Earnings Quality and Future Capital Investment: Evidence from

Discretionary Accruals

Kevin Ke Li
Haas School of Business
University of California at Berkeley
Phone: 510-643-1411
Email: kli@haas.berkeley.edu

Vicki Wei Tang*

McDonough School of Business
Georgetown University
Phone: 202-687-3773
Email: wt29@georgetown.edu

March 2008

* Corresponding author.

We would like to thank Bill Baber, Huafeng Chen, Guojin Gong, Prem Jain, Lee
Pinkowitz, Sundaresh Ramnath, Richard Sweeney and participants at the accounting-
finance joint workshop at Georgetown University and American University for their
comments and discussions.
 Earnings Quality and Future Capital Investment: Evidence from
Discretionary Accruals

Abstract: This paper examines how one aspect of earnings quality - discretionary
accruals - affects subsequent capital investment pattern and efficiency. We find that,
conditional on investment opportunities, investment in fixed assets in period t is less
sensitive to internal cash flows for firms with large positive discretionary accruals in
period t-1. We also find that, at a given level of capital investment in fixed assets in
period t, the return on assets in period t +1 is lower for firms with large positive
discretionary accruals in period t-1. The overall evidence suggests that firms with large
positive discretionary accruals misallocate resources, and thus, impose dead-weight
efficiency loss.

Keywords: earnings quality; discretionary accruals; capital investment; investment

efficiency

1
 Earnings Quality and Future Capital Investment: Evidence from
Discretionary Accruals

1. Introduction

Despite its fundamental importance, little empirical research investigates how

earnings quality affects capital investment decisions with the notable exception of Biddle

and Hilary (2006). They find that investment is more sensitive to internal cash flows for

firms with lower accrual quality as defined in Dechow and Dichev (2002), for which the

external cost of capital is higher (e.g., Francis et al., 2005).

This paper examines how another aspect of earnings quality - discretionary

accruals - affects capital investment pattern and efficiency in the subsequent period. This

paper is motivated by the mixed evidence on the market pricing of large positive

discretionary accruals. On one hand, firms with large unsigned discretionary accruals,

both positive and negative, have a higher external cost of capital (e.g., Francis et al.,

2005). On the other hand, firms with large positive discretionary accruals have a lower

stock return in the future, suggesting that firms with large positive discretionary accruals

have a lower external cost of capital at the time of portfolio formation (e.g., Xie, 2001;

Thomas and Zhang, 2002; Defond and Park, 2001). A higher external cost of capital

implies that firms with large positive discretionary accruals have a stronger incentive to

avoid external financing. Thus, capital investment is expected to be more sensitive to the

availability of internally generated cash flows. However, a lower cost of capital implies

that firms with large positive discretionary accruals are able to raise external capital at a

relatively low cost. Thus, capital investment is expected to be less sensitive to internally

generated cash flows.

1
 Using the data from 1988 to 2005, we find that, conditional on investment

opportunities, capital investment in fixed assets in period t is less sensitive to internal

cash flows for firms with large positive discretionary accruals in period t-1. We also find

that, at a given level of capital investment in period t, firms with large positive

discretionary accruals in period t-1 have a lower return on assets in period t+1. As

accounting performance following investment decisions is expected to be worse if a firm

misallocates resources, the result is consistent with the interpretation that firms with large

positive discretionary accruals misallocate resources to fixed assets in the subsequent

period.

The higher market valuation for firms with large positive discretionary accruals at

the time of portfolio formation is either due to investors’ information processing bias

(e.g., Hirshleifer and Teoh, 2003) or due to investors’ uncertainty about managerial

reporting incentives (e.g., Fischer and Verrecchia, 2000). To assess the validity of those

two alternative explanations, we examine whether the market pricing of large positive

discretionary accruals is related to external financing in period t. We find that, given the

level of discretionary accruals, the stock return in period t is lower for firms that raise

more external financing in period t, suggesting that the market valuation of discretionary

accruals is largely due to investors’ uncertainty about managerial reporting incentives.

The findings in this paper suggest that firms with large positive discretionary

accruals misallocate resources, and thus, impose dead-weight efficiency loss. The

findings have important implications for three strands of literature. First, a negative

association between earnings quality and cost of capital is usually interpreted as evidence

that lower earnings quality increases the information asymmetry component of cost of

2
capital (e.g., Francis et al., 2004). However, our findings suggests an alternative channel

to explain the negative association: firms with lower earnings quality make less efficient

investment decisions, which, in turn, adversely impacts future cash flows. Second, in

combination with Biddle and Hilary (2006), our findings suggest that a general statement

about earnings quality and future investment patterns is not valid unless a specific

attribute of earnings quality is specified. This is because the capital market prices

different attributes of earnings quality differently. Finally, our findings indicate that the

higher market valuation for firms with large positive discretionary accruals at the time of

portfolio formation is largely due to investors’ uncertainty about managerial reporting

incentives.

The paper is organized as the following. Section 2 discusses related literature.

Section 3 develops the hypothesis. Section 4 describes variable measurement and

research design. Section 5 presents the main empirical results on the association between

discretionary accruals and future investment pattern and efficiency. Section 6 presents

the robustness checks. Section 7 concludes the paper.

II. Related literature

2.1. Literature on firm investment

Neoclassical model of investment predicts that a firm will invest up to a level

where the marginal cost of investing is equal to managerial assessment of the marginal

profitability of capital (e.g., Lucas and Prescott, 1971; Mussa, 1977). Thus, firm

investment should only increase with its underlying investment opportunities as measured

by Tobin’s Q (e.g., Tobin, 1969). However, the misalignment of managerial and

3
shareholders incentives (e.g., Jensen and Meckling, 1976; Jensen, 1986) and asymmetric

information between corporate insiders and outside investors (e.g., Myers and Majluf,

1984) could cause investment to vary with internal cash flows. Under the agency view,

managers over-invest to reap private benefits such as “perks”, large empire and

entrenchment. Because external capital market disciplines managers from pursuing self-

interested investment, an influx of internal cash flows enables managers to invest more.

Under asymmetric information, managers in the interest of existing shareholders restrict

themselves from costly external financing, and thus investment is expected to increase

with internal cash flows.

A substantial body of empirical studies provides evidence consistent with that

corporate investment increases with internal cash flows (e.g., Blanchard, Lopez-de-

Silanes and Shleifer, 1994). Furthermore, starting with Fazzari, Hubbard and Peterson

(1988), a few studies find that firm investment is more sensitive to internal cash flows

for firms which are a priori classified as more financially constrained (e.g., Hoshi et al,

1991; Hubbard, Kashyap, and Whited, 1995). Based on the argument that managers are

more likely to forego positive NPV projects when firms are more financially constrained,

Biddle and Hilary (2006) interpret a higher sensitivity of investment to internal cash

flows as evidence of lower investment efficiency. Some recent papers, however, have

questioned the usefulness of the sensitivity of investment to cash flows in assessing the

effect of market frictions (e.g., Kaplan and Zingales, 1997 and Cleary, 1999). Given the

controversy on the sensitivity of investment to cash flows as a measure of investment

efficiency, we use accounting performance following investment decisions to gauge the

efficiency of investment decisions in the latter analysis.

4
 All the above studies share the common assumption that the capital market is

efficient with respect to either disciplining managers or incorporating all available

information. Another strand of literature investigates whether inefficient capital markets

affect corporate investment decisions. These studies use either time-series or cross-

sectional approach to examine whether non-fundamental noise in market valuation

impacts investment decisions (e.g., Barro, 1990; Morck, Shleifer and Vishny, 1990;

Blanchard, Rhee and Summers, 1993; Chrinko and Schaller, 2001; Baker, Stein and

Wurgler, 2003; Polk and Sapienza, 2006).

The empirical evidence from time-series approach, however, is rather mixed. On

one hand, Blanchard et al. (1993) uses the time-series economy-wide data in the U.S. and

concludes that stock market valuation appears to play a limited role, given fundamentals,

in the determination of investment decisions. On the other hand, Chirinko and Schaller

(2001) uses economy-wide data in Japan and concludes that, given structural assumption

about the fundamental and non-fundamental components of stock valuation, the market

bubble has a significant effect on investment in fixed assets.

Baker et al. (2003) find that stock valuation is relevant for contemporaneous

investment decisions for equity-dependent firms. Polk and Sapienza (2006) find that

managers cater to investor sentiment by investing more (less) when market valuation is

above (below) fundamentals, suggesting that non-fundamental noise in stock valuation is

relevant for investment decisions. While related, this paper, nevertheless, differs with

those two papers in two important aspects. First, this paper is based on the explicit

premise that market valuation is endogenous to earnings quality, while those papers

assumes that market valuation is exogenous. Thus, we examine the impact of large

5
positive discretionary accruals on subsequent investment decisions. Second, in contrast

to the level of investment, we are interested in the sensitivity of investment to internal

cash flows.

2.2. Literature on market valuation of discretionary accruals

Empirical evidence on the market pricing of large positive discretionary accruals

is mixed. On one hand, Francis at al. (2005) finds that firms with large unsigned

discretionary accruals, both positive and negative, have a higher cost of capital in the

long run. On the other hand, starting with Sloan (1996), voluminous studies document

that, conditional on the level of cash flows, firms with large positive discretionary

accruals have a lower stock return in the future, suggesting that those firms have a lower

cost of capital than otherwise similar firms at the time of portfolio formation (e.g., Xie,

2001; Thomas and Zhang, 2002; Defond and Park, 2001).1

Broadly speaking, there are two alternative theories to explain why firms with

large positive discretionary accruals have a higher market price at the time of portfolio

formation. Hirshleifer and Teoh (2003) suggest that it is due to investors’ information

processing bias, while Fischer and Verrecchia (2000) suggest that it is due to investors’

uncertainty about managerial reporting incentives. We are a priori uncertain about the

validity of those two theories, and thus, use a setting where investors are likely certain

about managerial reporting incentives to test the two explanations.

1
The reversal of stock prices for firms with large positive discretionary accruals is consistent with many
explanations, such as investors’ naïve fixation on reported earnings, trading frictions, survival bias, risk
estimation, and agency theory of overvalued equity (e.g., Khan, 2008; Mashruwala, Rajgopal, and Shevlin,
2006; Kraft, Leone and Wasley, 2006 and Kothari et al., 2006).

6
2.3. Other related studies

This paper is related to Fairfield et al. (2003) and Zhang (2007), which document

that firms with higher accruals tend to be growth firm, and thus, invest more in the future.

Their results are limited to the level of capital investment and are not interested in the

sensitivity of investment to internal cash flows. Our results are robust while explicitly

controlling for investment opportunities and correcting measurement error in Tobin’s Q.

This paper is also related to “real” earnings management literature. Stein (1989)

develops a model in which, faced with short-term market pressure, managers may forsake

good investments, such as cut R&D and advertising spending, to boost contemporaneous

earnings. Evidence in Bushee (1998) and Roychowdhury (2006) supports such

conjecture. Our paper differs in that we investigate the impact of earnings quality on

subsequent investment decisions.

Finally, this paper is related to Kothari, Louskina and Nikolaev (2006). They

document an unusually high level of capital expenditure prior to and during the year of

high discretionary accruals. Their results are not mutually exclusive from our results.

While their focus is on investment leading up to the formation of discretionary accruals

portfolio, our focus is on investment following the formation period.

III. Hypothesis development

Prior studies provided mixed evidence on the market pricing of discretionary

accruals, especially large positive discretionary accruals. Francis et al. (2005) find that

firms with large positive discretionary accruals have a higher cost of capital in the long

run, while a substantial body of studies find that firms with large positive discretionary

7
accruals have a lower stock return in the future, suggesting that firms with large positive

discretionary accruals have a lower cost of capital in the short run (e.g., Xie, 2001;

Thomas and Zhang, 2002; Defond and Park, 2001). A higher external cost of capital

implies that firms with large positive discretionary accruals have a stronger incentive to

avoid external financing. Thus, capital investment is expected to be more sensitive to the

availability of internally generated cash flows. However, a lower cost of capital implies

that firms with large positive discretionary accruals are able to raise external capital at a

relatively low cost. Thus, capital investment is expected to be less sensitive to internally

generated cash flows. We are a priori uncertain about how discretionary accruals impacts

future capital investment pattern. This leads to the first hypothesis stated in the null:

H1: Conditional on investment opportunities, the sensitivity of investment to internal cash

flows in period t is unrelated with discretionary accruals in period t-1.

Liang and Wen (2007) hypothesizes that more noise in accounting earnings

always leads to less efficient capital investment decisions, either in the form of

overinvestment or in the form of underinvestment, because more accounting noise leads

to more mispricing in the capital market. Based on their theoretical prediction, capital

investment decisions in firms with large positive discretionary accruals are expected to be

less efficient. Given the controversy on the sensitivity of investment to cash flows as a

measure of investment efficiency, we use accounting performance following investment

decisions to gauge the efficiency of investment decisions. If a firm misallocates

resources to fixed assets, accounting performance following investment decisions is

8
expected to be worse than firms that allocate resources properly to fixed assets. We

expect that, at a given level of capital investment in period t, firms with large positive

discretionary accruals in period t-1 have a lower return on assets in period t+1. This

leads to the second hypothesis stated in the null:

H2: At a given level of capital investment in period t, return on assets in period t +1 is

not lower for firms with large positive discretionary accruals in period t-1.

Finally, we examine whether the market pricing of large positive discretionary

accruals is related to subsequent external financing. We choose subsequent external

financing because of two reasons. First, the two alternative theories yield different

predictions on the market pricing of large positive discretionary accruals when investors

are aware of the associated reporting incentives. Second, prior studies find that firms

have large positive discretionary accruals prior to equity issuances (e.g., Teoh et al.,

1998a and 1998b).

If the pricing of discretionary accruals in the period of portfolio formation is due

to investors’ information processing bias, the valuation of discretionary accruals in period

t-1 is not expected to change with the status of external financing. However, if it is due

to investors’ uncertainty about the managerial reporting incentives, upon news of external

financing transactions, investors realize the reporting incentives for large positive

discretionary accruals and the valuation of discretionary accruals is adjusted accordingly.

We are a priori uncertain about which theory explains the market pricing of discretionary

accruals better. This leads to the third hypothesis stated in the null:

9
H3: The market pricing of large positive discretionary accruals is unrelated with the

amount of external financing in the subsequent period.

IV. Measurement and Research Design

4.1. Measure of discretionary accruals

Controlling for profitability is important in examining investment decisions. One

measure that controls for profitability is the performance-adjusted discretionary accruals

measure in Kothari, Leone, and Wasley (2005). We calculate discretionary accruals from

the following model, following Teoh et al. (1998a and 1998b):

CAit = α + β (ΔSalesit - ΔARit) + εit

where CAit is current accruals for firm i in year t, ΔSalesit (Compustat item Δ12) is the

annual change in sales for firm i in year t, and ΔARit (Compustat item Δ151) is the annual

change in accounts receivable for firm i in year t. Current accruals is the change in non-

cash current assets minus the change in current operating liabilities (Compustat item Δ(4-

1) - Δ(5-34-71)). Variables are scaled by total assets in period t-1 (Compustat item 6).

The regression residual εit captures the unexpected change in working capitals for firm i

in year t given the change in cash sales.

Next we assign each firm to a portfolio based industry membership and return on

assets.2 Portfolios are constructed by matching firms into return on asset (ROA) quartiles

and industry membership, where ROA is net income (Compustat item 172) deflated by

beginning book value of assets (Compustat item 6). The performance-adjusted

2
We use Fama-French 48 industry classification rather than 4-digit SIC codes to ensure a reasonable
number of observations for each ROA-industry portfolio.

10
discretionary accrual for firm i in year t (DAit) is the firm-year residual εit less the mean

residual for the corresponding portfolio excluding firm i.

Finally, we partition firms into deciles based on the magnitude of performance-

adjusted discretionary accruals (DAit). We identify firms in the top decile of

performance-adjusted discretionary accruals as firms with large positive discretionary

accruals. The indicator variable MAX_DAit is coded as 1 if firm i is in the top decile of

performance-adjusted discretional accruals in period t and 0 otherwise.

4.2. Research design on future investment patterns

In order to assess whether subsequent capital investment for firms with large

positive discretionary accruals is different from other firms, we use the following model:

INVESTMENTit = α + β1Qit-1 + β2CASHit-1 + β3MAX_DAit-1 + β4 Qit-1*MAX_DAit-1

+ β5 CASHit-1 *MAX_DAit-1 + INDUSTRY DUMMIES + YEAR DUMMIES + εit

We explicitly model capital investment in fixed assets in period t as a function of

investment opportunities at the beginning of the period (Qit-1) and internal cash flows

(CASHit-1) at the beginning of the period. We use Q and CASH at the beginning of the

period because the bulk of investment decisions are made at the beginning of the period

rather than at the end of period.3 We also include Fama-French 48 industry indicators

and year indicators in the model to industry factors or macro-economic conditions. The

indicator variable for large positive discretionary accruals in period t-1 (MAX_DAit-1) and

its interactions with investment opportunities and internal cash flows are added to the

3
If we use investment opportunities and internal cash flows at the end of the period as the explanatory
variables, the results are similar to those reported in the tables.

11
baseline model to test whether discretionary accruals impacts subsequent capital

investment.

INVESTMENTit is capital expenditure (Compustat item 128) scaled by beginning-

of-year book value of assets (Compustat item 6). Qit-1 is the ratio of the market value of

the company to the book value of the company at the beginning of period t. The market

value of the company is the sum of the market value of the equity (Compustat item

25*199), the value of short-term debt (Compustat item 9) and the value of long-term debt

(Compustat item 34). The book value of the company is the book value of assets

(Compustat item 6). CASHit-1 is measured as the income before extraordinary items

(Compustat item 18) plus depreciation and amortization expenses (Compustat item 14)

scaled by beginning-of-year book value of assets (Compustat item 6) at the beginning of

period t.

The interaction term β5 between MAX_DAit-1 and CASHit-1 is the variable of

interest. A negative (positive) coefficient on the interaction is consistent with the

interpretation that capital investment is less (more) sensitive to internal cash flows for

firms with large positive discretionary accruals. However, if Tobin’s Q, as a proxy of

investment opportunities, contains more measurement error for firms with large positive

discretionary accruals, it can also result in a higher sensitivity of investment to internal

cash for those firms because internal cash flow also reflects information about investment

opportunities (e.g., Poterba, 1988). Erickson and Whited (2000) suggest the generalized

method of moments (GMM) estimation yields measurement-error-consistent results. To

control for potential measurement errors in Tobin’s Q as a proxy for investment

opportunities, we use GMM to estimate the model. Finally, we use firm-fixed effect

12
model to assess whether the time-series variation in discretionary accruals has an impact

on time-series variation in subsequent capital investment.

4.3. Research design on future investment efficiency

If a firm misallocates resources to fixed assets, accounting performance following

investment decisions is expected to be worse than firms that allocate resources properly

to fixed assets. We use the accounting performance following investment decisions to

gauge the efficiency of investment decisions for firms with large positive discretionary

accruals. More specifically, we use the following model:

ROAit+1=α+β1ACCRUALSit+β2CASHit+β3INVESTMENTit+β4MAX_DAit-1+β5INVESTMENTit*MAX_DAit-1+εit

It is not valid to draw conclusions about the efficiency of investment decisions

from a univariate comparison of return on assets in period t+1 between firms with large

positive discretionary accruals and other firms. This is due to a number of confounding

factors. First, large positive discretionary accruals in period t (DAit) reverse in period

t+1. ACCRUALSit is the difference between earnings (Compustat item 172) and cash

flows from operating activities (Compustat item 308). Second, return on asset in period t

is positively correlated with return on assets in period t+1. Third, the reversal of large

positive discretionary accruals in period t-1 implies a lower return on assets in period t+1,

even in the absence of inefficient investment policies. Finally, return on assets in period

t+1 is negatively associated with investment in period t because of declining marginal

return to investment (e.g. Titman 2004). We control for all those confounding factors in

the model.

13
 Our variable of interest is the coefficient (β5) on the interaction between

investment level in period t and large positive discretionary accruals in period t-1. A

negative coefficient on the interaction indicates that firms with large positive

discretionary accruals have a worse return on assets in period t+1, at a given level of

capital investment in period t. A worse accounting performance following investment

decisions suggests that firms with large discretionary accruals misallocate resources to

fixed assets.

4.4. Market pricing of large positive discretionary accruals and external financing

We use the following model to examine whether the market pricing of large

positive discretionary accruals is related to external financing in the subsequent period:

BHARit=α + β1DAit-1 + β2CASHit-1 + β3XFINit + β4DAit-1 *XFINit + εit

The dependent variable is the buy-and-hold size-adjusted stock return in period t

(BHARit). The explanatory variables include discretionary accruals in period t-1, internal

cash flows in period t-1, external financing in period t (XFINit) and the interaction

between discretionary accruals in period t-1 and external financing in period t. External

financing (XFINit) includes both the debt issued and equity raised for a particular year.

Following Bradshaw, Richardson and Sloan (2006), we use information provided in the

statement of cash flows to derive a measure of external finance. More specifically,

XFINit is the sum of net cash inflows from the equity issues (Compustat item 108-115-

127), net cash inflows from the long-term debt issues (Compustat item 111-114) and cash

14
flows from the change in current debt (Compustat item 301), scaled by the beginning-of-

year book value of assets (Compustat item 6).

We include external financing in period t as a main effect because prior studies

suggest that the capital market reacts to news of external financing activities. However,

in this context, the variable of interest is the interaction between and DAit-1 and XFINit.

An insignificant interaction suggests that the pricing of discretionary accruals is unrelated

to external financing in period t. A significant negative interaction, however, suggests

that the pricing of discretionary accruals is related to external financing, which is

consistent with the interpretation that the pricing of discretionary accruals in the period of

portfolio formation is largely due to investors’ uncertainty about managerial reporting

incentives.

V. Empirical results

5.1. Descriptive statistics

The final sample to test the first (second) hypothesis consists of 60,728 (58,306)

firm-year observations from 1988 to 2005 where information on all variables is available

on the COMPUSTAT annual file. The data start from 1988 because statements of cash

flows are widely available after 1988. All variables, both dependent and independent

variables, are winsorized at both the top and the bottom one percentile. Firms in the

financial and utility industries are excluded. Table 1 reports descriptive statistics. The

magnitudes of most variables are in line with prior studies. For example, the median of

return on assets is around 3.8% and the median Tobin’s Q is 1.15. On average, capital

15
expenditures amount to 7.8% of total assets and all external financing activities, including

both equity issues and debt issues, provide 7.3% of total assets.

Table 1 also reports the descriptive statistics on the comparison between firms

with large positive discretionary accruals and all other firms. The median return on

assets in period t-1 is not statistically different from each other for the two groups of

firms, suggesting that performance-adjusted discretionary accruals virtually control for

the underlying profitability perfectly. Firms with large positive discretionary accruals in

period t-1 have a lower level of internally generated cash flows and a slightly higher

Tobin’s Q in period t-1. While firms with large positive discretionary accruals raise

more external capital, which is consistent with Teoh et al. (1998a, 1998b), those firms

invest less in fixed assets in period t. On a univariate basis, firms with large positive

discretionary accruals in period t-1 have a lower return on assets in period t+1.

5.2. Empirical results on investment pattern and efficiency

Panel A of Table 2 reports the correlation among all variables to test future

investment pattern. Capital investment in fixed assets increases with underlying

investment opportunities, as evident in the positive correlations between INVESTMENTt

and Qt-1. Capital investment in fixed assets also increases with the availability of

internally generated cash flows, as evident in the positive correlations between

INVESTMENTt and CASHt-1.

Table 3 reports the results on discretionary accruals and future investment-cash

flow sensitivity. Per Column 1, in the base model, capital investment increases with both

investment opportunities and internally generated cash flows, consistent with prior

16
literature. Column 2 reports the results under OLS specification. The interaction

between MAX_DAt-1 and CASHt-1 is -0.022 and is statistically significant (p-value =

0.005), which suggests that capital investment in firms with large positive discretionary

accruals is less sensitive to internal cash flows. Column 3 reports the results using

measurement-error consistent GMM estimation. The coefficient on the interaction

between discretionary accruals and internal cash flows is -0.022 (p-value = 0.015), which

is virtually identical to the results from the OLS estimation. Column 4 adds firm-fixed

effect and obtains virtually identical results as those without firm-fixed effect, suggesting

a possibility of over-identification. As the results are similar using different estimation

methods, we only report the GMM results for the remainder of the paper. The empirical

evidence in Table 3 indicates that investment is less sensitive to internal cash flows for

firms with large positive discretionary accruals.

Figure 1 presents the results in a graphical manner. We partition the sample into

two groups: firms with large positive discretionary accruals in period t-1 versus other

firms. We sort firm-year observations into 50 portfolios based on the value of internal

cash flows in period t-1 within each group. Then we plot the mean value of

INVESTMENTt and CASHt-1. It is obvious that firms with large positive discretionary

accruals in period t-1 have a flatter slope, i.e., a lower sensitivity of investment to internal

cash flows.

Panel B of Table 2 reports the correlation among all variables to test efficiency of

investment decisions. Consistent with prior studies, firms with higher accruals and

profitability in period t have a higher return on assets in period t+1. Consistent with the

17
reversal of discretionary accruals, firms with large positive discretionary accruals in

period t have a lower return on assets in period t+1.

Table 4 reports the results on efficiency of investment decisions. The baseline

model includes accruals and internal cash flows in period t. We observe a positive

coefficient on internal cash flows and a negative coefficient for accruals, which suggest

that, given the same profitability, firms with higher accruals have a lower return on assets

in the future. In the second column, the coefficient on investment level in period t is

significantly negative (coefficient = -0.207 and p-value = 0.001). Column 3 reports the

result on the efficiency of investment decisions. The negative main effect on MAX_DAt-1

is consistent with the reversal of discretionary accruals in the future. The variable of

interest, the coefficient on the interaction between MAX_DAt-1 and INVESTMENTt, is

significantly negative (coefficient = -0.046 and p-value = 0.068). As accounting

performance following investment decisions is expected to be worse if a firm

misallocates resources, the negative coefficient is consistent with the interpretation that

firms with large positive discretionary accruals misallocate resources to fixed assets.

Figure 2 presents the results in a graphical manner. We partition the sample into

two groups: firms with large positive discretionary accruals in period t-1 versus other

firms. We sort firm-year observations into 50 portfolios based on the value of investment

in period t within each group. Then we plot the mean value of INVESTMENTt and

ROAt+1. It is obvious that, given the same level of investment, return on assets in period

t+1 is consistently lower for firms with large positive discretionary accruals in period t-1.

5.3. Market pricing of large positive discretionary accruals in period t

18
 Table 5 presents the results on whether market pricing of large positive

discretionary accruals is related to subsequent external financing. The coefficient on the

interaction between DAt-1 and XFINt is significantly negative (coefficient = -1.565 and p-

value = 0.011). The negative coefficient indicates that, given the same level of

discretionary accruals in period t-1, the size-adjusted stock return in period t is

significantly lower for firms that raise more external capital in period t. The evidence is

consistent with the interpretation that market pricing of discretionary accruals in the

period of portfolio formation is largely due to investors’ uncertainty about managerial

reporting incentives.

VI. Robustness Checks

6.1. Large positive discretionary accruals and external financing

A lower cost of capital implies that, ceteris paribus, firms with large positive

discretionary accruals also raise more external capital in period t, which is consistent with

prior findings in Teoh et al. (1998). Although retained earnings, internal financing, rather

than equity issuance are by far the largest source of funds for capital investment (e.g.,

Froot, Scharfstein, and Stein, 1994; Mayer and Sussman, 2003), it is necessary to assess

the extent to which our main results are driven by the association between large positive

discretionary accruals and subsequent external financing. First, we explicitly control for

external financing in period t and its interactions with Qt-1 and CASHt-1. Second, we

examine whether firms with large positive discretionary accruals in period t-1 and

negative external financing in period t also have a lower sensitivity of investment to

internal cash flows.

19
 Table 6 Panel A Column 1 and Column 2 indicates that the coefficient on the

interaction between large positive discretionary accruals and internal cash flows

continues to be significantly negative. The coefficient (p-value) is -0.037 (0.001) after

controlling for the level of external financing. The coefficient on the interaction is -0.040

(0.001) after controlling for both the level of external financing and its interactions with

investment opportunities and internal cash flows. Per Table 6 Panel A Column 3, firms

with large positive discretionary accruals in period t-1 and negative external financing in

period t also have a lower sensitivity of investment to internal cash flows, as evident in

the statistically negative coefficient on the interaction between MAX_DAt-1 and CASHt-1.

The overall evidence suggests that our main empirical results are not solely driven by the

association between large positive discretionary accruals and more external financing in

the next period.

6.2. Characteristics for firms with large positive discretionary accruals

Although we document that our results is not solely driven by the association

between large positive discretionary accruals and external financing in period t, other

managerial reporting incentives also lead to large positive discretionary accruals. For

example, Kothari et al. (2006) suggest that firms that are overvalued in the past have

incentives to manage earnings upward to prolong the overvaluation. Managers have

incentives to use positive discretionary accruals to avoid debt covenant violation

constraints (e.g., Defond and Jimbalvo, 1994; Sweeney, 1994). Managers also have

incentives to use positive discretionary accruals to profit from their trading (e.g., McVay,

Nagar and Tang, 2006). In order to alleviate the concern that the empirical results are

20
driven by various managerial incentives, we include those variables and their interactions

with internal cash flows as additional explanatory variables in the model.

First, following Kothari et al. (2006), we use the buy-and-hold size-adjusted

return from period t-4 to in period t-1 (BHARt-4,t-1) as the proxy for prior overvaluation.

Buy-and-hold size adjusted return is raw return for the period minus the raw return for

the matching decile from CRSP database. We also include its interaction terms with

investment opportunity (Qt-1) and internal cash flow (CASHt-1). Table 6 Panel B Column

1 reports the result on the investment-cash flow sensitivity test after controlling for

BHARt-4,t-1. The interaction between MAX_DAt-1 and CASHt-1 is still negative and

statistically significant (coefficient = -0.039 and p-value = 0.001).

Second, we use the leverage in period t-1 as the proxy for the potential cost of

violating debt covenants. A higher cost of violating debt covenants implies a stronger

incentive to use positive discretionary accruals to avoid debt covenant violation.

Leverage is the ratio of debt to total assets. We also include its interaction terms with

investment opportunity (Qt-1) and internal cash flow (CASHt-1). Table 6 Panel B Column

2 reports the result on the investment-cash flow sensitivity test after controlling for

LEVERAGEt-1. The interaction between MAX_DAt-1 and CASHt-1 is still negative and

statistically significant (coefficient = -0.021 and p-value = 0.005).

Third, to alleviate the concern that our results are driven by managerial incentives

to profit from trading in period t, we include CEO’s realized net purchase of the

company’s stock in period t (CEOBUYt) as a proxy for CEO’s intention to trade in period

t. We also include its interaction terms with investment opportunity (Qt-1) and internal

21
cash flow (CASHt-1). Following McVay, Nagar and Tang (2006), we scale CEO net

transactions by CEO’s holdings:

CEOBUYi t= [∑SPh/SHh - ∑SSh/SHh]

where SPh, SSh, and SHh are shares purchased in open market, sold in open market and

held by CEO before each transaction h for firm i in year t. CEO trading data are obtained

from Thomson Financial. Table 6 Panel B Column 3 reports the results of investment-

cash flow sensitivity test after controlling for CEOBUYt. The interaction between

MAX_DAt-1 and CASHt-1 is negative and statistically significant (coefficient = -0.022 and

p-value = 0.005).

Finally, we include all those variables and their interaction terms in the model. Per

Table 6 Panel B Column 4, the coefficient on the interaction between MAX_DAt-1 and

CASHt-1 is -0.040 and statistically significant (p-value = 0.001). Overall, our main

results are robust to managerial incentives to report large positive discretionary accruals.

6.3. Measurement issue in the proxy for internal cash flows

Following prior literature, we measure internal cash flow (CASH) as net income

before extraordinary items plus depreciation. However, Bushman, Smith and Zhang

(2005) notice that this measure of internal cash flows can be disaggregated into cash

flows from operations (CFOs) and non-cash accounting components. To avoid making

erroneous inferences on the sensitivity of investment to cash flows, we also use cash flow

from operations (CFOs) as an alternative measure of internal cash flows. The empirical

results are presented in Table 6 Panel C and are largely consistent with the main results.

22
For example, the coefficient on the interaction between MAX_DAt-1 and CFOt-1 is -0.071

with a p-value of 0.001.

6.4. Discretionary accruals in period t

Polk and Sapienza (2006) document a positive association between discretionary

accruals and contemporaneous level of investment in fixed assets. As large discretionary

accruals in period t-1 reverse in period t, we examine the robustness of our results after

controlling for discretionary accruals in period t (DAt). Table 6 Panel D reports the

results. After controlling for discretionary accruals in period t, investment in fixed assets

is still less sensitive to internal cash flows for firms with large discretionary accruals in

period t-1 (coefficient = -0.019, p-value = 0.013). Interestingly, in contrast to the positive

association documented in Polk and Sapienza (2006), the main effect on

contemporaneous discretionary accruals is significantly negative (coefficient = -0.035

and p-value = 0.001) after controlling for large positive discretionary accruals in period t-

1. We also include the interaction terms of discretionary accruals in period t with

investment opportunity (Qt-1) and internal cash flow (CASHt-1), and the results are similar.

VII. Conclusions

This paper investigates how one aspect of earnings quality - discretionary accruals

- impacts future investment pattern and efficiency. We find that, conditional on

investment opportunities, investment in fixed assets in period t is less sensitive to internal

cash flows for firms with large positive discretionary accruals in period t-1. We also find

that, at a given level of capital investment in period t, firms with large positive

23
discretionary accruals in period t-1 have a lower return on assets in period t+1. As

accounting performance following investment decisions is expected to be worse if a firm

misallocates resources, the result is consistent with the interpretation that firms with large

positive discretionary accruals misallocate resources to fixed assets, thus imposes dead-

weight efficiency loss.

One of the potential areas for future research is to identify the specific channel(s)

through which large positive discretionary accruals result in less efficient investment

decisions. An exploration of such a channel facilitates identifying the nature of the

inefficient investment decisions: underinvestment or overinvestment for firms with large

positive discretionary accruals.

24
References
Baker, Malcolm, Jeremy Stein and Jeffery Wurgler, 2003, “When Does the Market
Matter? Stock Prices and the Investment of Equity-Dependent Firms,” Quarterly Journal
of Economics 118, 969-1005.

Barro, Robert J., 1990, “Government Spending in a Simple Model of Endogenous

Growth,” Journal of Political Economy 98(5), S103-S125.

Biddle, Gary and Gilles Hilary, 2006, "Accounting Quality and Firm-level Capital
Investment," The Accounting Review Vol. 81, No. 5, 963-982.

Blanchard, Olivier, Florencio Lopez-de-Silanes, and Andrei Sheifler, 1994, “What do

Firms do with Cash Windfalls?” Journal of Financial Economics 36, 337-360.

Blanchard, Olivier, Changyong Rhee, and Lawrence Summers, 1993, “The Stock Market,
Profit, and Investment,” Quarterly Journal of Economics 108, 115-136.

Bradshaw, Mark, Scott Richardson and Richard Sloan, 2006, “The relation between
corporate financing activities, analysts’ forecast and stock returns,” Journal of
Accounting and Economics 42: 53-85.

Bushee, Brian J., 1998, “The Influence of Institutional Investors on Myopic R&D
Investment Behavior,” The Accounting Review 73(3), 305-333

Bushman, Robert M., Abbie J. Smith, and Frank Zhang, 2005, “Investment-Cash Flow
Sensitivities are Really Investment-Investment Sensitivities,” Available at SSRN:
http://ssrn.com/abstract=842085

Chirinko, Robert S. and Huntley Schaller, 2001, “Business Fixed Investment and
‘Bubbles’: The Japanese Case,” American Economic Review 91(3), 663-680.

Cleary, Sean, 1999, ‘The Relationship between Firm Investment and Financial Status,”
Journal of Finance 54(2), 673-692.

Dechow, Patricia M. and Ilia D. Dichev, 2002, “The Quality of Accruals and Earnings:
The Role of Accrual Estimation Errors,” The Accounting Review 77 (Supplement), 35-59.

DeFond, Mark and Chul Park, 2001, “The Reversal of Abnormal Accruals and the
Market Valuation of Earnings Surprises,” The Accounting Review 76(3), 375-404.

DeFond, Mark and James Jimbalvo, 1994, “Debt covenant violation and manipulation of
accruals: Accounting choice in troubled companies,” Journal of Accounting and
Economics 17, 145–176.

Erickson, Timothy and Toni M. Whited, 2000, “Measurement error and the relationship
between investment and q,” Journal of Political Economy 108, 1027–1057.

25
Fairfield, Patricia M., J. Scott Whisenant, and Teri Lombardi Yohn, 2003, “Accrued
Earnings and Growth: Implications for Future Profitability and Market Mispricing,” The
Accounting Review 78 (1), 353-371.

Fazzari, Steven, R. Glenn Hubbard, and Bruce Peterson, 1988, “Financing Constraints
and Corporate Investment,” from Brookings Papers on Economic Activity 1, 141-195.

Fischer, Paul and Robert Verrecchia, 2000, “Reporting Bias,” The Accounting Review
75(2), 229-245.

Francis, Jennifer, Ryan LaFond, Per Olsson, and Katherine Schipper, 2004, “Costs of
Equity and Earnings Attributes,” The Accounting Review 79(4), 967-1010.

Francis, Jennifer, Ryan LaFond, Per Olsson, and Katherine Schipper, 2005, “The Market
Pricing of Accruals Quality,” Journal of Accounting and Economics 39(2), 295-327.

Froot, Kenneth A., David S. Scharfstein, and Jeremy C. Stein, 1994, “A new approach to
risk management,” Harvard Business Review (November-December) 91-102.

Hirshleifer, David and Siew Hong Teoh, 2003, “Limited Attention, Information
Disclosure, and Financial Reporting,” Journal of Accounting and Economics 36(1-3),
337-386.

Hoshi, Takeo, Anil Kashyap and David Scharfstein, 1991, “Corporate Structure,
Liquidity, and Investment: Evidence from Japanese Industrial Groups,” Quarterly
Journal of Business and Economics 106, 33-60.

Hubbard, R. Glenn, Anil K. Kashyap, and Toni M. Whited, 1995, “Internal Finance and
Firm Investment,” Journal of Money, Credit and Banking 27(3), 683-701.

Jensen, Michael C., and William H. Meckling, 1976, “Theory of the Firm: Managerial
Behavior, Agency Costs and Ownership Structure,” Journal of Financial Economics
3(4), 305-360.

Jensen, Michael C., 1986, “The Agency Costs of Free Cash Flow: Corporate Finance and
Takeovers,” American Economic Review 76(2), 323-329.

Kaplan, Steven, and Luigi Zingales, 1997, “Do Financing Constraints Explain Why
Investment Is Correlated with Cash Flow?" Quarterly Journal of Economics CXII, 169-
215.

Khan, Mozaffar, 2008, “Are accruals mispriced? Evidence from tests of an Intertemporal
Capital Asset Pricing Model,” Journal of Accounting and Economics 45, 55-77.

Kothari, S.P., Andrew Leone, and Charles Wasley, 2005, “Performance Matched
Discretionary Accrual Measures,” Journal of Accounting and Economics 39(1), 163-197.

26
Kothari, S.P., Elena Loutskina, and Valeri Nikolaev, 2006, “Agency Theory of
Overvalued Equity as an Explanation for the Accrual Anomaly,” Available at SSRN:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=871750

Kraft, Arthur, Andrew J. Leone, and Charles Wasley, 2006, “An Analysis of the Theories
and Explanations Offered for the Mispricing of Accruals and Accrual Components,”
Journal of Accounting Research 44(2), 297-339.

Lucas, Jr., Robert and Edward Prescott, 1971, “Investment Under Uncertainty,”
Econometrica 39(5), 659-681.

Liang, P. and Xiaoyan Wen, 2007, “Accounting Measurement Basis, Market Mispricing,
and Firm Investment Efficiency,” Journal of Accounting Research 45(1), 155-197.

Mashruwala, Christina A., Shivaram Rajgopal, and Terry J. Shevlin, 2006, “Why is the
Accrual Anomaly not Arbitraged Away? The Role of Idiosyncratic Risk and Transaction
Costs,” Journal of Accounting and Economics 42, 3-33.

Mayer, Colin and Oren Sussman, 2003, “A new test of capital structure,” Oxford
Financial Research Centre Working Paper, 2003-FE-16.

McVay, Sarah, Venky Nagar, and Vicki Wei Tang, 2006, “Trading incentives to meet
analyst forecast,” Review of Accounting Studies 11, 575-598.

Morck, Randall, Andrei Shleifer, and Robert Vishny, 1990, “The Stock Market and
Investment: Is the Market a Sideshow?” Brookings Papers on Economic Activity 1990(2),
157-215.

Myers, Stewart C. and, Nicholas S. Majluf, 1984, “Corporate financing and investment
decisions when firms have information that investors do not have,” Journal of Financial
Economics 13(2), p.187.

Mussa, Michael, 1977, “External and Internal Adjustment Costs and the Theory of
Aggregate and Firm Investment,” Economica 44(174) 163-178.

Poterba, James M., 1988, “'Comments and discussion' following 'Financing constraints
and corporate investment.'” Brookings Papers on Economic Activity I, 200-4

Polk, Christopher and Paola Sapienza, 2006, “The Stock Market and Corporate
Investment: A Test of Catering Theory,” Review of Financial Studies, forthcoming.

Rajan, Raghuram G. and Luigi Zingales, 1995, "What Do We Know about Capital
Structure? Some Evidence from International Data," Journal of Finance vol 50, pp 1421-
1460.

27
Roychowdhury, Sugata, 2006, “Earnings management through real activities
manipulation,” Journal of Accounting and Economics 42(3), 335-370.

Sloan, Richard, 1996, “Do stock prices fully reflect information in accruals and cash
flows about future earnings?” The Accounting Review 71, 289-315.

Stein, Jeremy, 1989, “Overreactions in the Options Market,” Journal of Finance 44,
1011-1023.

Sweeney, Amy P., 1994, “Debt covenant violations and managers accounting responses,”
Journal of Accounting and Economics 17, 281–308.

Teoh, Siew Hong, Ivo Welch, and T. J. Wong, 1998a, “Earnings management and the
underperformance of seasoned equity offerings,” Journal of Financial Economics 50, 63-
100.

Teoh, Siew Hong, Ivo Welch, and T. J. Wong, 1998b, “Earnings management and the
long-run market performance of initial public offerings,” Journal of Finance 53, 1935-
1974.

Thomas, Jacob and Hui Zhang, 2002, “Inventory Changes and Future Returns,” Review
of Accounting Studies 7, 163-187.

Titman, Sheridan, 2004, “Capital Investments And Stock Returns,” Journal of Financial
and Quantitative Analysis 39, 677-700.

Tobin, James, 1969, “A General Equilibrium Approach to Monetary Theory,” Journal of

Money, Credit and Banking 1(1), 15-29.

Xie, Hong, 2001, “The mispricing of abnormal accruals,” The Accounting Review 76(3),
357–373

Zhang, Frank, 2007, “Accruals, Investment, and the Accrual Anomaly,” Yale University
working paper.

28
 Figure 1
Investment-Cash Flows Sensitivity
by Discretionary Accruals

Figure 1 is constructed in the following steps. First, all observations are divided into two groups based on
MAX_DAt-1. Within each group, observations are evenly sorted into 50 portfolios based on the value of
CASHt-1. The mean values of INVESTMENTt and CASHt-1 in each portfolio are plotted.

29
 Figure 2
Future Return on Assets
by Discretionary Accruals

Figure 2 is constructed in the following steps. First, all observations are divided into two groups based on
MAX_DAt-1. Within each group, observations are evenly sorted into 50 portfolios based on the value of
INVESTMENTt. The mean values of ROAt+1 and INVESTMENTt in each portfolio are plotted.

30
 Table 1
Descriptive Statistics

Panel A: Descriptive statistics of variables used to test Hypothesis 1

Full Sample MAX_DAt-1 Others MAX_DAt-1 versus Others

Difference Difference
Variables N Mean Std Median Min Max Mean Median Mean Median in Mean p-value in Median p-value
DAt-1 60,728 0.000 0.076 0.000 -0.346 0.304 0.141 0.125 -0.016 -0.005 0.156 (0.001) 0.130 (0.001)
ROAt-1 60,728 0.004 0.178 0.038 -1.654 0.413 -0.001 0.035 0.005 0.038 -0.005 (0.059) -0.003 (0.539)
Qt-1 60,728 1.723 1.881 1.154 0.123 30.752 1.763 1.163 1.719 1.153 0.044 (0.100) 0.010 (0.084)
CASHt-1 60,728 0.062 0.172 0.088 -1.219 0.507 0.055 0.078 0.063 0.089 -0.008 (0.003) -0.011 (0.001)
INVESTMENTt 60,728 0.078 0.091 0.049 0.000 0.731 0.062 0.037 0.079 0.050 -0.017 (0.001) -0.013 (0.001)

Panel B: Descriptive statistics of variables used to test Hypothesis 2

Full Sample MAX_DAt-1 Others MAX_DAt-1 versus Others

Difference Difference
Variables N Mean Std Median Min Max Mean Median Mean Median in Mean p-value in Median p-value
ACCRUALSt 58,306 -0.066 0.134 -0.056 -1.307 0.402 -0.060 -0.049 -0.066 -0.056 0.006 (0.004) 0.007 (0.001)
CASHt 58,306 0.050 0.187 0.084 -1.289 0.504 0.015 0.057 0.053 0.086 -0.038 (0.001) -0.029 (0.001)
XFINt 58,306 0.073 0.284 0.001 -0.335 2.943 0.090 0.008 0.071 0.000 0.019 (0.001) 0.008 (0.001)
ROAt+1 58,306 -0.003 0.177 0.034 -1.215 0.394 -0.035 0.017 0.000 0.036 -0.034 (0.001) -0.019 (0.001)

31
 Table 1
(Continued)
New investment (INVESTMENT) is capital expenditure (Compustat item 128) deflated by beginning-of-year book value of assets (Compustat item 6).
Discretionary accruals is the residual from the regression CAit = α + β (ΔSalesit - ΔARit ) + εit , where CA is the current accrual of firm i in year t (Compustat
item Δ (4-1) - Δ (5-34-71)) , Δ Sales is the annual change in sales for firm i in year t, ΔAR is the annual change in accounts receivable for firm i in year t, where
all variables are scaled by beginning-of-year book value of total assets. The performance-adjusted discretionary accrual (DA) is the difference between the firm-
specific discretionary accruals and the average discretionary accrual of a portfolio of firms (excluding the sample firm itself) matched on Fama-French 48
industry classifications and current return on assets quartiles. MAX_DA is an indicator variable, which takes the value of 1 if a firm is in the highest quartile of
performance-adjusted discretionary accruals and 0 otherwise. Investment opportunities (Q) is measured as the ratio of the market value to the replacement cost
of the assets, where the market value is measured as the sum of the market value of the equity (Compustat item 25*199) and the value of debt (Compustat item 9
+ 34) and the replacement cost of the assets is measured as the end-of-year book value of assets (Compustat item 6). Internally generated cash flows (CASH) is
measured as the income before extraordinary items (Compustat item 18) plus depreciation and amortization expenses (Compustat item 14) scaled by beginning-
of-year book value of assets (Compustat item 6). External finance (XFIN) is the sum of net cash inflows from the equity issuances (Compustat item 108-115-
127), net cash inflows from the long-term debt issuances (Compustat item 111-114) and cash flows from the change in current debt (Compustat item 301),
deflated by the beginning-of-year book value of assets (Compustat item 6). ROA is measured as net income (Compustat item 172) deflated by beginning-of-year
book value of assets (Compustat item 6). ACCRUALS is measured as the difference between earnings (Compustat item 172) and cash flows from operating
activities (Compustat item 308). Numbers in the parenthesis are p-value of t-statistics for mean test and z-statistics for median test

32
 Table 2
Correlation Table
(Pearson above diagonal and Spearman below diagonal)

Panel A: Correlations of variables used to test Hypothesis 1

INVESTMENTt Qt-1 CASHt-1 MAX_DAt-1

INVESTMENTt 1 0.128 0.230 -0.054

(0.001) (0.001) (0.001)

Qt-1 0.209 1 -0.006 0.007

(0.001) (0.146) (0.087)

CASHt-1 0.430 0.304 1 -0.014

(0.001) (0.001) (0.001)

MAX_DAt-1 -0.076 0.007 -0.018 1

(0.001) (0.084) (0.001)

Panel B: Correlations of variables used to test Hypothesis 2

ROA t+1 ACCRUALSt CASHt INVESTMENTt MAX_DAt-1

ROA t+1 1 0.191 0.646 0.046 -0.054

(0.001) (0.001) (0.001) (0.001)

ACCRUALSt 0.120 1 0.440 -0.136 0.013

(0.001) (0.001) (0.001) (0.001)

CASHt 0.628 0.228 1 0.179 -0.057

(0.001) (0.001) (0.001) (0.001)

INVESTMENTt 0.156 -0.102 0.397 1 -0.054

(0.001) (0.001) (0.001) (0.001)

MAX_DAt-1 -0.059 0.020 -0.075 -0.076 1

(0.001) (0.001) (0.001) (0.001)

New investment (INVESTMENT) is capital expenditure (Compustat item 128) deflated by beginning-of-year book value
of assets (Compustat item 6). Discretionary accruals is the residual from the regression CAit = α + β (ΔSalesit - ΔARit )
+ εit , where CA is the current accrual of firm i in year t (Compustat item Δ (4-1) - Δ (5-34-71)) , Δ Sales is the annual
change in sales for firm i in year t, ΔAR is the annual change in accounts receivable for firm i in year t, where all
variables are scaled by beginning-of-year book value of total assets. The performance-adjusted discretionary accrual
(DA) is the difference between the firm-specific discretionary accruals and the average discretionary accrual of a
portfolio of firms (excluding the sample firm itself) matched on Fama-French 48 industry classifications and current
return on assets quartiles. MAX_DA is an indicator variable, which takes the value of 1 if a firm is in the highest
quartile of performance-adjusted discretionary accruals and 0 otherwise Investment opportunities (Q) is measured as
the ratio of the market value to the replacement cost of the assets, where the market value is measured as the sum of the
market value of the equity (Compustat item 25*199) and the value of debt (Compustat item 9 + 34) and the
replacement cost of the assets is measured as the end-of-year book value of assets (Compustat item 6). Internally
generated cash flows (CASH) is measured as the income before extraordinary items (Compustat item 18) plus
depreciation and amortization expenses (Compustat item 14) scaled by beginning-of-year book value of assets
(Compustat item 6). ROA is measured as net income (Compustat item 172) deflated by beginning-of-year book value
of assets (Compustat item 6). ACCRUALS is measured as the difference between earnings (Compustat item 172) and
cash flows from operating activities (Compustat item 308).

33
 Table 3
Discretionary Accruals and Future Capital Investment

Model: INVESTMENTit = α + β1Qit-1 + β2CASHit-1 + β3MAX_DAit-1 + β4 Qit-1*MAX_DAit-1

+ β5 CASHit-1 *MAX_DAit-1 + INDUSTRY DUMMIES + YEAR DUMMIES + εit

Predicted Coefficients
Sign (p-value)
OLS OLS GMM OLS
Intercept 0.104 0.104 0.104 0.104
(0.001) (0.001) (0.001) (0.001)
Qt-1 (+) 0.008 0.008 0.008 0.008
(0.001) (0.001) (0.001) (0.001)
CASHt-1 (+) 0.085 0.087 0.087 0.087
(0.001) (0.001) (0.001) (0.001)
MAX_DAt-1 ? -0.006 -0.006 -0.006
(0.001) (0.001) (0.001)
Qt-1*MAX_DAt-1 ? 0.001 0.001 0.001
(0.370) (0.419) (0.370)
CASHt-1 *MAX_DAt-1 (-) -0.022 -0.022 -0.022
(0.005) (0.015) (0.005)
Industry dummies Included Included Included Included

Year dummies Included Included Included Included

Fixed firm effect Included

Adjusted R**2 23.75% 23.81% 23.81% 23.88%

N 60,728 60,728 60,728 60,728

New investment (INVESTMENT) is capital expenditure (Compustat item 128) deflated by beginning-of-year book value
of assets (Compustat item 6). Discretionary accruals is the residual from the regression CAit = α + β (ΔSalesit - ΔARit )
+ εit , where CA is the current accrual of firm i in year t (Compustat item Δ (4-1) - Δ (5-34-71)) , Δ Sales is the annual
change in sales for firm i in year t, ΔAR is the annual change in accounts receivable for firm i in year t, where all
variables are scaled by beginning-of-year book value of total assets. The performance-adjusted discretionary accrual
(DA) is the difference between the firm-specific discretionary accruals and the average discretionary accrual of a
portfolio of firms (excluding the sample firm itself) matched on Fama-French 48 industry classifications and current
return on assets quartiles. MAX_DA is an indicator variable, which takes the value of 1 if a firm is in the highest
quartile of performance-adjusted discretionary accruals and 0 otherwise. Investment opportunities (Q) is measured as
the ratio of the market value to the replacement cost of the assets, where the market value is measured as the sum of the
market value of the equity (Compustat item 25*199) and the value of debt (Compustat item 9 + 34) and the
replacement cost of the assets is measured as the end-of-year book value of assets (Compustat item 6). Internally
generated cash flows (CASH) is measured as the income before extraordinary items (Compustat item 18) plus
depreciation and amortization expenses (Compustat item 14) scaled by beginning-of-year book value of assets
(Compustat item 6).

34
 Table 4
Discretionary Accruals, Investment and Future Return on assets

Model: ROAit+1=α+β1ACCRUALSit+β2CASHit+β3INVESTMENTit+β4MAX_DAit-1
+β5INVESTMENTit*MAX_DAit-1+εit

Predicted Coefficients
Sign (p-value)
Intercept -0.052 -0.040 -0.039
(0.001) (0.001) (0.001)
ACCRUALSt (-) -0.120 -0.158 -0.157
(0.001) (0.001) (0.001)
CASHt (+) 0.688 0.720 0.718
(0.001) (0.001) (0.001)
INVESTMENTt (-) -0.207 -0.206
(0.001) (0.001)
MAX_DAt-1 (-) -0.009
(0.001)
INVESTMENTt*MAX_DAt-1 (-) -0.046
(0.068)
Adjusted R**2 42.33% 43.34% 43.38%

N 58,306 58,306 58,306

New investment (INVESTMENT) is capital expenditure (Compustat item 128) deflated by beginning-of-year book value
of assets (Compustat item 6). Discretionary accruals is the residual from the regression CAit = α + β (ΔSalesit - ΔARit )
+ εit , where CA is the current accrual of firm i in year t (Compustat item Δ (4-1) - Δ (5-34-71)) , Δ Sales is the annual
change in sales for firm i in year t, ΔAR is the annual change in accounts receivable for firm i in year t, where all
variables are scaled by beginning-of-year book value of total assets. The performance-adjusted discretionary accrual
(DA) is the difference between the firm-specific discretionary accruals and the average discretionary accrual of a
portfolio of firms (excluding the sample firm itself) matched on Fama-French 48 industry classifications and current
return on assets quartiles. MAX_DA is an indicator variable, which takes the value of 1 if a firm is in the highest
quartile of performance-adjusted discretionary accruals and 0 otherwise. Internally generated cash flows (CASH) is
measured as the income before extraordinary items (Compustat item 18) plus depreciation and amortization expenses
(Compustat item 14) scaled by beginning-of-year book value of assets (Compustat item 6). ROA is measured as net
income (Compustat item 172) deflated by beginning-of-year book value of assets (Compustat item 6) in year t.
ACCRUALS is measured as the difference between earnings (Compustat item 172) and cash flows from operating
activities (Compustat item 308).

35
 Table 5
Stock Return, Discretionary Accruals and External Financing

Model: BHARit=α + β1DAit-1 + β2CASHit-1 + β3XFINit + β4DAit-1 *XFINit + εit

Predicted Coefficients
Sign (p-value)
Intercept -0.044
(0.200)
DAt-1 (-) -0.591
(0.009)
CASHt-1 (+) 0.174
(0.004)
XFINt (+) 0.741
(0.001)
DAt-1 *XFINt (-) -1.565
(0.011)
Adjusted R**2 2.82%

N 5,610

External finance (XFIN) is the sum of net cash inflows from the equity issuances (Compustat item 108-115-127), net
cash inflows from the long-term debt issuances (Compustat item 111-114) and cash flows from the change in current
debt (Compustat item 301), deflated by the beginning-of-year book value of assets (Compustat item 6). New
investment (INVESTMENT) is capital expenditure (Compustat item 128) deflated by beginning-of-year book value of
assets (Compustat item 6). Discretionary accruals is the residual from the regression CAit = α + β (ΔSalesit - ΔARit ) +
εit, where CA is the current accrual of firm i in year t (Compustat item Δ (4-1) - Δ (5-34-71)) , ΔSales is the annual
change in sales for firm i in year t, ΔAR is the annual change in accounts receivable for firm i in year t, where all
variables are scaled by beginning-of-year book value of total assets. The performance-adjusted discretionary accrual
(DA) is the difference between the firm-specific discretionary accruals and the average discretionary accrual of a
portfolio of firms (excluding the sample firm itself) matched on Fama-French 48 industry classifications and current
return on assets quartiles. MAX_DA is an indicator variable, which takes the value of 1 if a firm is in the highest
quartile of performance-adjusted discretionary accruals and 0 otherwise. Investment opportunities (Q) is measured as
the ratio of the market value to the replacement cost of the assets, where the market value is measured as the sum of the
market value of the equity (Compustat item 25*199) and the value of debt (Compustat item 9 + 34) and the
replacement cost of the assets is measured as the end-of-year book value of assets (Compustat item 6). Internally
generated cash flows (CASH) is measured as the income before extraordinary items (Compustat item 18) plus
depreciation and amortization expenses (Compustat item 14) scaled by beginning-of-year book value of assets
(Compustat item 6). ROA is measured as net income (Compustat item 172) deflated by beginning-of-year book value
of assets (Compustat item 6) in year t. ACCRUALS is measured as the difference between earnings (Compustat item
172) and cash flows from operating activities (Compustat item 308). BHAR is the buy-and-hold size adjusted stock
return of firm i in fiscal year t.

36
 Table 6
Robustness Checks

Panel A: Investment at time t, controlling for external financing at time t

Predicted Coefficients
Sign (p-value)
Intercept 0.106 0.103 0.104
(0.001) (0.001) (0.001)
Qt-1 (+) 0.004 0.007 0.009
(0.001) (0.001) (0.001)
CASHt-1 (+) 0.122 0.110 0.089
(0.001) (0.001) (0.001)
XFINt ? 0.099 0.133
(0.001) (0.001)
Qt-1*XFINt ? -0.011
(0.001)
CASHt-1*XFINt ? 0.038
(0.001)
MAX_DAt-1 ? -0.004 -0.005 -0.020
(0.040) (0.007) (0.001)
Qt-1*MAX_DAt-1 ? -0.002 -0.001 0.000
(0.021) (0.142) (0.861)
CASHt-1* MAX_DAt-1 (-) -0.037 -0.040 -0.026
(0.001) (0.001) (0.046)
Industry dummies Included Included Included

Year dummies Included Included Included

Adjusted R**2 31.17% 32.10% 24.61%

N 60,728 60,728 57,385

37
 Table 6
(Continued)

Panel B: Investment at time t, controlling for abnormal returns from year t-4 to
year t-1, leverage at time t-1, and CEO trading in year t-1.

Coefficients
(p-value)
Predicted Control for Control for Control for
Sign BHARt-4,t-1 LEVERAGEt-1 CEOBUYt All Combined
Intercept 0.099 0.109 0.114 0.112
(0.001) (0.001) (0.001) (0.001)
Qt-1 (+) 0.009 0.006 0.009 0.006
(0.001) (0.001) (0.001) (0.001)
CASHt-1 (+) 0.088 0.068 0.088 0.068
(0.001) (0.001) (0.001) (0.001)
BHARt-4,t-1 ? 0.009 0.007
(0.001) (0.001)
Qt-1*BHARt-4,t-1 ? -0.002 -0.002
(0.001) (0.001)
CASHt-1*BHARt-4,t-1 ? 0.008 0.011
(0.001) (0.001)
LEVERAGEt-1 ? -0.053 -0.038
(0.001) (0.001)
Qt-1* LEVERAGEt-1 ? 0.037 0.028
(0.001) (0.001)
CASHt-1*LEVERAGEt-1 ? 0.123 0.138
(0.001) (0.001)
CEOBUYt ? -0.006 0.000
(0.001) (0.815)
Qt-1*CEOBUYt ? 0.003 0.000
(0.001) (0.771)
CASHt-1* CEOBUYt ? -0.007 -0.008
(0.157) (0.148)
MAX_DAt-1 ? -0.004 -0.004 -0.006 -0.003
(0.040) (0.015) (0.003) (0.178)
Qt-1*MAX_DAt-1 ? 0.000 -0.001 0.000 -0.001
(0.841) (0.522) (0.636) (0.193)
CASHt-1* MAX_DAt-1 (-) -0.039 -0.021 -0.022 -0.040
(0.001) (0.005) (0.005) (0.001)
Industry dummies Included Included Included Included
Year dummies Included Included Included Included
Adjusted R**2 26.48% 24.70% 23.66% 26.87%
N 46,455 60,728 60,728 46,455

38
 Table 6
(Continued)

Panel C: Substitute CASH with CFO to control for measurement error

Coefficients
Predicted Sign (p-value)
Intercept 0.099
(0.001)
Qt-1 (+) 0.008
(0.001)
CFOt-1 (+) 0.112
(0.001)
MAX_DAt-1 ? -0.001
(0.632)
Qt-1*MAX_DAt-1 ? 0.003
(0.008)
CFOt-1 *MAX_DAt-1 (-) -0.071
(0.001)
Industry dummies Included

Year dummies Included

Adjusted R**2 24.70%

N 57,862

39
 Table 6
(Continued)

Panel D: Investment at time t, controlling for discretionary accruals at time t

Predicted Coefficients
Sign (p-value)
Intercept 0.104 0.104
(0.001) (0.001)
Qt-1 (+) 0.008 0.008
(0.001) (0.001)
CASHt-1 (+) 0.087 0.087
(0.001) (0.001)
DAt ? -0.035 -0.020
(0.001) (0.001)
Qt-1*DAt ? -0.008
(0.004)
CASHt-1*DAt ? -0.071
(0.004)
MAX_DAt-1 ? -0.007 -0.007
(0.001) (0.001)
Qt-1*MAX_DAt-1 ? 0.001 0.001
(0.220) (0.175)
CASHt-1* MAX_DAt-1 (-) -0.019 -0.020
(0.013) (0.010)
Industry dummies Included Included

Year dummies Included Included

Adjusted R**2 23.90% 23.92%

N 60,728 60,728
External finance (XFIN) is the sum of net cash inflows from the equity issuances (Compustat item 108-115-127), net cash
inflows from the long-term debt issuances (Compustat item 111-114) and cash flows from the change in current debt
(Compustat item 301), deflated by the beginning-of-year book value of assets (Compustat item 6). New investment
(INVESTMENT) is capital expenditure (Compustat item 128) deflated by beginning-of-year book value of assets (Compustat
item 6). Discretionary accruals is the residual from the regression CAit = α + β (ΔSalesit - ΔARit ) + εit, where CA is the
current accrual of firm i in year t (Compustat item Δ (4-1) - Δ (5-34-71)) , ΔSales is the annual change in sales for firm i in
year t, ΔAR is the annual change in accounts receivable for firm i in year t, where all variables are scaled by beginning-of-year
book value of total assets. The performance-adjusted discretionary accrual (DA) is the difference between the firm-specific
discretionary accruals and the average discretionary accrual of a portfolio of firms (excluding the sample firm itself) matched
on Fama-French 48 industry classifications and current return on assets quartiles. MAX_DA is an indicator variable, which
takes the value of 1 if a firm is in the highest quartile of performance-adjusted discretionary accruals and 0 otherwise.
Investment opportunities (Q) is measured as the ratio of the market value to the replacement cost of the assets, where the
market value is measured as the sum of the market value of the equity (Compustat item 25*199) and the value of debt
(Compustat item 9 + 34) and the replacement cost of the assets is measured as the end-of-year book value of assets
(Compustat item 6). Internally generated cash flows (CASH) is measured as the income before extraordinary items
(Compustat item 18) plus depreciation and amortization expenses (Compustat item 14) scaled by beginning-of-year book
value of assets (Compustat item 6). BHARt-4,t-1 is the buy-and-hold size adjusted stock return of firm i from year t-4 to t-1.
CEOBUY is CEO’s net transaction (purchase – sales) size on firm’s stock in year t, where the size of each transaction is
measured as the number of shares traded, deflated by the number of shares held by CEO before transaction. CFO is cash
flows from operating activities (Compustat item 308). LEVERAGE is measured as the ratio of the book value of debt
(Compustat item 34+9) to the book value of assets (Compustat item 6).

40