 94: 150-69.
Fields, T. D., T. Z. Lys, and L. Vincent. 2001. Empirical
research on accounting choice. Journal
of Accounting and Economics 31: 255-307.
Fishman, M. J., and K. M. Hagerty. 1989. Disclosure decisions by
firms and the competition for
price efficiency. The Journal of Finance 44: 633-46.
Frankel, R., and C. Lee. 1998. Accounting valuation, market
expectation, and cross-sectional
stock returns. Journal of Accounting and Economics 25:
283-319.
Gavetti, G. 2005. Cognition and hierarchy: Rethinking the
microfoundations of capabilities development. Organization Science
16: 599-617.
Goldman, E., and S. L. Slezak. 2006. An equilibrium model of
incentive contracts in the
presence of information manipulation. Journal of Financial
Economics 80: 603-26.
-
32
Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. The economic
implications of corporate
financial reporting. Journal of Accounting and Economics 40:
3-73.
Grossman, S. J., and J. E. Stiglitz. 1980. On the impossibility
of informationally efficient
markets. The American Economic Review 70: 393-408.
Hasbrouck, J. 2009. Trading costs and returns for U.S. equities:
Estimating effective costs from
daily data. The Journal of Finance 64: 1445-77.
Healy, P. M., and K. G. Palepu. 2003. The fall of Enron. The
Journal of Economic Perspectives
17: 3-26.
Healy, P. M., and J. M. Wahlen. 1999. A review of the earnings
management literature and its
implications for standard setting. Accounting Horizons 13:
365-83.
Hirshleifer, D., S. H. Teoh, and J. J. Yu. 2011. Short
arbitrage, return asymmetry, and the accrual
anomaly. Review of Financial Studies 24: 2429-61.
Holmstrm, B., and J. Tirole. 1993. Market liquidity and
performance monitoring. Journal of
Political Economy 101: 678-709.
Hribar, P., and D. W. Collins. 2002. Errors in estimating
accruals: Implications for empirical
research. Journal of Accounting Research 40: 105-34.
Jensen, M. C. 2005. Agency costs of overvalued equity. Financial
Management 34: 5-19.
Jiang, G., C. M. Lee, and Y. Zhang. 2005. Information
uncertainty and expected returns. Review
of Accounting Studies 10: 185-221.
Jiang, J., K. R. Petroni, and I. Y. Wang. 2010. CFOs and CEOs:
Who have the most influence on
earnings management? Journal of Financial Economics 96:
513-26.
Jones, J. J. 1991. Earnings management during import relief
investigations. Journal of
Accounting Research 29: 193-228.
Karpoff, J. M., D. S. Lee, and G. S. Martin. 2008. The cost to
firms of cooking the books.
Journal of Financial and Quantitative Analysis 43: 581-612.
Karpoff, J. M., and X. Lou. 2010. Short sellers and financial
misconduct. The Journal of Finance
65: 1879-913.
Kasznik, R., and M. F. McNichols. 2002. Does meeting earnings
expectations matter? Evidence
from analyst forecast revisions and share prices. Journal of
Accounting Research 40: 727-59.
Kyle, A. S., and J.-L. Vila. 1991. Noise trading and takeovers.
The RAND Journal of Economics
22: 54-71.
Lang, M., K. V. Lins, and M. Maffett. 2012. Transparency,
liquidity, and valuation: International
evidence on when transparency matters most. Journal of
Accounting Research 50: 729-74.
Lee, C. 2001. Market efficiency and accounting research: a
discussion of capital market research in accounting by SP Kothari.
Journal of Accounting and Economics 31: 233-53.
Maug, E. 1998. Large shareholders as monitors: Is there a
trade-off between liquidity and control?
The Journal of Finance 53: 65-98.
-
33
Ns, R., J. A. Skjeltorp, and B. A. degaard. 2011. Stock market
liquidity and the business
cycle. The Journal of Finance 66: 139-76.
O'Hara, M. 2003. Presidential address: Liquidity and price
discovery. The Journal of Finance 58:
1335-54.
Peng, L., and A. Rell. 2008. Manipulation and equity-based
compensation. The American
Economic Review 98: 285-90.
RhodesKropf, M., D. T. Robinson, and S. Viswanathan. 2005.
Valuation waves and merger activity: The empirical evidence.
Journal of Financial Economics 77: 561-603.
Roberts, M. R., and T. Whited. 2011. Endogeneity in empirical
corporate finance. In Handbook
of the Economics of Finance. Eds. G. Constantinides, Rene Stulz
and M. Harris. Amsterdam:
Elsevier.
Roll, R. 1984. A simple implicit measure of the effective bidask
spread in an efficient market. The Journal of Finance 39:
1127-39.
Shleifer, A., and R. W. Vishny. 1997. The limits of arbitrage.
The Journal of Finance 52: 35-55.
Skinner, D. J., and R. G. Sloan. 2002. Earnings surprises,
growth expectations, and stock returns
or don't let an earnings torpedo sink your portfolio. Review of
Accounting Studies 7: 289-312.
Yadav, M. S., J. C. Prabhu, and R. K. Chandy. 2007. Managing the
future: CEO attention and
innovation outcomes. Journal of Marketing 84-101.
Yan, X. S., and Z. Zhang. 2009. Institutional investors and
equity returns: Are short-term
institutions better informed? Review of Financial Studies 22:
893-924.
Zang, A. Y. 2012. Evidence on the trade-off between real
activities manipulation and accrual-
based earnings management. The Accounting Review 87:
675-703.
Zhang, X. 2006. Information uncertainty and stock returns. The
Journal of Finance 61: 105-37.
-
34
TABLE 1
Descriptive statistics
Table 1 reports descriptive statistics for the sample used in
the main test of H1. The sample is
constructed from the intersection of COMPUSTAT (accounting and
auditor data), CRSP (stock
price), and Thomson CDA/Spectrum Institutional 13f Holdings
(institutional ownership) for the
period of 1988 2012. The sample covers a total of 52,516
firm-year observations. Panel A presents summary statistics. Panel
B presents Pearson and Spearman correlations. All variables
are as defined in Appendix 1.
Panel A: Summary statistics
Variable N Mean Std 25th Median 75th
ADA 52,516 7.16 7.66 2.14 4.80 9.29
LIQ_HL 52,516 4.34 0.76 3.88 4.41 4.88
Big8 52,516 0.86 0.35 1.00 1.00 1.00
AUD_TEN 52,516 9.73 7.67 4.00 7.00 14.00
AUD_TEN x AUD_TEN 52,516 153.57 235.13 16.00 49.00 196.00
NOA 52,516 0.80 0.89 0.34 0.55 0.89
OP_Cycle 52,516 141.39 91.90 79.96 123.04 179.58
Market_Share 52,516 0.06 0.15 0.00 0.01 0.04
Z-Score 52,516 5.35 5.70 2.30 3.59 5.95
INST_OWN 52,516 0.46 0.30 0.19 0.45 0.71
REM 52,516 -0.09 0.25 -0.21 -0.09 0.02
LMV 52,516 5.56 2.11 4.04 5.50 6.95
BTM 52,516 0.70 0.63 0.31 0.53 0.87
LEV 52,516 0.19 0.18 0.02 0.16 0.31
Firm_Age 52,516 19.78 16.29 8.00 14.00 27.00
ROA 52,516 0.01 0.16 -0.01 0.04 0.08
RET 52,516 0.16 0.67 -0.24 0.05 0.38
Capital 52,516 0.26 0.21 0.10 0.20 0.37
Intangible 52,516 0.08 0.26 0.00 0.02 0.08
Std_CashFlow 52,516 0.08 0.08 0.03 0.05 0.09
Std_Sale 52,516 0.22 0.23 0.08 0.15 0.28
-
35
Panel B: Pairwise Pearson (Spearman) correlations in upper
(lower) triangle
Correlations significantly different from zero at p-values less
than 0.05 are in boldface type
Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
(14) (15) (16) (17) (18) (19) (20) (21)
ADA (1) -0.23 -0.11 -0.15 -0.13 0.03 0.09 -0.11 0.10 -0.21 0.14
-0.21 -0.04 -0.10 -0.19 -0.30 0.00 -0.16 0.12 0.26 0.19
LIQ_HL (2) -0.24 0.19 0.30 0.27 -0.02 -0.11 0.23 -0.02 0.54
-0.09 0.71 -0.23 0.02 0.41 0.35 0.05 0.09 -0.10 -0.30 -0.26
Big8 (3) -0.10 0.19 0.16 0.13 -0.02 -0.07 0.10 -0.03 0.28 -0.04
0.32 -0.06 0.06 0.06 0.10 0.01 0.07 -0.01 -0.12 -0.07
AUD_TEN (4) -0.14 0.29 0.17 0.95 -0.10 -0.02 0.16 -0.11 0.24
-0.04 0.30 -0.01 0.04 0.56 0.14 -0.01 0.07 -0.07 -0.21 -0.22
AUD_TEN x AUD_TEN (5) -0.14 0.29 0.17 1.00 -0.09 -0.02 0.16
-0.10 0.21 -0.03 0.28 -0.01 0.04 0.57 0.12 -0.01 0.05 -0.06 -0.18
-0.18
NOA (6) -0.06 0.06 0.00 -0.03 -0.03 0.14 -0.07 0.00 -0.01 -0.07
0.05 0.07 0.22 -0.12 -0.18 -0.06 0.26 0.11 0.04 -0.13
OP_Cycle (7) 0.08 -0.10 -0.05 0.01 0.01 0.24 -0.10 0.10 -0.12
-0.03 -0.13 0.01 -0.07 -0.03 -0.12 -0.01 -0.27 0.19 0.12 -0.12
Market_Share (8) -0.22 0.49 0.25 0.26 0.26 -0.07 -0.16 -0.09
0.18 0.03 0.27 -0.01 0.10 0.25 0.10 -0.02 0.07 -0.09 -0.15
-0.07
Z-Score (9) 0.07 0.03 -0.02 -0.05 -0.05 -0.21 0.07 -0.14 -0.02
-0.08 0.06 -0.22 -0.47 -0.17 0.03 -0.10 -0.21 0.20 0.21 0.15
INST_OWN (10) -0.20 0.55 0.29 0.26 0.26 0.05 -0.10 0.41 0.02
-0.11 0.66 -0.11 0.01 0.27 0.21 -0.06 0.02 -0.06 -0.27 -0.24
REM (11) 0.07 -0.10 -0.04 -0.03 -0.03 -0.18 -0.06 0.03 -0.09
-0.10 -0.14 0.03 0.00 -0.04 -0.30 -0.01 0.01 0.19 0.19 0.11
LMV (12) -0.22 0.71 0.32 0.27 0.27 0.11 -0.13 0.50 0.08 0.69
-0.15 -0.38 0.02 0.39 0.33 0.15 0.08 -0.03 -0.26 -0.25
BTM (13) -0.05 -0.18 -0.05 0.03 0.03 0.10 0.02 0.07 -0.28 -0.08
0.10 -0.39 0.11 0.00 -0.12 -0.31 0.10 -0.09 -0.08 -0.03
LEV (14) -0.12 0.09 0.07 0.07 0.07 0.29 -0.05 0.30 -0.67 0.02
0.02 0.05 0.12 0.10 0.02 0.02 0.31 -0.14 -0.17 -0.10
Firm_Age (15) -0.19 0.42 0.01 0.54 0.54 -0.06 0.01 0.35 -0.13
0.29 -0.02 0.30 0.09 0.14 0.17 -0.01 0.11 -0.11 -0.26 -0.25
ROA (16) -0.10 0.36 0.08 0.12 0.12 -0.12 -0.04 0.22 0.33 0.20
-0.28 0.37 -0.32 -0.13 0.14 0.21 0.07 -0.39 -0.36 -0.12
RET (17) -0.06 0.18 0.04 0.06 0.06 -0.07 -0.03 0.06 -0.09 0.03
-0.05 0.24 -0.37 0.02 0.08 0.31 -0.03 -0.02 0.00 0.00
Capital (18) -0.17 0.14 0.09 0.10 0.10 0.23 -0.23 0.24 -0.24
0.03 0.04 0.09 0.12 0.36 0.16 0.03 0.00 -0.17 -0.22 -0.19
Intangible (19) 0.07 -0.09 0.02 -0.01 -0.01 -0.02 0.34 -0.33
0.22 0.02 0.02 0.04 -0.24 -