Real Activities Manipulation and Accrual-Based E

Evidence on the trade-off between real activities manipulation and accrual-based earnings

management

Amy Y. Zang

The Hong Kong University of Science and Technology

Abstract: I study whether managers use real activities manipulation and accrual-based

earnings management as substitutes in managing earnings. I find that managers trade

off the two earnings management methods based on their relative costs and that

managers adjust the level of accrual-based earnings management according to the

level of real activities manipulation realized. Using an empirical model that incorporates

the costs associated with the two earnings management methods and captures

managers’ sequential decisions, I document large sample evidence consistent with

managers using real activities manipulation and accrual-based earnings management

as substitutes.

Keywords: real activities manipulation, accrual-based earnings management, trade-off

Data Availability: Data are available from public sources indicated in the text.

I am grateful for the guidance from my dissertation committee members, Jennifer Francis (chair),

Qi Chen, Dhananjay Nanda, Per Olsson and Han Hong. I am also grateful for the suggestions

and guidance received from Steven Kachelmeier (senior editor), Dan Dhaliwal and two

anonymous reviewers. I thank Allen Huang, Moshe Bareket, Yvonne Lu, Shiva Rajgopal,

Mohan Venkatachalam and Jerry Zimmerman for helpful comments. I appreciate the comments

from the workshop participants at Duke University, University of Notre Dame, University of

Utah, University of Arizona, University of Texas at Dallas, Dartmouth College, University of

Oregon, Georgetown University, University of Rochester, Washington University in St. Louis

and the HKUST. I gratefully acknowledge the financial support from the Fuqua School of

Business at Duke University, the Deloitte Foundation, University of Rochester and the HKUST.

Errors and omissions are my responsibility.

0

Electronic copy available at: /abstract=1891759

I. INTRODUCTION

I study how firms trade off two earnings management strategies, real activities manipulation

and accrual-based earnings management, using a large sample of firms over 1987–2008. Prior

studies have shown evidence of firms altering real activities to manage earnings (e.g.,

Roychowdhury 2006; Graham et al. 2005) and evidence that firms make choices between the

two earnings management strategies (Cohen et al. 2008; Cohen and Zarowin 2010; Badertscher

2011). My study extends research on the trade-off between real activities manipulation and

accrual-based earnings management by documenting a set of variables that explain the costs of

both real and accrual earnings management. I provide evidence for the trade-off decision as a

function of the relative costs of the two activities and show that there is direct substitution

between them after the fiscal year end due to their sequential nature.

Real activities manipulation is a purposeful action to alter reported earnings in a particular

direction, which is achieved by changing the timing or structuring of an operation, investment or

financing transaction, and which has suboptimal business consequences. The idea that firms

engage in real activities manipulation is supported by the survey evidence in Graham et al.

(2005).1 They report that 80 percent of surveyed CFOs stated that, in order to deliver earnings,

they would decrease research and development (R&D), advertising and maintenance

expenditures, while 55 percent said they would postpone a new project, both of which are real

activities manipulation.

In particular, Graham et al. (2005) note that: “The opinion of many of the CFOs is that every company

would/should take actions such of these [real activities manipulation] to deliver earnings, as long as the real

sacrifices are not too large and as long as the actions are within GAAP.” Graham et al. further conjecture that CFOs’

greater emphasis on real activities manipulation rather than accrual-based earnings management may be due to their

reluctance to admit to accounting-based earnings management in the aftermath of the Enron and Worldcom

accounting scandals.

11

Electronic copy available at: /abstract=1891759

Unlike real activities manipulation, which alters the execution of a real transaction taking

place during the fiscal year, accrual-based earnings management is achieved by changing the

accounting methods or estimates used when presenting a given transaction in the financial

statements. For example, changing the depreciation method for fixed assets and the estimate for

provision for doubtful accounts can bias reported earnings in a particular direction without

changing the underlying transactions.

The focus of this study is on how managers trade off real activities manipulation and accrual-based earnings management. This question is important for two reasons. First, as mentioned by

Fields et al. (2001), examining only one earnings management technique at a time cannot explain

the overall effect of earnings management activities. In particular, if managers use real activities

manipulation and accrual-based earnings management as substitutes for each other, examining

either type of earnings management activities in isolation cannot lead to definitive conclusions.

Second, by studying how managers trade off these two strategies, this study sheds light on the

economic implications of accounting choices; that is, whether the costs that managers bear for

manipulating accruals affect their decisions about real activities manipulation. As such, the

question has implications about whether enhancing SEC scrutiny or reducing accounting

flexibility in GAAP, for example, might increase the levels of real activities manipulation

engaged in by firms.

I start by analyzing the implications for managers’ trade-off decisions due to the different

costs and timing of the two earnings management strategies. First, because both are costly

activities, firms trade off real activities manipulation versus accrual-based earnings management

based on their relative costliness. That is, when one activity is relatively more costly, firms

engage in more of the other. Because firms face different costs and constraints for the two

2

earnings management approaches, they show differing abilities to use the two strategies. Second,

real activities manipulation must occur during the fiscal year and is realized by the fiscal year

end, after which managers still have the chance to adjust the level of accrual-based earnings

management. This timing difference implies that managers would adjust the latter based on the

outcome of real activities manipulation. Hence, there is also a direct, substitutive relation

between the two: if real activities manipulation turns out to be unexpectedly high (low),

managers will decrease (increase) the amount of accrual-based earnings management they carry

out.

Following prior studies, I examine real activities manipulation through overproduction and

cutting discretionary expenditures (Roychowdhury 2006; Cohen et al. 2008; Cohen and Zarowin

2010). I test the hypotheses using a sample of firms that are likely to have managed earnings. As

suggested by prior research, earnings management is likely to occur when firms just beat/meet an

important earnings benchmark (Burgstahler and Dichev 1997; DeGeorge et al. 1999). Using a

sample containing more than 6,500 earnings management suspect firm-years over the period

1987–2008, I show the empirical results that real activities manipulation is constrained by firms’

competitive status in the industry, financial health, scrutiny from institutional investors, and the

immediate tax consequences of manipulation. The results also show that accrual-based earnings

management is constrained by the presence of high-quality auditors; heightened scrutiny of

accounting practice after the passage of the Sarbanes-Oxley Act (SOX); and firms’ accounting

flexibility, as determined by their accounting choices in prior periods and the length of their

operating cycles. I find significant positive relations between the level of real activities

manipulation and the costs associated with accrual-based earnings management, and also

between the level of accrual-based earnings management and the costs associated with real

3

activities manipulation, supporting the hypothesis that managers trade off the two approaches

according to their relative costliness. There is a significant and negative relation between the

level of accrual-based earnings management and the amount of unexpected real activities

manipulation, consistent with the hypothesis that managers “fine-tune” accruals after the fiscal

year end based on the realized real activities manipulation. Additional Hausman tests show

results consistent with the decision of real activities manipulation preceding the decision of

accrual-based earnings management.

Two recent studies have examined the trade-off between real activities manipulation and

accrual-based earnings management. Cohen et al. (2008) document that, after the passage of

SOX, the level of accrual-based earnings management declines, while the level of real activities

manipulation increases, consistent with firms switching from the former to the latter as a result of

the post-SOX heightened scrutiny of accounting practice. Cohen and Zarowin (2010) show that

firms engage in both forms of earnings management in the years of a seasoned equity offering

(SEO). They show further that the tendency for SEO firms to use real activities manipulation is

positively correlated with the costs of accrual-based earnings management in these firms.2

Compared to prior studies, this study contributes to the earnings management literature by

providing a more complete picture of how managers trade off real activities manipulation and

accrual-based earnings management. First, it documents the trade-off in a more general setting

by using a sample of firms that are likely to have managed earnings to beat/meet various

earnings targets. The evidence for the trade-off decisions discussed in this study does not depend

on a specific period (such as around the passage of SOX, as in Cohen et al. 2008) or a significant

corporate event (such as a SEO, as in Cohen and Zarowin 2010).

Cohen and Zarowin (2010) do not examine how accrual-based earnings management for SEO firms varies based

on the costs of real and accrual earnings management.

24

Second, to my knowledge, mine is the first study to identify a set of costs for real activities

manipulation and to examine their impact on both real and accrual earnings management

activities. Prior studies (Cohen et al. 2008; Cohen and Zarowin 2010) only examine the costs of

accrual-based earnings management. By including the costs of real activities manipulation, this

study provides evidence for the trade-off as a function of the relative costs of the two approaches.

That is, the level of each earnings management activity decreases with its own costs and

increases with the costs of the other. In this way, I show that firms prefer different earnings

management strategies in a predictive manner, depending on their operational and accounting

environment.

Third, I consider the sequential nature of the two earnings management strategies. Most prior

studies on multiple accounting and/or economic choices implicitly assume that managers decide

on multiple choices simultaneously without considering the sequential decision process as an

alternative process (Beatty et al. 1995; Hunt et al. 1996; Gaver and Paterson 1999; Barton 2001;

Pincus and Rajgopal 2002; Cohen et al. 2008; Cohen and Zarowin 2010). In contrast, my

empirical model explicitly considers the implication of the difference in timing between the two

earnings management approaches. Because real activities manipulation has to occur during the

fiscal year, but accrual manipulation can occur after the fiscal year end, managers can adjust the

extent of the latter based on the realized outcomes of the former. I show that, unlike the trade-off

during the fiscal year, which is based on the relative costliness of the two strategies, there is a

direct substitution between the two approaches at year end when real activities manipulation is

realized. Unexpectedly high (low) real activities manipulation realized is directly offset by a

lower (higher) amount of accrual earnings management.

5

Section II reviews relevant prior studies. Section III develops the hypotheses. Section IV

describes the research design, measurement of real activities manipulation, accrual-based

earnings management and independent variables. Section V reports sample selection and

empirical results. Section VI concludes and discusses the implications of my results.

II. RELATED LITERATURE

The extensive literature on earnings management largely focuses on accrual-based earnings

management (reviewed by Schipper 1989; Healy and Wahlen 1999; Fields et al. 2001). A

smaller stream of literature investigates the possibility that managers manipulate real transactions

to distort earnings. Many such studies examine managerial discretion over R&D expenditures

(Baber et al. 1991; Dechow and Sloan 1991; Bushee 1998; Cheng 2004). Other types of real

activities manipulation that have been explored include cutting advertising expenditures (Cohen

et al. 2010), stock repurchases (Hribar et al. 2006), sales of profitable assets (Herrmann et al.

2003; Bartov 1993), sales price reductions (Jackson and Wilcox 2000), derivative hedging

(Barton 2001; Pincus and Rajgopal 2002), debt-equity swaps (Hand 1989), and securitization

(Dechow and Shakespeare 2009).

The prevalence of real activities manipulation as an earnings management tool was not well

understood until recent years. Graham et al. (2005) survey more than 400 executives and

document the widespread use of real activities manipulation. Eighty percent of the CFOs in their

survey stated that, in order to meet an earnings target, they would decrease expenditure on R&D,

advertising and maintenance, while 55 percent said they would postpone a new project, even if

such delay caused a small loss in firm value. Consistent with this survey, Roychowdhury (2006)

documents large-sample evidence suggesting that managers avoid reporting annual losses or

6

missing analyst forecasts by manipulating sales, reducing discretionary expenditures, and

overproducing inventory to decrease the cost of goods sold, all of which are deviations from

otherwise optimal operational decisions, with the intention of biasing earnings upward.

Recent research has started to examine the consequence of real activities manipulation.

Gunny (2010) finds that firms that just meet earnings benchmarks by engaging in real activities

manipulation have better operating performance in the subsequent three years than do firms that

do not engage in real activities manipulation and miss or just meet earnings benchmarks. Bhojraj

et al. (2009), on the other hand, show that firms that beat analyst forecasts by using real and

accrual earnings management have worse operating performance and stock market performance

in the subsequent three years than firms that miss analyst forecasts without earnings management.

Most previous research on earnings management examines only one earnings management

tool in settings where earnings management is likely to occur (e.g., Healy 1985; Dechow and

Sloan 1991; Roychowdhury 2006). However, given the portfolio of earnings management

strategies, managers probably use multiple techniques at the same time. A few prior studies

(Beatty et al. 1995; Hunt et al. 1996; Gaver and Paterson 1999; Barton 2001; Pincus and

Rajgopal 2002; Cohen et al. 2008; Cohen and Zarowin 2010; Badertscher 2011) examine how

managers use multiple accounting and operating measures to achieve one or more goals.

Beatty et al. (1995) study a sample of 148 commercial banks. They identify two accrual

accounts (loan loss provisions and loan charge-offs) and three operating transactions (pension

settlement transactions, miscellaneous gains and losses due to asset sales, and issuance of new

securities) that these banks can adjust to achieve three goals (optimal primary capital, reported

earnings and taxable income levels). The authors construct a simultaneous equation system, in

which the banks minimize the sum of the deviations from the three goals and from the optimal

7

levels of the five discretionary accounts.3 They find evidence that some, but not all, of the

discretionary accounts (including both accounting choices and operating transactions) are

adjusted jointly for some of the objectives identified.

Barton (2001) and Pincus and Rajgopal (2002) study how firms manage earning volatility

using a sample of Fortune 500, and oil and gas, firms respectively. Both studies use simultaneous

equation systems, in which derivative hedging and accrual management are simultaneously

determined to manage earnings volatility. Barton (2001) suggests that the two activities are used

as substitutes, as evidenced by the negative relation between the two after controlling for the

desired level of earnings volatility. Pincus and Rajgopal (2002) find similar negative relation, but

only in the fourth quarter.

There are two limitations in the approach taken by the above studies. First, in the empirical

tests, they assume that the costs of adjusting discretionary accounts are constant across all firms

and hence do not generate predictions or incorporate empirical proxies for the costs. In other

words, they do not consider that discretion in some accounts is more costly to adjust for some

firms. Hence, these studies fail to consider the trade-off among different tools due to their

relative costs. Second, they assume all decisions are made simultaneously. If some decisions are

made before others, this assumption can lead to misspecification in their equation system.

Badertscher (2011) examines overvaluation as an incentive for earnings management. He

finds that during the sustained period of overvaluation, managers use accrual earnings

management in early years, real activities manipulation in later years, and non-GAAP earnings

management as a last resort. He claims that the duration of overvaluation is an important

determinant in managers’ choice of earnings management approaches, but he does not model the

3

Hunt et al. (1996) and Gaver and Paterson (1999) follow Beatty et al. (1995) and construct similar simultaneous

equation systems.

8

trade-off between real activities manipulation and accrual-based earnings management based on

their relative costliness, nor does his study examine the implication of the sequential nature of

the two activities during the year.

Two recent studies examine the impact of the costs of accrual-based earnings management

on the choice of earnings management strategies. Cohen et al. (2008) show that, on average,

accrual-based earnings management declines, but real activities manipulation increases, after the

passage of SOX. They focus on one cost of accrual-based earnings management, namely the

heightened post-SOX scrutiny of accounting practice, and its impact on the levels of real and

accrual earnings management. Using a sample of SEO firms, Cohen and Zarowin (2010)

examine several costs of accrual-based earnings management and show that they are positively

related to the tendency to use real activities manipulation in the year of a SEO. Neither study

examines the costs of real activities manipulation or considers the sequential nature of the two

strategies. Hence, they do not show the trade-off decision as a function of the relative costs of the

two strategies or the direct substitution between the two after the fiscal year end.

III. HYPOTHESES DEVELOPMENT

Consistent with prior research on multiple earnings management strategies, I predict that

managers use real activities manipulation and accrual-based earnings management as substitutes

to achieve the desired earnings targets. Unlike prior research, however, I investigate the

differences in the costs and timing of real activities manipulation and accrual-based earnings

management, and their implications for managers’ trade-off decisions.

Both real activities manipulation and accrual-based earnings management are costly

activities. Firms are likely to face different levels of constraints for each strategy, which will lead

9

to varying abilities to use them. A manager’s trade-off decision, therefore, depends on the

relative costliness of the two earnings management methods, which is in turn determined by the

firm’s operational and accounting environment. That is, given the desired level of earnings, when

discretion is more constrained for one earnings management tool, the manager will make more

use of the other. This expectation can be expressed as the following hypothesis:

H1: Other things being equal, the relative degree of accrual-based earnings management vis-à-vis real activities manipulation depends on the relative costs of each action.

Accrual-based earnings management is constrained by scrutiny from outsiders and the

available accounting flexibility. For example, a manager might find it harder to convince a high-quality auditor of his/her aggressive accounting estimates than a low-quality auditor. A manager

might also feel that accrual-based earnings management is more likely to be detected when

regulators heighten scrutiny of firms’ accounting practice. Other than scrutiny from outsiders,

accrual-based earnings management is constrained by the flexibility within firms’ accounting

systems. Firms that are running out of such flexibility due to, for example, their having made

aggressive accounting assumptions in the previous periods, face an increasingly high risk of

being detected by auditors and violating GAAP with more accrual-based earnings management.

Hence, I formulate the following two subsidiary hypotheses to H1:

H1a: Other things being equal, firms facing greater scrutiny from auditors and regulators

have a higher level of real activities manipulation.

H1b: Other things being equal, firms with lower accounting flexibility have a higher level of

real activities manipulation.

Real activities manipulation, as a departure from optimal operational decisions, is unlikely

to increase firms’ long-term value. Some managers might find it particularly costly because their

10

firms face intense competition in the industry. Within an industry, firms are likely to face various

levels of competition and, therefore, are under different amounts of pressure when deviating

from optimal business strategies. Management research (as reviewed by Woo 1983) shows that

market leaders enjoy more competitive advantages than do followers, due to their greater

cumulative experience, ability to benefit from economies of scale, bargaining power with

suppliers and customers, attention from investors, and influence on their competitors. Therefore,

managers in market-leader firms may perceive real activities manipulation as less costly because

the erosion to their competitive advantage is relatively small. Hence, I predict the following:

H1c: Other things being equal, firms without market-leader status have a higher level of

accrual-based earnings management.

For a firm in poor financial health, the marginal cost of deviating from optimal business

strategies is likely to be high. In this case, managers might perceive real activities manipulation

as relatively costly because their primary goal is to improve operations. This view is supported

by the survey evidence documented by Graham et al. (2005), who find that CFOs admit that if

the company is in a “negative tailspin,” managers’ efforts to survive will dominate their

reporting concerns. This reasoning leads to the following subsidiary hypothesis to H1:

H1d: Other things being equal, firms with poor financial health have a higher level of

accrual-based earnings management.

Managers might find it difficult to manipulate real activities when their operation is being

monitored closely by institutional investors. Prior studies suggest that institutional investors play

a monitoring role in reducing real activities manipulation.4 Bushee (1998) finds that, when

However, there is also evidence that “transient” institutions, or those with high portfolio turnover and highly

diversified portfolio holdings, increase managerial myopic behavior (e.g., Porter 1992; Bushee 1998; Bushee 2001).

In this study, I focus on the average effect of institutional ownership on firms’ earnings management activities

without looking into the investment horizon of different institutions.

411

institutional ownership is high, firms are less likely to cut R&D expenditure to avoid a decline in

earnings. Roychowdhury (2006) also finds a negative relation between institutional ownership

and real activities manipulation to avoid losses. Unlike accrual-based earnings management, real

activities manipulation has real economic consequences for firms’ long-term value. Institutional

investors, being more sophisticated and informed than other investors, are likely to have a better

understanding of the long-term implication of firms’ operating decisions, leading to more effort

to monitor and curtail real activities manipulation than accrual-based earnings management, as

predicted in the following subsidiary hypothesis:

H1e: Other things being equal, firms with higher institutional ownership have a higher level

of accrual-based earnings management.

Real activities manipulation is also costly due to tax incentives. It might be subject to a

higher level of book-tax conformity than accrual-based earnings management, because the

former has a direct cash flow effect in the current period, while the latter does not. Specifically,

when firms increase book income by cutting discretionary expenditures or by overproducing

inventory, they also increase taxable income and incur higher tax costs in the current period.5 In

contrast, management of many accrual accounts increases book income without current-period

tax consequences. For example, increasing the estimated useful lives of long-term assets,

decreasing write-downs for impaired assets, recognizing unearned revenue aggressively, and

decreasing bad debt expense can all increase book income without necessarily increasing

current-year taxable income. Therefore, for firms with higher marginal tax rates, the net present

value of the tax costs associated with real activities manipulation is likely to be higher than that

of accrual-based earnings management, leading to the following prediction:

5

Other types of real activities manipulation, such as increasing sales by discounts and price cuts, and sale of long-term assets, are also book-tax conforming earnings management.

12

H1f: Other things being equal, firms with higher marginal tax rates have a higher level of

accrual-based earnings management.

Another difference between the two earnings management strategies that will influence

managers’ trade-off decisions is their different timing. H1 predicts that the two earnings

management strategies are jointly determined and the trade-off depends on their relative

costliness. However, a joint decision does not imply a simultaneous decision. Because real

activities manipulation changes the timing and/or structuring of business transactions, such

decisions and activities have to take place during the fiscal year. Shortly after the year end, the

outcome of the real activities manipulation is revealed, and managers can no longer engage in it.

Note that, when a manager alters real business decisions to manage earnings, s/he does not have

perfect control over the exact amount of the real activities manipulation attained. For example, a

pharmaceutical company cuts current-period R&D expenditure by postponing or cancelling

development of a certain drug. This real decision can include a hiring freeze and shutting down

the research site. The manager may be able to make a rough estimate of the dollar amount of the

impact on R&D expenditure from these decisions, but s/he does not have perfect information

about it.6 Therefore, managers face uncertainty when they execute real activities manipulation.

After the fiscal year end, the realized amount of the real activities manipulation could be higher

or lower than the amount originally anticipated.

On the other hand, after the fiscal year end but before the earnings announcement date,

managers can still adjust the accruals by changing the accounting estimates or methods. In

addition, unlike real activities manipulation, which distorts earnings by executing transactions

Another example is reducing travelling expenditures by requiring employees to fly economy class instead of

allowing them to fly business class. This change could be suboptimal because employees might reduce the number

of visit they make to important clients or because employees’ morale might be adversely impacted, leading to greater

turnover. The manager cannot know for certain the exact amount of SG&A being cut, as s/he does not know the

number of business trips taken by employees during the year.

613

differently, accrual management impacts reported earnings in a more immediate and certain

manner. Therefore, when managers observe the impact of real activities manipulation on

earnings at the fiscal year end, they can offset an unexpectedly high (low) impact by using less

(more) accrual management. This prediction, which is my next hypothesis, is based on the

premise that managers will use the two earnings management methods as substitutes:

H2: Managers adjust the amount of accrual-based earnings management after real activities

manipulation is realized; the level of accrual-based earnings management is negatively

related to the unexpected amount of real activities manipulation.

IV. RESEARCH DESIGN

Real activities manipulation

Following Roychowdhury (2006), I examine the following manipulation of real activities:

increasing earnings by reducing the cost of goods sold by overproducing inventory, and cutting

discretionary expenditures, including R&D, advertising and selling, general and administrative

(SG&A) expenditures.7 The former is measured by the abnormal level of production costs, the

latter by the abnormal level of discretionary expenditures. Subsequent studies using the same

metrics (Cohen et al. 2008; Cohen and Zarowin 2010) provide further evidence that these

measures capture real activities manipulation.

I estimate the normal level of production costs following Roychowdhury (2006):