Accounting Conservatism and Corporate Investment Decisions: Evidence from a Structural Assessment Chen Chen The University of Auckland Owen G Glenn Building 12 Grafton Road Auckland Tel: (+64) 9.923 7321 Email: [email protected]Yi Hu Chinese Academy of Science Email: [email protected]Karen Jingrong Lin University of Massachusetts Lowell Manning School of Business 1 University Ave Lowell, MA, 01854 Tel: 978.934.2406 Email: [email protected]
46
Embed
Accounting Conservatism and Corporate Investment Decisions ... · Accounting Conservatism and Corporate Investment Decisions: Evidence from a ... relevant because of the contingent
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Accounting Conservatism and Corporate Investment Decisions: Evidence from a Structural Assessment
Accounting Conservatism and Corporate Investment Decisions:
Evidence from a Structural Assessment
Abstract
We investigate whether accounting conservatism solves the misalignment of interest between managers and shareholders by increasing hurdle rates used by managers during project selections. We argue that accounting conservatism raises managerial cautiousness in project screening. By incorporating bad news timely into earnings, conservative accounting system increases the likelihood of early termination of unsuccessful projects, increasing personal costs of the manager and thus deterring managers from investing in projects merely to enjoy private benefits. Using a structural model and GMM estimation, we find that conservative accounting increases hurdle rates and such increases are more pronounced for firms that exhibit higher degree of agency problem. We also find conservatism adds value to firms when investment is under consideration. Our work sheds lights on the literature attempting to identify the relation between accounting conservatism and managers’ investment decisions.
We examine the effect of accounting conservatism on managers’ investment
decisions. We predict that managers evaluates projects with higher degree of
cautiousness and thus make less risky investment decisions when accounting
conservatism is in place. Our prediction is consistent with the view that
accounting conservatism reduces agency problem (Ball 2001; Ball and
Shivakumar 2005; Ahmed and Duellman 2007).
Corporate investment is an important decision made by managers. In 2009,
the total amount of corporate capital expenditures was $1,090.1 billion, and in
2008 it was close to $1,400 billion1 (Census Bureau 2011). However, corporate
investment decisions are not always made in the best interest of shareholders
because of the misalignment of managerial and shareholder interests: managers
invest in risky and even negative net present value projects because they enjoy
private benefit of investment (e.g., Jensen and Meckling 1976; Jensen 1986).
Accounting conservatism could potentially alleviate CEOs’ empire-building
incentives. Accounting conservatism is a reporting mechanism that reflects
accountants’ tendency to require a lower threshold to recognize unfavorable
1 The GDP (Gross Domestic Product) of the U.S. in 2008 was $14,369.1 billion and in 2009 was $13,939 billion. Corporat in the U.S. were about 10% of GDP, which is a significant portion.
4
news than favorable news (Basu 1997)2. We argue that accounting conservatism
improve corporate investment efficiency by changing managers’ investment
decisions. In the presence of accounting conservatism, losses from unsuccessful
investment projects are revealed timely in earnings, leading to early
terminations of such projects. The early terminations of projects penalize
managers by reduced compensation, impaired reputations, or even increased
likelihood of job turnover (Hirshleifer 1993; Weisbach 1995).
Ex ante, if managers are rational to anticipate the accelerated recognition of
losses and thus the timely termination of unsuccessful projects, they will be
more cautious to make the investment decisions from the outset. The increased
cautiousness to invest will be reflected in the hurdle rate used to evaluate an
investment project. In theory, hurdle rate just equals the cost of capital;
managers should not invest in a project if the return is lower than the cost of
financing. In practice, hurdle rate is observed to be either lower than the cost of
capital because of managerial discretion, or higher because of managerial
cautiousness of the irreversibility nature of the decision (e.g., Driver and Temple
2010).
2 We recognize that there are two types of accounting conservatism: conditional and
unconditional conservatism. In the context of investment, conditional conservatism is more
relevant because of the contingent nature of the turnouts of investment decisions. Therefore we
focus on the conditional conservatism.
5
To model the managerial cautiousness to invest, we following Whited and
Wu (2006) and use a structural approach to examine how accounting
conservatism affects hurdle rate and investment behaviors. In this intertemporal
investment model, the manager evaluates the options to invest today versus
tomorrow by comparing costs and benefits of the decision. An Euler investment
equation solves this utility-maximization decision problem: it is the first-order
condition in which the marginal cost of investing today equals that of tomorrow.
If a manager is unwilling to invest in new and risky projects, he/she will use a
lower discount factor to discount the cost of investing tomorrow, making
investing today less appealing compared to investing tomorrow. Hurdle rate is
the inverse function of this stochastic discount factor. In other words, a manager
who is more likely to reject the new risky investment projects today will have a
higher hurdle rate. In section 2, we present a model that formalizes this intuition.
To implement the Euler equation, we parameterized the stochastic discount
factor as a linear function of accounting conservatism and firm-level variables
that are shown to be correlated with the discount factor as in Kang et al. (2010).
We measure accounting conservatism using four proxies, TIMELINESS, CSCORE,
ACCRUAL and SKEWNESS. TIMELINESS captures the asymmetric timeliness of
bad news over good news as measured by the Basu (1997) model; CSCORE is
6
compiled following Khan and Watts (2009) to measure accounting conservatism
based on firm characteristics such as size, leverage, and market-to-book ratio.
ACCRUAL and SKEWNESS are constructed from the statistical properties of
accounting numbers (or of their components) following Beatty et al.(2008).
We find evidences supporting our hypothesis that accounting conservatism
improves firms’ investment efficiency by using a panel of firm-year observations
during 1982 -2009. We document that the presence of conservative accounting
significantly increases managers’ investment hurdle rates and induces
conservative investment decisions, and this effect is more pronounced when the
perceived agency problems are more severe. We also show that accounting
conservatism affects the stochastic discount factor in both financially
constrained and unconstraint firms, alleviating the concern that the observed
positive correlation between accounting conservatism and hurdle rate is mainly
driven by the effect of conservatism on external financing.
Prior studies have established the corporate governance role of accounting
conservatism in resolving information asymmetry, reducing agency costs, and
improving debt contracting efficiency (Ball 2001; Ball and Shivakumar 2005;
Ahmed and Duellman 2007; Beatty et al. 2008; Lafond and Roychowdhury 2008;
LaFond and Watts 2008; García Lara et al. 2009). On how accounting
7
conservatism affects corporate investment, Francis and Martin (2010)
(hereinafter FM2010) provide evidences that higher level of accounting
conservatism is associated with better M&A investment decisions measured by
greater market return around M&A announcements. Our study is closely related
to FM2010 but different in the following important ways. First, an expressed
concern about the relation between corporate investment and accounting
conservatism observed in FM2010 is that such relation could be associational
rather than causal3. We build on this study and provide a channel through which
accounting conservatism affects corporate investment behaviors: by altering
manager’s hurdle rate. This evidence more directly points to the relation
between accounting conservatism and corporate investment efficiency. Second,
FM2010’s results cannot be interpreted as supporting evidence on the impact of
conservatism on M&A profitability if managers cannot commit to use the same
accounting practice after the investment is made (Roychowdhury 2010). Our
theoretical framework avoids this pitfall because managers will not increase
hurdle rates if they anticipate future modification of accounting practices. Finally,
our study is generic in nature and does not rely on the market reaction of
3 As Francis and Martin pointed out, their result “cannot rule out the alternative explanation that
governance mechanisms are independently associated with both accounting conservatism and
better acquisition policies, in which case conservatism per se may not have a direct effect on
acquisition profitability” (Francis and Martin 2010, p.162).
8
investment decisions as FM2010 do; the latter often relays to other assumptions
such as market efficiency and market conditions such as bull or bear markets.
Our study contributes to the broader literature on how financial reporting
could affect firms’ real investment decisions (Biddle and Hilary 2006; Biddle et al.
2009; Beatty et al. 2010). We join the discussion of the role played by a specific
accounting mechanism, accounting conservatism, in managerial investment
decisions and highlight the monitoring role of conservatism (Francis and Martin
2010; Ma 2010; Roychowdhury 2010; Bushman et al. 2011). Our evidences
directly support the view that accounting conservatism has a deterrent effect on
managers’ empire-building incentives. The timely loss recognition imposes
personal costs to managers when investment projects turn out to be
unsuccessful, which in turn changes managers’ attitudes towards risks. Our
study, however, does not address the broader question on whether and how
other accounting quality proxies may change corporate investment behaviors.
This is because theoretically it is less clear how managers’ investment decisions
are related to other accounting quality proxies such as accrual quality, whereas
the ex post as well as the ex ante monitoring role of accounting conservatism is
clear (Roychowdhury 2010).
9
The remainder of this paper is organized as follows. Section 2 discusses prior
literature and develops testable hypotheses. Section 3 describes the model and
the research design. In Section 4 we discuss the construction of our variables and
describe our sample. Section 5 presents the empirical results, and Section 6
concludes the study.
2. HYPOTHESIS DEVELOPMENT
2.1The Effect of Accounting Conservatism on Hurdle Rate
Corporate investment is efficient when internal funds flow to investment
projects with the highest returns. However, managers deviate from the optimal
level of investment because of dysfunctional investment incentives (Stein et al.
2003). These dysfunctional incentives can be reflected in the hurdle rates used to
screen projects. Finance textbooks prescribe that managers should benchmark
the investment return to the cost of capital so that projects with returns lower
than the cost of capital are rejected. In practice, managers use hurdle rates that
are either higher or lower than the cost of capital, depending on the managerial
incentives to invest. For example, Driver and Temple (2010) show that a hurdle
rate that is higher than the cost of capital will be used if the manager is aware of
the embedded risk or is deterred by the irreversibility of the investment project;
10
a hurdle rate that is lower than the cost of capital will be used if the manager has
discretionary power and empire-building incentives such as those depicted in
the study by Jensen (1986). The low hurdle rate leads to overinvestment, which
is often at the cost of shareholders (Shleifer and Vishny 1989; Stulz 1990; Stein
et al. 2003).
Good corporate governance could alleviate the agency problem in
investment (e.g., Billett et al. 2011; Lin et al. 2011), and accounting conservatism
has been documented as an effective monitoring and contracting mechanism
(Ball 2001; Ball and Shivakumar 2005; Ahmed and Duellman 2007; Beatty et al.
2008; Lafond and Roychowdhury 2008; LaFond and Watts 2008; García Lara et
al. 2009). We argue that accounting conservatism improves corporate
investment efficiency by increasing hurdle rates and curbing managers’
incentives to invest in unsound projects merely to enjoy private benefits from
investment.
Accounting conservatism increases corporate investment efficiency in two
ways. On the one hand, by timelier incorporating bad news into earnings,
accounting conservatism triggers the board’s investigation of the unsuccessful
investment project that may lead to its early abandonment, saving shareholders
from potential future losses (Watts 2003). On the other hand, the expected
11
personal costs related to project abandonment such as compensation, reputation
and career concerns would increase managers’ cautiousness in choosing
investment projects at the outset.
If a manager is more cautious about project selections at the beginning, it
will be reflected in the hurdle rate used to evaluate projects. That is, the
stochastic discount factor, an inverse function of hurdle rate, used by the
manager to discount the adjustment costs of investing tomorrow will be lower so
that investing today appears to be less appealing. Therefore, we expect to
observe that the increased level of accounting conservatism is associated with
higher level of hurdle rates used by managers. The first hypothesis is as follows:
H1. The hurdle rate used in screening investment projects is increased with the
level of accounting conservatism.
2.2 Cross-sectional Variation of Ex Ante Agency Costs
The value of accounting conservatism in curbing managers’ incentives to
invest in unsound projects is likely to vary across firms depending on the ex ante
agency costs. Therefore, we consider three firm-level characteristics that could
be associated with higher ex ante agency costs: firms that are less transparent,
12
firms with more free cash flows, and firms that have higher growth
opportunities.
2.2.1 Information Transparency
Corporate transparency refers to the availability of information to market
participants outside the publicly-traded company. Transparency is beneficial to
capital allocation both at country and firm level (Bushman et al. 2004). It is
easier for managers in an opaque firm to cover up the unsuccessful projects and
operate loss-generating project at the cost of the shareholders (Watts 2003).
Therefore, conservatism in these firms is more valuable if it has the monitoring
role as we argued. Empirically, we use firm size and the number of analyst
following the firm are to proxy for the information environment of the firms
(Atiase 1985; Welker 1995; Aboody and Lev 2000; Brown et al. 2004).
2.2.2 Growth Opportunities: Book-to-Market Ratio
Firms with higher growth opportunities relative to assets have higher
contracting costs because growth opportunities are often intangible in nature
and information related to them are less verifiable. As a result, managers have
greater discretionary power and are more likely to manage earnings through
accruals, and thus agency costs are higher in high-growth firms. (Smith and
Watts 1992; Kwon and Yin 2006). In addition, high growth firms are more likely
13
to have more volatile returns and higher probability of lawsuits, which suggests
a higher demand of accounting conservatism (Khan and Watts, 2009).
Empirically, we use book-to-market ratio (an inverse measure of growth
opportunity) to measure the growth opportunity of firms.
2.2.3 Free Cash Flow
Managers are keen to expand the firm to the size beyond the optimal size,
often at the costs of shareholders. Excessive free cash flow allows managers to
do so without costly external financing. Overinvestment problem is found to be
most severe in firms with the highest level of free cash flows (Jensen 1986;
Richardson 2006).
In sum, managers in firms with lower transparency, greater growth
opportunities and higher level of free cash flow are more likely to use their
discretionary powers and enjoy private benefit from investments. Therefore, the
second hypothesis is derived as the follows:
H2. The effect of accounting conservatism on increasing hurdle rate is more
pronounced for firms with higher ex ante agency costs, measured by lower
transparency, greater growth opportunity, or higher level of free cash flow.
14
3. MODEL AND EMPIRICAL FRAMEWORK
Following Kang et al. (2010), we offer a partial-equilibrium model in which a
manager maximizes expected utility by choosing investment and consumption.
The following sections briefly discuss the model and lay out the empirical
framework.
3.1 A Simple Model
Suppose firm i uses capital to produce goods in each period t. A typical utility
maximization decision faced by an owner-manager is based on the expected
value of the firm (Eq. (1)), which is the sum of discounted future dividends,
subject to the dividend payout (Eq. (2)) as well as the stock accumulation (Eq.
(3)) constraints:
,0 ,0 , ( )i i i t
t
V Max E u d
, (1)
subject to
titititititi IKICKd ,,,,,, ),(),( , and
(2)
titiiti IKK ,,1, )1( ; (3)
whereio
E is the expectations operator conditional on the manager’s time 0
information set; is the one-period discount factor common to all firms; ( )u is
the manager’s utility which is concave if the manager is risk averse; andit
d is the
dividend paid by firm i in period t. In the constraints, Ki,t is the
15
beginning-of-period capital stock; ζi,t is a shock to the profit function; П(Ki,t , ζi,t )
is the firm’s profit function. Ii,t is investment during time t; C(Ii,t ,Ki,t) is the real
cost of adjusting the capital stock. δi is the firm-specific constant rate of
economic depreciation.
Solving the utility maximization problem yields the Euler condition for Ki,t:
1,1,1,
,
'
1,
'
,, )(1)(1()()()(
)()(1 tiititi
ti
ti
titiI
C
K
C
Kdu
duE
I
C , (4)
where CI
is the marginal adjustment cost of investment; the term '
, 1
'
,
( )
( )
i t
i t
u d
u d is
the marginal rate of substitution of dividends, or the pricing kernel from a
consumption-based asset pricing model; and K is the marginal profit of capital.
The intuition of Eq. (4) is that, when deciding whether to invest today or
tomorrow, managers evaluate the discounted benefit of harvesting in the future
net of costs versus the investment costs incurred today. This is analogous to the
typical capital allocation problem where a project’s net present value is
calculated.
In the heart of this utility maximization problem is the term)(
)(
,
'
1,
'
ti
ti
du
du , the
marginal rate of substitution of dividends. For notation purposes, we define
)(
)(
,
'
1,
'
1,
ti
ti
tidu
du
and re-write Eq. (4) as:
))(1)(1()()()(1 1,1,1,1,,, tiitititititi
I
C
K
C
KE
I
C . (5)
16
Eq. (5) presents the options that managers face: investing today or
tomorrow. It shows the marginal adjustment cost of investing in this period,
versus expected discounted cost of waiting to invest until next period. The
pricing kernel,)(
)(
,
'
1,
'
1,
ti
ti
tidu
du
, describes the essence of the managers’
investment decisions. On the margin, the manager must be indifferent to the
choice between investing in the current period and transferring those resources
to the next period. The perceived hurdle rate that firm managers use for optimal
investment can be defined as:
11
,
,
ti
tir . (6)
That is, if the manager is more cautious, the discount factor he/she uses will
be lower (and hence the hurdle rate will be higher), making investing today
appears to be less appealing.
3.2 Estimation of the Model
3.2.1 The Investment Decision
Eq. (5) can be converted into the following equation by replacing the
expectation operator with an error, ei,t+1:
M
m
ti
m
ti
ti
m
M
m
M
m
m
ti
ti
mi
m
ti
ti
m
ti
titi
ti
eK
I
K
I
K
I
m
m
K
VCY
2
1,
1
,
,
2 2
1
1,
1,
1,
1,
1,
1,1,
1,
.)(1
))(1)(1()(1
(7)
17
where αm, m=2,…., M are coefficients to be estimated, and M is a truncation
parameter that sets the highest power of ti
ti
K
I
,
, in the expansion. We follow prior
studies to set M = 3. Assuming perfect competition and a constant markup
attached to the cost to obtain the sales prices, we derive the marginal profit of
capital (Eq. (8)) and the real adjustment costs of investment (Eq. (9)):
ti
titi
titiK
VCYK
K ,
,,
,, ),(
, (8)
ti
M
m
m
ti
ti
mtiti KK
I
mKIC ,
2 ,
,
,, ))(1
(),(
, (9)
3.2.2 Hurdle Rate
The stochastic discount factor, Γi,t+1 is parameterized to estimate Eq. (7). We
specify Γi,t+1 as a function of accounting conservatism (CONSERVATISMi,t,) and
several firm-level characteristics that are known to be related to the discount
factor (Fama and French 1992; Jegadeesh and Titman 1993; Kang et al. 2010).
In Eq. (15), variables Xi,t, DRi,t, Ri,t are defined as in Eq. (12). Sizei,t is the
natural logarithm of market value of equity, MBi,t is the market-to-book ratio, and
Levi,t is the leverage. We calculate the firm-year level proxy for accounting
conservatism, CSCOREi,t, by estimating Eq. (15). The advantage of using CSCOREi,t
is that this measure controls for the time- and firm- specific characteristics of
firms’ reporting conservatism.
23
Accounting conservatism based on Givoly and Hayn (2000) and Beatty
et al. (2008). Our third and fourth accounting conservatism measures,
ACCRUALi,t and SKEWNESSi,t are estimated based on the time-series properties of
earnings and accruals. Specifically, we estimate the accumulated non-operating
accruals (ACCRUALi,t) that captures the extent to which earnings incorporate
negative non-operating accruals. It is calculated as the ratio of non-operating
accruals6 to total assets cumulated over the previous three years and multiplied
by -1. We estimate the level of skewness of earnings distribution resulted from
the asymmetric timeliness of good news and bad news incorporated in earnings
due to accounting conservatism: SKEWNESSi,t is calculated by the difference
between the skewness in cash flows from operations and the that in earnings
over the preceding three years window.
4.2.2 Variables used in the investment Euler equation
We estimate Equations (7) and (10) simultaneously. The key variable for the
empirical analysis is corporate investment, Ii,t, which is defined as the difference
6 Nonoperating accruals are defined as net income (NI) plus depreciation (DP) minus cash flow from operations (OANCF) minus the change in accounts receivable (RECT) minus the change in inventories (INVT) minus the change in prepaid expenses (XPP) plus the change in accounts payable (AP) plus the change in tax payable (TXP). Data on cash flow from operations (OANCF)since 1987 is available. If OANCF is missing, we estimate cash flow from operations as funds from operations (FOPT) minus the change in current assets (ACT) minus the change in short term debt (DLC) plus the change in current liabilities (LCT) plus the change in cash (CHE). Accumulated nonoperating accruals are multiplied by -1 so that the value of the conservatism proxy increases with a firm’s level of conservatism.
24
between capital expenditure and the sales of property, plant, and equipment in
year t for firm i.7 Capital stock, Ki,t, is measured by beginning-of-year total assets,
and is used to deflate other firm-level characteristic variables in order to reduce
heteroskedasticity. Firm i's output Yi,t in year t is measured by its sales, and
variable cost (VCi,t) is the sum of the costs of goods sold and the selling, general,
and administrative expenses. Both Yi,t and VCi,t are deflated by beginning-of-year
total assets. To parameterize the discount factor (i.e., Eq. (10)), market model
beta, i,t, is calculated by fitting the CAPM on a rolling basis with monthly stock
returns over the previous five years for every firm-year observations. The
market return (rm) and the risk-free rate (rf) are calculated by the CRSP
value-weighted market index return and by the one-month Treasury bill rate,
respectively. Total stock return volatility (Annvoli,t) is the annualized standard
deviation of monthly stock returns over the past five years. Firm size (Sizei,t) is
the logarithm of the year-end market value of the firm's common equity,
Book-to-market equity (btmi,t) is the book value of the firm's equity divided by its
market value of common equity, cumulative stock return (Annreti,t) is calculated
over the previous twelve months for every firm-year observations, and leverage
(Levi,t) is the ratio of the book value of the firm's debts to its total assets. In the
7 We also use alternative measures of investment defined as the sum of the capital expenditure net
of the sale of PPE and Research and Development cost. We get the similar results.
25
GMM estimation, we use rf,K
Y , K
I ,K
VC ,, Size, Annret, Annvol, and Lev,all lagged
by two periods, as instruments. A constant is also included as an instrumental
variable. The GMM estimation is implemented in first difference.
4.3 Descriptive Statistics
Table1 exhibits the descriptive statistics of the variables used in our main
tests. All the variables in our investment Euler equation have the similar sample
distribution to that presented in Kang et al. (2010). The mean of our main
interested variables, TIMELINESSi,t, CSCOREi,t, ACCRUALi,t, and SKEWNESSi,t, are
also comparable to those presented in studies by Hui et al. (2011), Khan and
Watts (2009), Beatty et al. (2008) and Givoly and Hayn (2000), respectively. The
Pearson and Spearman correlation matrix with each pair of variables are
presented in Table 2. Variable pairs that are correlated within 5% significance
are starred.
[Insert Table 1 and Table 2 about here]
5. EMPIRICAL RESULTS FROM MULTIPLE REGRESSIONS
5.1 The Relation between Accounting Conservatism and Hurdle Rates
26
We test hypothesis H1 by simultaneous estimations of models Eq. (7) and Eq.
(10), and the results are presented in Table 3. Columns (1)-(4) present different
conservatism measures and their effects on the discount factor (i.e., the inverse
function of hurdle rate) that governs a manager’s investment decisions. Column
(1) reports that the parameter estimate of TIMELINESSi,t is -0.243, and is
statistically significant at the 1% level. Column (2)-(4) present estimates for
CSCOREi,t, ACCRUALi,t, and SKEWNESSi,t are also negative and significant at 5%
level or better. Taken together, Columns (1)-(4) suggest that managers who
work for firms with higher level of accounting conservatism use a lower discount
factor (i.e., a higher hurdle rate) when making investment decisions. These
evidences are consistent with monitoring role of accounting conservatism on
curbing managers’ overinvestment incentives. Signs of parameters on all other
control variables are consistent with those documented in prior studies (Whited
1998; Whited and Wu 2006; Kang et al. 2010).
J-test of overidentifying restrictions is commonly included in GMM
estimations to detect the possible misspecification of models. The null
hypothesis is that these specifications do not have overidentifying restrictions
and the test statistics is chi-squared distributed with degrees of freedom equal to
the number of overidentifying restrictions. J-test reported in Column (1) does
27
not reject the exclusion restrictions, indicating a proper specification of the
model and orthogonal property of the instruments but J-test statistics reported
in columns (2)-(4) produce rejections of the exclusion restrictions. Therefore,
although accounting conservatism measures CSCOREi,t, ACCRUALi,t, and
SKEWNESSi,t used to estimate Eq. (10) produce coefficients consistent with the
theoretical derivation, we do not proceed to use these variables in the rest of our
empirical tests. That is, in the following tests we use TIMELINESSi,t as our key
variable to proxy for accounting conservatism.
[Insert Table 3 about here]
5.2 The Moderating Effect of Accounting Conservatism on Hurdle Rate
Conditional ex Ante Agency Costs
Hypothesis H2 is tested by simultaneous estimations of models Eq. (7) and
Eq. (10) in different subsamples with various degrees of ex ante agency costs,
and Table 4 presents these regression results. Columns (1)-(2) present the
parameter estimates based on the subsamples formed according to size. The
coefficient of TIMELINESSi,t estimated in the small-firm subsample is negative
and significant at the 10% level, while this statistic is insignificant when
estimated in the large-firm subsample. This result suggests that accounting
28
conservatism increases managers’ hurdle rates in a greater level in small firms
than in large firms. Similarly, Columns (3) – (8) show that accounting
conservatism has a stronger effect in raising managers’ hurdle rates if the firm is
more opaque (i.e., has a smaller number of analysts following), has greater
growth opportunities, and higher level of free cash flows. Overall, these results
indicate a stronger moderating effect of accounting conservatism in firms with
higher ex ante agency costs. Note that, in seven out of eight cases, J-tests do not
reject the overidentifying restrictions, suggesting the good fit of these models.
[Insert Table 4 about here]
5.3 Additional Tests
To facilitate our understanding of how accounting conservatism affects
managers’ investment decisions and the consequences related to such decisions
and robustness of our tests, we conduct several additional tests as the follows.
5.3.1 Accounting Conservatism, Future Financing Costs, and Hurdle Rate
One concern on the observed relation between accounting conservatism and
hurdle rates is that accounting conservatism may affect cost of financing and
thus change the hurdle rate used by managers: hurdle rate is the average cost of
29
capital plus managers’ discretionary adjustments (Driver and Temple 2010).
That is, if employing a more conservative accounting system could reduce future
cost of financing, managers will delay investment to future, which decreases the
stochastic discount factor and increases the hurdle rate. This suggests that
higher level of accounting conservatism per se does not lead to the lower hurdle
rate because of better monitoring of managers.
If cheaper future external financing is a plausible alternative explanation, we
should observe the moderating effect of accounting conservatism on hurdle rates
pronounced in the financially constrained firms but not in the financially
unconstrained firms. We hence divide our samples into two subsamples, firms
that are financially constrained and unconstrained, using three different proxies
following Denis and Sibilkov 2010 and replicate our analyses concerning the
effect of accounting conservatism on hurdle rate based on these two groups of
firms separately. Results reported in Table 5 demonstrate the moderating effect
of accounting conservatism on hurdle rates in both financially constrained and
unconstrained subsamples. This addresses the concern that accounting
conservatism, by lowering future financing costs, is the key driving factor of a
lower hurdle rate used by the manager.
[Insert Table 5 about here]
30
5.3.2 Accounting Conservatism, Hurdle Rate and Underinvestment
Prior studies express the concern that conservative accounting could induce
managers’ dysfunctional incentives to underinvest (Leuz 2001; Watts 2003). Our
study appears to imply the same conclusion. If accounting conservatism
increases hurdle rates in a disproportional way, managers will give up investing
and enjoy the “quiet life” (Stein et al. 2003). A priori, our theoretical framework
cannot reject the possibility that the increase of hurdle rate resulted from
heightened accounting conservatism leads to underinvestment. However, we are
keen to find out whether such an increase in hurdle rates would hurt
shareholder value.
We test whether the increased level of accounting conservatism has a
negative impact on firm value when investment is under consideration. We test
the following regression model:
, , , ,0 1 , 1 2 , 1 3 4
, , , ,5 76 , 1 8 9 10 , 1
, ,11 , 1*
*i t i t i t i ti t i t
i t i t i t i ti t i t
i t i ti tLev Cash YFE IFE
Ret TIMELINESS TIMELINESS I I Div
Cash Cash Earn NetFin Int Lev
, (16)
where Reti,t is the market-adjusted stock return over a fiscal year and the
benchmark portfolios used are twenty five Fama-French value-weighted
portfolios. ΔIi,t, the change of the capital minus the sale of property, plant, and
equipment. TIMELINESSi,t-1 is the firm-specific coefficient estimates from the
Basu (1997) model in year t-1; ΔCashi,t, is the change in cash, ΔIncomei,t is the
31
change in earnings before interest and extraordinary items, ΔRDi,t is the change
in R&D expenses, ΔDividendi,t is the change in dividends, ΔInteresti,t, is the
change in interest expenses, Cashi,t-1, is the lagged cash holdings, Levi,t , is the
leverage ratio, and Netfini,t is the net financing. The explanatory variables except
leverage are standardized by lagged market equity. Regressions are estimated
using OLS. Statistical significance is computed using heteroskedasticity and
autocorrelation robust standard errors that are clustered at the firm level.
If accounting conservatism, by increasing the hurdle rate, induces
dysfunctional underinvestment incentives for managers, we should observe β2 to
be negative and significant. In addition, if the underinvestment incentives
induced by accounting is detrimental, it will affect firms regardless of the ex ante
agency costs of the firms. On the contrary, if accounting conservatism plays
important monitoring roles, the firm value should increase, and the increments
should be larger for firms with higher perceived agency costs.
Table 6 reports the results of the regression (Eq. (16)). The coefficient of the
interaction term TIMELINESSi,t *∆Ii,t is significantly positive at 5%, indicating on
average the stock market places greater value on capital investments made by
firms with more conservative accounting than it does for capital investments of
the firms with less conservative accounting.
32
[Insert Table 6 about here]
Table 7 presents results from estimations of Eq. (16) based on subsamples of
firms that have various degrees of ex ante agency costs. Columns (1)-(2) present
results when Sizei,t is used to partition the sample. The estimated coefficient on
TIMELINESSi,t *∆Ii,t, is significantly positive for the small size group, whereas it is
insignificant for the large size group. This evidence is consistent with the
prediction that when firms have higher degrees of agency problems, accounting
conservatism adds more value. Columns (3)-(8) convey the same message when
the sample is partitioned according to analyst following, growth opportunity and
free cash flow holdings.
[Insert Table 7 about here]
5.3.3 Controlling for other accounting quality measures and corporate
governance measures
We further control for other accounting quality measures and firm-level
corporate governance measures in Equation (10) to see whether our results still
hold. In particular, we include the performance-adjusted abnormal accrual
suggested by Kothari et al. (2005), the accrual quality measure of Dechow and
Dichev (2002), the financial statement readability measure suggested by Li
(2008), the GSCORE developed by Gompers et al. (2003), the CEO-president
33
duality dummy and board independence ratio suggested by Harford et al. (2008).
The untabulated results show that our conservatism measures are still positively
correlated to managers’ hurdle rates and other accounting matrix are negatively
correlated with managers’ hurdle rates. This is not surprising given that
managers’ hurdle rate is the sum of firms’ cost of capital and managers’
subjective adjustment and Francis et al. (2004) already finds that the accounting
quality measures except for conservatism measures are negatively correlated
with firms’ cost of capital. In addition, we find the corporate governance
variables are also positively correlated with managers’ hurdle rates, indicating
that the better the corporate governance of the firm, the higher hurdle rates
managers use to screen the projects.
6. CONCLUSIONS
Misalignment of interest as well as asymmetric information between
managers and shareholders are the main reasons for investment inefficiency.
Accounting conservatism is documented as an importing monitoring and
contracting mechanism, and we investigate how this mechanism is related to
corporate investment decisions. We argue that accounting conservatism could
increase managerial cautiousness during project. By recognize bad news timely,
accounting conservatism accelerates terminations of unsuccessful projects,
34
increase personal costs of managers and thus deter managers from investing in
projects merely to enjoy private benefits.
Using a structural model and GMM estimation, we find that conservative
accounting increases hurdle rates and such increases are more pronounced for
firms that exhibit higher degree of agency problem. Our additional tests rule out
the alternative explanation that conservatism changes hurdle rate by reducing
the cost of future financing rather than by increasing the efficiency of contracting
with the managers. We also find conservatism adds value to firms when
investment is under consideration. Our results are robust when we further
control for other accounting quality and corporate governance measures. Our
work sheds lights on the literature attempting to identify the relation between
accounting conservatism and managers’ investment decisions.
35
Reference
Aboody, D., and B. Lev. 2000. Information asymmetry, r&d, and insider gains. The Journal
of Finance 55 (6): 2747-2766.
Ahmed, A. S., and S. Duellman. 2007. Accounting conservatism and board of director
characteristics: An empirical analysis. Journal of Accounting and Economics 43
(2-3): 411-437.
Atiase, R. K. 1985. Predisclosure information, firm capitalization, and security price
behavior around earnings announcements. Journal of Accounting Research 23 (1):
21-36.
Ball, R. 2001. Infrastructure requirements for an economically efficient system of public
financial reporting and disclosure. In Brookings-wharton papers on financial
Ball, R., and L. Shivakumar. 2005. Earnings quality in uk private firms: Comparative loss
recognition timeliness. Journal of Accounting and Economics 39 (1): 83-128.
Basu, S. 1997. The conservatism principle and the asymmetric timeliness of earnings.
Journal of Accounting and Economics 24 (1): 3-37.
Beatty, A., W. Scott Liao, and J. Weber. 2010. The effect of private information and
monitoring on the role of accounting quality in investment decisions.
Contemporary Accounting Research 27 (1): 17-47.
Beatty, A., J. Weber, and J. J. Yu. 2008. Conservatism and debt. Journal of Accounting and
Economics 45 (2-3): 154-174.
Biddle, G. C., and G. Hilary. 2006. Accounting quality and firm-level capital investment.
The Accounting Review 81 (5): 963-982.
Biddle, G. C., G. Hilary, and R. S. Verdi. 2009. How does financial reporting quality relate
to investment efficiency? Journal of Accounting and Economics 48 (2-3): 112-131.
Billett, M. T., J. A. Garfinkel, and Y. Jiang. 2011. The influence of governance on
investment: Evidence from a hazard model. Journal of Financial Economics 102
(3): 643-670.
Brown, S., S. A. Hillegeist, and K. Lo. 2004. Conference calls and information asymmetry.
Journal of Accounting and Economics 37 (3): 343-366.
Bushman, R. M., J. D. PIotroski, and A. J. Smith. 2004. What determines corporate
transparency? Journal of Accounting Research 42 (2): 207-252.
Bushman, R. M., J. D. Piotroski, and A. J. Smith. 2011. Capital allocation and timely
accounting recognition of economic losses: Blackwell Publishing Ltd, 1-33.
Census Bureau. 2011. 2011 capital spending report:U.S. Capital spending
patterns—2000-2009.
36
Dechow, P. M. and I. D. Dichev. 2002. The quality of accruals and earnings: the role of
accrual estimation errors. The Accounting Review 77 (Supplement): 35-59.
Denis, D. J., and V. Sibilkov. 2010. Financial Constraints, Investment, and the Value of
Cash Holdings. Review of Financial Studies 23 (1):247-269.
Driver, C., and P. Temple. 2010. Why do hurdle rates differ from the cost of capital?
Cambridge Journal of Economics 34 (3): 501-523.
Fama, E. F., and K. R. French. 1992. The cross-section of expected stock returns. The
Journal of Finance 47 (2): 427-465.
Francis, J., R. LaFond, P. M. Olsson, and K. Schipper. 2004. Costs of equity and earnings
attributes. The Accounting Review 79 (4): 967-1010.
Francis, J. R., and X. Martin. 2010. Acquisition profitability and timely loss recognition.
Journal of Accounting and Economics 49 (1-2): 161-178.
García Lara, J., B. García Osma, and F. Penalva. 2009. Accounting conservatism and
corporate governance. Review of Accounting Studies 14 (1): 161-201.
Givoly, D., and C. Hayn. 2000. The changing time-series properties of earnings, cash
flows and accruals: Has financial reporting become more conservative? Journal
of Accounting and Economics 29 (3): 287-320.
Gompers, P. A., J. Ishii, and A. Metrick. 2003. Corporate governance and equity prices.
Quarterly Journal of Economics 118:107.55.
Harford, J., S. Mansi., and W. Maxwell. 2008. Corporate governance and firm cash holding
in the US. Journal of Financial Economics 87: 535-555.
Hirshleifer, D. 1993. Managerial reputation and corporate investment decisions.
Financial Management 22 (2): 145-160.
Hui, K. W., S. Klasa, and E. Yeung. 2011. Corporate suppliers and customers and
accounting conservatism. Journal of Accounting and Economics 53
(1)-(2):115-135.
Jegadeesh, N., and S. Titman. 1993. Returns to buying winners and selling losers:
Implications for stock market efficiency. The Journal of Finance 48 (1): 65-91.
Jensen, M. C. 1986. Agency costs of free cash flow, corporate finance, and takeovers. The
American Economic Review 76 (2): 323.
Jensen, M. C., and W. H. Meckling. 1976. Theory of the firm: Managerial behavior, agency
costs and ownership structure. Journal of Financial Economics 3 (4): 305-360.
Kang, Q., Q. Liu, and R. Qi. 2010. The sarbanes-oxley act and corporate investment: A
structural assessment. Journal of Financial Economics 96 (2): 291-305.
Khan, M., and R. L. Watts. 2009. Estimation and empirical properties of a firm-year
measure of accounting conservatism. Journal of Accounting and Economics 48
(2-3): 132-150.
37
Kwon, S. S., and Q. J. Yin. 2006. Executive compensation, investment opportunities, and
earnings management: High-tech firms versus low-tech firms. Journal of
Accounting, Auditing and Finance 21 (2): 119-148.
Lafond, R., and S. Roychowdhury. 2008. Managerial ownership and accounting
conservatism: Blackwell Publishing Inc, 101-135.
LaFond, R., and R. L. Watts. 2008. The information role of conservatism. The Accounting
Review 83 (2): 447-478.
Leuz, C. 2001. Comments on infrastructure requirements for an economically efficient
system of public financial reporting and disclosure. In Brookings-wharton papers
on financial services, 2001: Brookings Institution Press, 170-178.
Li,F. 2008. Annual report readability, current earnings and earnings persistence. Journal
of Accounting and Economics 45: 221-247.
Lin, C., Y. Ma, and Y. Xuan. 2011. Ownership structure and financial constraints:
Evidence from a structural estimation. Journal of Financial Economics 102 (2):
416-431.
Ma, T. 2010. Accounting conservatism and corporate investment. Working Paper,
Washington University in St. Louis.
Pope, P. F., and M. Walker. 1999. International differences in the timeliness,
conservatism, and classification of earnings. Journal of Accounting Research 37:
53-87.
Richardson, S. 2006. Over-investment of free cash flow. Review of Accounting Studies 11
(2): 159-189.
Roychowdhury, S. 2010. Discussion of: "Acquisition profitability and timely loss
recognition" by j. Francis and x. Martin. Journal of Accounting and Economics 49
(1-2): 179-183.
Shleifer, A., and R. W. Vishny. 1989. Management entrenchment: The case of
manager-specific investments. Journal of Financial Economics 25 (1): 123-139.
Smith, C. W., and R. L. Watts. 1992. The investment opportunity set and corporate
financing, dividend, and compensation policies. Journal of Financial Economics
32 (3): 263-292.
Stein, J. C., M. H. G.M. Constantinides, and R. M. Stulz. 2003. Agency, information and
corporate investment. In Handbook of the economics of finance: Elsevier,
111-165.
Stulz, R. 1990. Managerial discretion and optimal financing policies. Journal of Financial
Economics 26 (1): 3-27.
Watts, R. L. 2003. Conservatism in accounting part i: Explanations and implications. In
Accounting Horizons: American Accounting Association, 207-221.
38
Weisbach, M. S. 1995. Ceo turnover and the firm's investment decisions. Journal of
Financial Economics 37 (2): 159-188.
Welker, M. 1995. Disclosure policy, information asymmetry, and liquidity in equity
markets*. Contemporary Accounting Research 11 (2): 801-827.
Whited, T. M. 1998. Why do investment euler equations fail? Journal of Business &
Economic Statistics 16 (4): 479-488.
Whited, T. M., and G. Wu. 2006. Financial constraints risk. Review of Financial Studies 19
(2): 531-559.
39
Appendix I Variable Definition
Variable Conservatism Variables
TIMELINESSi,t The coefficient estimate of the negative return when regressed on earnings following the Basu (1997) model, estimating from at least an eight-year rolling window of each firm-year.
CSCOREi,t Firm-specific conservatism measure estimated following Khan and Watts (2009). ACCRUALi,t Accumulated non-operating accruals estimated following Givoly and Hayn (2000).
SKEWNESSi,t the difference between the skewness in cash flows and that in earnings distributions, estimated following Beatty et al. (2008).
Variables in Investment Euler Equation
Ii,t Investment, defined as capital expenditure minus the sale of property, plant, and equipment (CAPX – SPPE).
Ki,t Capital stock, = total assets of a fiscal year (AT).
VCi,t Variable costs incurred in a given year, including cost of goods sold (COGS) and selling, general, and administrative costs, i.e., = COGS + SGA.
Yi,t Output, defined as total sales (Sale)
i,t A firm’s beta computed from fitting a CAPM on a rolling basis with monthly returns over the previous five years.
Sizei,t Defined as the logarithm of a firm’s market capitalization, which is the product of the number of shares outstanding at year-end and the stock price at year-end (PRCC_F*CSHO).
btmi,t Book-to-market Ratio, calculated as book equity (CEQ) over market equity (PRCC_F*CSHO).
Annreti,t Annual return. Calculated as the past 12-month cumulative stock return.
Annvoli,t Annual stock volatility. Defined as the annualized standard deviation in monthly stock returns over the previous five years.
rf The risk-free rate, measured by the one-month Treasury bill rate. Levi,t Leverage. The ratio of book value of long term debt to total assets. Other variables Analyst The number of analyst following. FCFi,t Free cash flow of the firm defined as Richardson 2006. Variables in Market Valuation Test
ΔIi,t Change in investment, defined as capital expenditure minus the sale of property, plant, and equipment (CAPX – SPPE) scaled by lagged market value of equity (PRCC_F*CSHO).
ΔCashi,t, Change in cash (CHE) scaled by lagged market value of equity (PRCC_F*CSHO).
ΔIncomei,t, Change in earnings before interest and extraordinary items (IB) scaled by lagged market value of equity (PRCC_F*CSHO).
ΔDividendi,t, Change in dividend. (DVC) scaled by lagged market value of equity (PRCC_F*CSHO).
ΔInteresti,t Change in Interest Expenses (EXP) scaled by lagged market value of equity (PRCC_F*CSHO).
Netfini,t, The sum of issuance of equity and debt during the fiscal year (DLTIS and SSTK) scaled by lagged market value of equity (PRCC_F*CSHO).
40
Table 1. Descriptive Statistics This table presents the descriptive statistics of the variables used in the empirical analysis on an annual basis over 1982–2009. The sample is constructed from the Compustat and CRSP databases. Investment (I), total variable costs (VC) and output (Y) are scaled by the beginning-of-the-year total assets. The conservatism measures used are TIMELINESS, CSCORE, ACCRUAL and SKEWNESS. TIMELINESS is the coefficient estimates following the Basu (1997) model, estimating from at least an eight-year rolling window of each firm-year.; CSCORE is estimated by following Khan and Watts (2009); ACCRUAL is the accumulated nonoperating accruals estimated by following Givoly and Hayn (2000); and SKEWNESS is the difference between the skewness in cash flows and the skewness in earnings, estimated by following Beatty et al. (2008). All other variables are defined in Appendix I. Variable Obs Mean Median SD Q1 Q3 Min Max
Note: This table presents the Spearman/ Pearson correlation of each variable used in the empirical analysis on an annual basis over 1982–2009. All variables
are defined in Table 1.
42
Table 3. GMM Estimates of Investment Euler Equation This table reports the results from GMM estimations. We estimate our investment Euler equation on the sample consisting of firms during 1982–2009. The number of firms varies as various conservatism measures require different data availability.
, 1 , 1 , 1 , 1 1
,
2 2, 1 , 1 , 1
Mi,t m-1
m i,t+1
m-2 i,t
,
1* ( ) (1 )(1 ( ) )
I=1+ α ( ) +e ., and
K
(
M Mi t i t i t i tm m
i t m i m
m mi t i t i t
i t
Y VC I Im
K m K K
l
0 1 , 2 , 3 , 4 , 5 , )i t i t i t i t i tl l Size l btm l Annret l Conservatism
I/K is used to measure investment, and α2 and α3 are adjustment cost parameters. We use four different measures to proxy for Conservatism, namely, TIMELINESS, CSCORE, ACCRUAL, and SKEWNESS. Specifically, TIMELINESS is the coefficient estimates of negative returns following the Basu (1997) model, estimating from at least an eight-year rolling window of each firm-year; CSCORE is the firm-specific conservatism measure estimated following Khan and Watts (2009; ACCRUAL is the accumulated non-operating accruals estimated following Givoly and Hayn (2000), and SKEWNESS is the difference between the skewness in cash flows and the skewness in earnings, estimated following Beatty et al. (2008). All other variables are defined in Appendix I. We conduct the GMM estimation on the model in the first differences with twice-lagged instrumental variables. For all estimations, we include the following instruments: constant; rf, Y/K, VC/K, I/K, , size, btm, annvol, annret, leverage, each lagged by two periods; The λ2 values, degrees of freedom, and p-values from the J-test (overidentifying restriction test) are reported in the bottom section of the table.
(1) TIMELINESSi,t
(2) CSCOREi,t
(3) ACCRUALi,t
(4) SKEWNESSi,t
μi,t -0.002 (-0.13)
0.012** (2.07)
-0.005 (-0.56)
0.023** (2.03)
α2 0.490*** (2.73)
0.309*** (4.88)
0.257** (2.08)
0.662*** (4.81)
α3 -0.169*** (-2.85)
-0.098*** (-4.88)
-0.088** (-2.28)
-0.205*** (-4.65)
i,t 0.0239 (1.03)
-0.175*** (-19.75)
-0.038*** (-5.68)
-0.278*** (-13.87)
Sizei,t -0.079*** (-9.54)
-0.085*** (-28.04)
-0.071*** (-11.34)
-0.086*** (-13.72)
btmi,t -0.668*** (-12.24)
-0.584*** (-24.15)
-0.564*** (-17.35)
-0.348*** (-10.70)
Annreti,t 0.036 (0.11)
0.415*** (4.98)
-1.257*** (-10.41)
0.850*** (4.64)
Conservatismi,t -0.243*** (-3.16)
-0.442*** (-4.55)
-0.679** (-2.40)
-0.792*** (-5.35)
λ2 of J Test 2.430 4.441 15.097 65.227 Degree of freedom 2 2 2 2 p-value of J Test 0.2967 0.1081 0.0005 0 N 27,978 42,749 25,128 45,443
Note: *** , ** and * denote significance at the 1, 5, and 10% levels, respectively. Coefficient estimates are reported with t-statistics in parentheses.
43
Table 4. Value of Conservatism When Ex Ante Agency Costs Vary: GMM Estimates of Investment Euler Equation for Subsamples
This table reports the results from GMM estimations from different subsets of our sample. We estimate our investment Euler equation on the sample consisting of firms during 1982–2009. The number of firms varies as various conservatism measures require different data availability.
, 1 , 1 , 1 , 1 1
,
2 2, 1 , 1 , 1
Mi,t m-1
m i,t+1
m-2 i,t
,
1* ( ) (1 )(1 ( ) )
I=1+ α ( ) +e ., and
K
(
M Mi t i t i t i tm m
i t m i m
m mi t i t i t
i t
Y VC I Im
K m K K
l
0 1 , 2 , 3 , 4 , 5 , )i t i t i t i t i tl l Size l btm l Annret l Conservatism
I/K is used to measure investment, and α2 and α3 are adjustment cost parameters. We use four different measures to proxy for Conservatism, namely, TIMELINESS, CSCORE, ACCRUAL, and SKEWNESS. Specifically, TIMELINESS is the coefficient estimates of negative returns following the Basu (1997) model, estimating from at least an eight-year rolling window of each firm-year; CSCORE is the firm-specific conservatism measure estimated following Khan and Watts (2009; ACCRUAL is the accumulated non-operating accruals estimated following Givoly and Hayn (2000), and SKEWNESS is the difference between the skewness in cash flows and the skewness in earnings, estimated following Beatty et al. (2008). All other variables are defined in Appendix I. We conduct the GMM estimation on the model in the first differences with twice-lagged instrumental variables. For all estimations, we include the following instruments: constant; rf, Y/K, VC/K, I/K, , size, btm, annvol, annret, leverage, each lagged by two periods; The λ2 values, degrees of freedom, and p-values from the J-test (overidentifying restriction test) are reported in the bottom section of the table.
(1) Size-Small
(2) Size-Large
(3) Analyst-Low
(4) Analyst-High
(5)
BTM-Small
(6)
BTM-Large
(7) FCF-Small
(8) FCF-Large
μi,t -0.018 (-0.63)
-0.072** (-2.11)
-0.171*** (-2.57)
-0.003 (-0.18)
0.181** (2.12)
-0.045** (-2.56)
-0.014 (-0.44)
-0.069 (-0.69)
α2 -0.071 (-0.22)
-1.199*** (-2.74)
-1.800*** (-3.50)
0.011 (0.16)
2.425** (2.54)
-1.125*** (-3.67)
0.476** (1.97)
0.209 (0.24)
α3 0.015 (0.19)
0.433*** (2.85)
0.645*** (3.50)
-0.005 (-0.52)
-0.725** (-2.41)
0.335*** (3.17)
-0.014** (-2.01)
-0.099 (-0.30)
i,t -0.096*** (-5.11)
-0.133 (-1.52)
-0.171*** (-5.42)
-0.003 (-0.61)
0.012 (0.34)
-0.119*** (-3.24)
0.009 (0.21)
0.304* (1.73)
Sizei,t N/A N/A 0.008 (0.40)
0.083*** (-44.69)
-0.141*** (-4.96)
-0.031*** (-3.87)
-0.082*** (-7.08)
-0.051 (-1.28)
btmi,t -0.419*** (-7.23)
0.291 (0.82)
-0.052 (-0.62)
0.418*** (-18.80)
N/A N/A -0.767*** (-7.02)
0.178 (0.75)
Annreti,t -0.521*** (-10.91)
-1.626*** (-4.79)
-0.799* (-1.80)
0.568*** (9.47)
0.245 (0.30)
-2.521*** (-4.26)
0.083 (0.15)
-0.148** (-2.48)
TIMELINESSi,t -0.228* (-1.85)
0.276 (1.28)
-0.215** (-2.11)
0.00 (0.12)
-0.268** (-2.51)
-0.009 (-0.18)
-0.218 (-1.56)
-0.761** (-2.48)
λ2 values of J Test 5.191 0.127 9.362 1.603 1.829 2.554 1.823 2.231
Degree of freedom 2 2 2 2 2 2 2 2
p-value of J Test 0.075 0.938 0.009 0.449 0.401 0.279 0.402 0.326
N 9,025 9,173 8,650 11,358 8,701 9,284 9,093 8,980
Note: *** , ** and * denote significance at the 1, 5, and 10% levels, respectively. Coefficient estimates are reported with t-statistics in parentheses.
44
Table 5. Additional Test: GMM Estimates of Investment Euler Equation for Financially Constrained and Unconstrained Firms
This table reports the results from GMM estimations from different subsamples partitioned by various financial constrained criteria. We estimate our investment Euler equation on the sample consisting of firms during 1982–2009. The number of firms varies as various conservatism measures require different data availability.
, 1 , 1 , 1 , 1 1
,
2 2, 1 , 1 , 1
Mi,t m-1
m i,t+1
m-2 i,t
,
1* ( ) (1 )(1 ( ) )
I=1+ α ( ) +e ., and
K
(
M Mi t i t i t i tm m
i t m i m
m mi t i t i t
i t
Y VC I Im
K m K K
l
0 1 , 2 , 3 , 4 , 5 , )i t i t i t i t i tl l Size l btm l Annret l Conservatism
I/K is used to measure investment, and α2 and α3 are adjustment cost parameters. We use TIMELINESS as our measure of conservatism. TIMELINESS is the coefficient estimates of negative returns following the Basu (1997) model, estimating from at least an eight-year rolling window of each firm-year. All other variables are defined in Appendix I. We conduct the GMM estimation on the model in the first differences with twice-lagged instrumental variables. For all estimations, we include the following instruments: constant; rf, Y/K, VC/K, I/K, , size, btm, annvol, annret, leverage, each lagged by two periods; The λ2 values, degrees of freedom, and p-values from the J-test (overidentifying restriction test) are reported in the bottom section of the table. We use three different measures of financial constraint following Denis and Sibilkov 2010. For each year, we assign those firms in the bottom (top) three deciles of the annual dividend payout ratio distribution to the financially constrained (unconstrained) group. Firms are classified as financially unconstrained if they have had their short-term debt rated by S&P’s and their debt is not in default. Firms are classified as constrained if they have debt outstanding that year, but have never had their short-term debt rated before (or the rating is unavailable). Firms are classified as financially unconstrained if they have had their long term debt rated by Standard & Poor’s (S&P Long-term Senior Debt Rating is available on Compustat) and their debt is not in default (rating of “D” or “SD”). Firms are classified as constrained if they have debt outstanding that year, but have never had their public debt rated before (or the long-term debt rating is unavailable).
(1)
Divdend- Constrained
(2) Divdend- Unconstrained
(3) Paperrating- Constrained
(4) Paperrating- Unconstrained
(5) Bondrating- Constrained
(6) Bondrating- Unconstrained
μi,t -0.069 (-0.88)
-0.048 (-0.82)
-0.257 (-1.61)
0.011 (0.19)
-0.045 (-1.33)
-0.063** (-2.33)
α2 -0.457 (-0.59)
0.060 (0.12)
-0.105 (-0.13)
0.535 (1.04)
0.192 (0.71)
-0.447*** (-1.49)
α3 0.141 (0.58)
-0.063 (-0.32)
0.059 (0.22)
-0.189 (-1.13)
-0.062 (-0.73)
0.136 (1.30)
i,t -0.057 (-0.84)
0.182 (1.49)
0.188 (1.03)
-0.161** (-2.05)
0.002 (0.07)
0.092 (1.35)
Sizei,t -0.057*** (-4.24)
-0.054*** (-3.24)
-0.1.7* (-1.85)
-0.068*** (-3.01)
-0.089*** (-5.86)
-0.037*** (-2.98)
btmi,t -0.247** (-2.33)
-0.456*** (-2.65)
-0.216*** (-2.11)
-0.416** (-2.49)
-0.542*** (-6.12)
-0.179** (-2.29)
Annreti,t 0.146 (0.21)
0.353 (0.41)
-0.989 (-0.85)
0.725 (0.73)
0.286 (0.41)
-1.171** (-2.40)
TIMELINESSi,t -0.544*** (-3.18)
-0.594** (-2.28)
-0.604** (-1.98)
-0.666*** (-3.54)
-0.251* (-1.91)
-0.323** (-2.11)
λ2 values of J Test 3.531 3.306 3.748 3.115 0.608 4.366 Degree of freedom 2 2 2 2 2 2 p-value of J Test 0.171 0.191 0.153 0.211 0.738 0.112 N 7,489 9,268 4,047 16,975 8,464 16,948
Note: *** , ** and * denote significance at the 1, 5, and 10% levels, respectively. Coefficient estimates are reported with t-statistics in parentheses.
45
Table 6. Additional Test: The Value of Accounting Conservatism in Capital Investment
This table reports the valuation test using the following model.
, 0 1 , 2 , , 3 , 4 ,
5 , 6 , 1 7 , 8 , 9 , 10 , 1
11 , 1 , ,
*
*
i t i t i t i t i t i t
i t i t i t i t i t i t
i t i t i t
ret TIMELINESS TIMELINESS I I Div
Cash Cash Earn NetFin Int Lev
Lev Cash YFE IFE
The dependent variable reti,t is stock return over fiscal year minus the return on a benchmark portfolio. The benchmark portfolios are 25 Fama-French value-weighted portfolios. β2 is the coefficient we are interested in. All other variables are defined as in Appendix I. The explanatory variables except leverage are standardized by lagged market equity. Regressions are estimated using OLS. Statistical significance is computed using heteroskedasticity and autocorrelation robust standard errors that are clustered at the firm level.
Coefficient t-stat
TIMELINESSi,t-1 -0.015 (-0.46)
ΔIi,t *TIMELINESSi,t-1 0.025** (2.41)
ΔIi,t 0.817*** (3.87)
ΔCashi,t 2.088*** (7.12)
ΔIncomei,t 2.553*** (11.46)
ΔRDi,t 3.228*** (5.28)
ΔDividendi,t 0.771*** (4.18)
ΔInteresti,t -0.912** (-1.99)
Cashi,t-1 0.939*** (8.25)
Levi,t 0.059 (0.37)
Netfinancei,t -0.057* (-1.72)
Cashi,t-1*ΔCashi,t -0.188*** (-2.69)
Levi,t*ΔCashi,t 0.325 (0.78)
N 40,312
Adjusted-Rsquare 0.088
Note: *** , ** and * denote significance at the 1, 5, and 10% levels, respectively. Coefficient estimates are reported with t-statistics in parentheses.
46
Table 7. Additional Test: The Value of Accounting Conservatism in Capital Investment When Ex Ante Agency Cost Vary
This table reports the valuation test using the following model.
, 0 1 , 2 , , 3 , 4 ,
5 , 6 , 1 7 , 8 , 9 , 10 , 1
11 , 1 , ,
*
*
i t i t i t i t i t i t
i t i t i t i t i t i t
i t i t i t
ret TIMELINESS TIMELINESS I I Div
Cash Cash Earn NetFin Int Lev
Lev Cash YFE IFE
The dependent variable reti,t is stock return over fiscal year minus the return on a benchmark portfolio. The benchmark portfolios are 25 Fama-French value-weighted portfolios. β2 is the coefficient we are interested in. All other variables are defined as in Appendix I. All explanatory variables except leverage are standardized by lagged market equity. Regressions are estimated using OLS. Statistical significance is computed using heteroskedasticity and autocorrelation robust standard errors that are clustered at the firm level.
Note: *** , ** and * denote significance at the 1, 5, and 10% levels, respectively. Coefficient estimates are reported with t-statistics in parentheses.