University of Amsterdam
Faculty Economics and Business
Master Accountancy en Control
Masters Thesis
The effect of auditor tenure and auditor size on the
earnings management trade-off
Name
Max Ferguson
Student No.
10824715
Date
20
thof June 2015
Supervisor
Wim Janssen
Abstract
This paper examines the effect of auditor size and auditor tenure on accrual-based and real earnings management, as well as the trade-off between the two. Prior literature does not find consistent results in regard to auditor tenure and audit quality. To measure accrual-based earnings management the modified Jones (1991) model was used. In order to measure real earnings management Roychowdhury’s (2006) model was used to determine abnormal cash flows from operations; abnormal discretionary expenses; and, abnormal production. This thesis finds a distinct difference in how auditor tenure and size affects earnings management between pre and post Sarbanes Oxley (SOx) periods. Pre SOx auditor tenure was found to increase both accrual-based and real earnings management, post SOx auditor tenure had no significant impact. Auditor size was not found to have a significant effect, except in in post-SOx period when it was found to increases real earnings management.
This document is written by Student [Max Ferguson] who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
Table of Contents
1.0 INTRODUCTION ...4
!
2.0 LITERATURE REVIEW ...7
2.1
P
OSITIVE ACCOUNTING THEORY...7
2.2
A
CCRUALS-
BASED AND REAL EARNING MANAGEMENT...7
2.3
T
RADE-
OFF BETWEEN ACCRUALS-
BASED AND REAL EARNINGS MANAGEMENT..8
2.4
F
ACTORS AFFECTING THE EARNINGS MANAGEMENT TRADE-
OFF...10
2.4.1
A
UDIT QUALITY...10
2.4.2
A
UDITOR TENURE...10
2.4.3
A
UDIT FIRM SIZE(B
IGN
AND NON-B
IGN
AUDIT FIRMS) ...11
2.4.3
S
ARBANES-O
XLEYA
CT2002 ...12
!
3.0 HYPOTHESIS DEVELOPMENT ...14
!
4.0 RESEARCH METHODOLOGY ...15
4.1
A
CCRUAL-
BASED EARNINGS MANAGEMENT...15
4.2
R
EAL EARNINGS MANAGEMENT...15
4.3
S
AMPLE...18
4.4
E
STIMATION MODELS FOR ACCRUAL-
BASED EARNINGS MANAGEMENT AND REAL EARNINGS MANAGEMENT...18
4.4.1
P
ROXIES OF AUDIT QUALITY...18
4.4.2
C
ONTROL VARIABLES...19
4.4.3
R
EGRESSIONM
ODELS...20
4.4.4
E
XPECTATIONS...21
!5.0 EMPIRICAL FINDINGS ...22
5.1
D
ESCRIPTIVES
TATISTICS...22
5.2
C
ORRELATION...25
5.3
R
EGRESSIONR
ESULTS...28
!6.0 CONCLUSION ...34
!7.0 BIBLIOGRAPHY ...37
1.0 Introduction
Auditor tenure has been a contentious topic worldwide for many years. European Union guidelines stipulate that audit firm rotation must occur every ten years, while in the Netherlands auditor tenure is limited to eight years. Although such regulation has been growing in popularity, it still faces opposition from many parts of the auditing world. Auditing firms argue that when a firm changes auditor the new partner will lack experience and familiarity with the company, which will lead to lower reporting quality (Litt et al., 2014).
The case for auditor rotation is based on the assumption that an extended tenure will lead to a loss of independence and thus a decreased level of reporting quality. If an audit partner becomes too familiar with a company and thereby loses their independence they are less likely to issue a going concern (Carey, 2006). This impedes auditors from performing one of their primary functions. Furthermore, Myers (2005) found that long-term tenure was associated with increased levels of abnormal accruals, a sign of accrual-based management. Such practices have been deemed to be a contributing factor in many high profile accounting scandals such as Enron, WorldCom and Lehman Brothers. These practices lead to a loss of confidence and legitimation of the auditing profession.
Earnings management is the practice of misleading stakeholders in regard to a company’s performance through either over/under stating accruals (accrual-based earnings management) or through real activities manipulation (real earnings management). Accrual-based earnings management has no cash flow consequences. Accruals are defined as the difference between cash flow and reported earnings (Healy, 1985). In comparison real earnings management is a divergence from normal operating procedure for the motivation of misleading stakeholders (Roychowdhury, 2006).
Both methods are harmful to long-term performance and are frequently associated as being enabled to varying degrees by, independence and tenure (Johnson, 2002; Meyers, 2005; Chi, 2011). There is an apparent trade-off between the two; according to Zang (2012) management will choose the less costly alternative. Thereby meaning that the two act as substitutes of one
another and are simply two different methods of achieving the same goal. Corroborating the notion of substitution, Chi (2011) finds evidence that enhanced reporting quality will make accruals manipulation too arduous and therefore lead to a higher use of real earnings management.
In this research I aim to examine the relationship between auditor tenure and auditor size, on the trade-off between real and accruals earnings management. I will focus on the degree to which auditor tenure and auditor size increases audit quality and thereby effects use of accrual and real earnings management. I expect an increase in audit quality through auditor tenure and auditor size to be associated with a decrease in accruals-based earnings management and an increase in real earnings management.
The research question of this paper aims to further investigate; the effect of auditor tenure and auditor size on audit quality; and, the effect of audit quality on the earnings management trade-off.
!
Previous literature associated with this field focuses primarily on auditor tenure and its effect on accruals earnings management. Zang (2012) investigated the trade-off between accruals-based and real earnings management, I intend on expanding on this research by testing the trade-off in relation to auditor tenure and auditor size. Furthermore, Chi (2011) found a relationship between reporting quality and accruals-based earnings management, I will extend this research by analysing the relationship between earnings management and reporting quality with a focus on auditor tenure and size. Through this research I aim to build upon previous knowledge of firms’ earnings management trade-off.
!
In the last ten years there have been increasing steps to mandate auditor rotation both at a partner and firm level. There has been much discussion regarding the benefits and detriments of auditor rotation. Proponents argue that extended periods of auditor tenure lead to a decrease in audit quality and an increase in accruals-based earnings management (Myers, 2005). However, opponents argue that audit quality suffers upon rotation due to the new partner’s inexperience with that firm (Johnson et al., 2002). My paper will discuss the effect of auditor tenure and size
on the trade-off between real and accruals-based management. I believe this to be a novel and interesting endeavour that will give new insight into the effect of auditor tenure and size. Furthermore, this research also has the possibility to be beneficial to investors by increasing knowledge of firms’ earnings management tendencies, this could lead to investors being more capable of accurately assessing a firm’s financial reports.
2.0 Literature review
2.1 Positive accounting theory
Within accounting much is left to the manager’s discretion, thus the opportunity exists to account for events in the manner that best benefits management. Positive accounting theory, written by Watts and Zimmerman (1990) describes the relationship between a manager’s decisions and their personal interests. Positive accounting theory assists in understanding the nature of accounting choice and the motives for engaging in earnings management. The underlying assumption of Watts and Zimmerman is that a manager seeks to increase his own personal well-being. Thus a manager will choose the accounting policies that provide him with the best possible performance review. Managers are typically judged on their ability to increase profits, therefore, income increasing earnings management is the most common. Watts and Zimmerman’s 1978 and 1990 papers clearly show that management is inclined to manipulate their choice of accounting policy in order to maximise their own utility. This research provides a solid grounding for the motives of earnings management and shows how an agency problem exists within firms.
2.2 Accruals-based and real earning management
Earnings management can be accomplished through different methods; effectively it is the practice of managers misleading shareholders and other stakeholders for their personal gain. It is best described by Healy and Wahlen (1999), “Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting practices”. Positive accounting theory gives us an insight into the motivation for engaging in earnings management, due to Watts and Zimmerman we know managers have the incentives to engage in earnings management. This paper will focus on two forms of earnings management: accrual-based earnings management and real earnings management.
Accrual-based earnings management is achieved through using discretionary accounting policies in order to increase/decrease accruals with the intention of misleading stakeholders (Healy and Wahlen, 1999). In this scenario, there is no deviation from normal business procedure; instead managers use the accounting policies that best suit their desired outcomes. The second method, real earnings management, is achieved through a deviation of normal business procedure; it is the act of misleading stakeholders through a departure from normal operational practices, thereby creating the illusion of certain reporting goals being achieved through normal operations (Rochowdhury, 2006). Both methods seek the same outcomes; it is only the methods that differ. Accruals are intended to give a true and fair view of a firm through representing those transactions that have been completed but not finalised with a cash transaction. Accrual-based earnings management seeks to exploit this practice by using biased estimates with the intention of distorting the firm’s perceived economic performance (Healy and Wahlen, 1999).
Real earnings management utilises the resulting accounting outcomes of real business decisions to manipulate financial reports. By deviating from normal operating procedure managers have the ability to alter the perceived profitability of their firm or department. Examples of this include cutting expenses such as, research and development, and selling, general and administrative, below required levels. This creates the short-term illusion of greater profitability but will negatively affect the firm in the long term (Cazavan-Jeny, 2011).
2.3 Trade-off between accruals-based and real earnings management
!
As both methods provide the same outcome this raises the question, which method to use? As with all business decisions the logical method is based on a cost-benefit analysis. Roychowdhury (2006) describes the cost of real earnings management to be the cost of deviating from operating procedure. Due the fact that accrual-based earnings management does not change business procedure but merely the way it is represented, its costs are more difficult to define. For the manager in question this is most likely to be the cost of being exposed. Therefore, it would be logical to assume as a manager’s ability to employ accrual-based earnings management is reduced they will switch to real earnings management.
Ewert & Wagenhofer (2005) investigated the trade-off between real earnings management and accruals-based earnings management with regard to the introduction of tighter accounting restrictions. They argue that a standard setter has the capability to make standards more restrictive by giving less accounting choices to management. This reduces a manager’s opportunity to engage in accrual-based earnings management. Nevertheless, this does not influence a firm’s ability to engage in real earnings management. It is due to this that Ewert & Wagenhofer attribute the substitution effect they witness, which is seen through an increase in real earnings management with the introduction of tighter standards. They find that a firm will simply substitute real earnings management when its ability to use accrual-based techniques is constricted. Watts and Zimmerman (1990) outline how a manager is incentivised to choose the accounting policies that give them the best outcome. By restricting a manager’s ability to do so, the opportunity to do so has been reduced, however, it has not affected their motivation. It is therefore that a manager may instead utilise real earnings management to achieve the same result.
This substitution effect of trade-off has been further researched by Chan, Chen, Tai Yuan and Yangxin (2014) who found the introduction of claw back provisions was associated with a decrease in accrual-based earnings management and an increase in real earnings management. This is the same relationship as found by Ewert & Wagenhofer, as accrual-based earnings management is constricted (in these cases tighter regulation and claw back provisions) the firm will turn to real earnings management. In this instance claw back provisions guard shareholders against the future negative reversal of accruals, this method focuses on constricting a manager’s incentive to engage in accruals-based earnings management, however, real earnings management is still considered a viable alternative as its effects are typically more long term and likely to not become apparent until the claw back provisions have expired.
Zang (2012) further analyses the trade-off between accruals earnings management and real earnings management. She found evidence that pointed to managers adjusting the degree of accrual-based earnings management when real earnings management was more effective. The paper underlines that real earnings management is conducted during the fiscal year but realised at
the end of financial year, this suggests that managers have the capability of adjusting accruals and thus engaging in accrual-based earnings management depending on the effectiveness of their real earnings management strategy. Furthermore, Zang (2012) finds evidence to show that managers trade-off earnings management methods depending on their respective cost. For example, evidence is provided to show firms facing tighter regulation substitute for real earnings management and firms for whom real earnings management is too costly do the reverse.
2.4 Factors affecting the earnings management trade-off 2.4.1 Audit quality
Audit quality is the probability that an existing material error is detected and reported by the auditor (DeAngelo, 1981). Thus, we can see that audit quality is determined by two factors, the ability of the auditor to detect material misstatements and the probability that they will report the error. The ability of an auditor to detect misstatements is influenced by their skill and previous experience with the firm. The probability that the auditor will report the error is determined by the auditor’s independence. An auditor’s independence can then be further broken up into independence in fact and independence in appearance (Arens et al., 2011). Independence in fact, is an auditor’s true state of independence, an auditor who is independent in fact can make independent decisions regarding to their audit client. Independence in appearance is the whether outsiders perceive the auditor to be independent. Both are necessary for an audit to be successful as without independence in appearance investors will be unable to trust the audit findings and without independence in fact audit quality is severely impaired.
As audit quality is increased it also drives reporting quality (Johnson, 2002). Therefore, with greater audit quality the ability to employ accrual-based earnings management is reduced. It becomes more costly than to switch to real earnings management. Thus, factors that increase audit quality are also likely to increase real earnings management.
Auditor tenure has long been connected with affecting audit quality. It has been argued that short-term auditor tenure causes lower audit quality due to the auditor’s lack of experience with the company. Long-term auditor tenure is believed to cause lower audit quality as too much familiarity the auditor loses their independence. Johnson (2002) found that there were higher levels of unexpected accruals in the first two to three years of auditor tenure. In the long term, which Johnson defined as longer than nine years, no increase in accruals-based earnings management was observed. This is in accordance with Myers (2005) who found long term auditor tenure (five years) was associated with fewer unexpected accruals. In contrast to this Davis (2009) found that both short-term and long-term tenure were in fact associated with higher levels of unexpected accruals. This could well be due to Davis’s alternative definition of long-term tenure, greater than 15 years. It is quite possible that an auditor’s relationship with a client takes considerably longer to develop to the extent where audit quality diminishes than Johnson and Myers tested. Furthermore, Davis (2009) found that the relationship between higher accruals and long-term auditor tenure was no longer present after the introduction of Sarbanes-Oxley Act 2002. This is possibly a result of the stricter reporting regulations placed upon managers and auditors. It has increased the cost of accrual-based earnings management past the cost of real earnings management, thus affecting the trade-off.
From prior literature we can clearly see that relationship between audit quality and auditor tenure exists. In accordance with Watts and Zimmerman (1990) a manager’s incentives to choose the most favourable accounting policies should not be affected by audit quality. Thus, this change in audit quality will be related to a switch in accrual-based earnings management and real earnings management. Assuming Zang’s (2012) findings hold true, we would expect to see a direct trade-off between accruals-based and real earnings management, as auditor tenure increases.
2.4.3 Audit firm size (Big N and non-Big N audit firms)
Previous research has found that the size of an audit firm has a direct relationship with audit quality. According to DeAngelo (1981) the smaller an audit firm’s client is as percentage of the auditor’s quasi rent, the less incentive the auditor had to behave opportunistically and the higher
the perceived quality of the audit. Larger firms are seen as less likely to compromise their independence, as every client is only a small fraction of total revenue. Big N auditors are more likely to have many clients and therefore, no one client is so valuable as to affect their independence ‘in fact’.
Becker et al. (1998) found accounting flexibility to be associated with lower audit quality. It was theorised that Big N auditors would be less flexible in their audits than non-Big N audit firms, due to their greater resources for providing high quality audits. Big N auditors were found to be more capable of stopping accrual-based earnings management and thereby had greater audit quality (Becker et al., 1998). It was observed that clients of non-Big N auditors were more likely to have significantly higher levels of income increasing discretionary accruals. The higher levels of audit quality appear to be also caused by the greater reputational incentives Big N auditors possess (Francis, 2004). It is not only their greater resources that increase audit quality but also their perceived reputation and independence in appearance and fact.
Increased levels of audit quality leading to decreased levels in accrual-based earnings management are a highly beneficial trait. Nevertheless, Chi et al. (2011) found evidence to suggest that the reductions in accrual-based earnings management caused by Big N auditors was also coupled with an increase in real earnings management. Such results would be in line with Zang’s (2012) evidence of the earnings management trade-off. Big N auditors create greater audit quality that reduces managers’ ability to employ accrual-based earnings management, thus affecting the trade-off.
2.4.3 Sarbanes-Oxley Act 2002
The Sarbanes Oxley Act (SOx) was an act introduced in the U.S in 2002 in response to recent corporate accounting scandals; namely, Enron and WorldCom. In the aftermath of these scandals investor confidence in audit quality was uncertain at best. The scandals led to a lack in confidence of financial disclosures (Wang et al. 2010). The hope of the U.S. Congress was that tightened regulation would and increased penalties would restore trust within the public. Under the new SOx regulation board members are more accountable, the CFO and CEO must
personally sign off on financial reports, audit committees are required to be independent and financially knowledgeable. The introduction of SOx caused a substantive change not only within, the physical practice of financial reporting and auditing, but also to the mentality of accountants and other financial professionals.
3.0 Hypothesis Development
The primary function of financial reporting regulation and auditing as a profession is to increase reporting quality. Nonetheless, this may have the unintended consequence of increasing real earnings management. As a manager’s ability to use accrual-based earnings management is restricted one would expect real earnings management to increase. Prior literature indicates a strong relationship between auditor tenure and increased audit quality, I expect that as auditor tenure increases accrual-based earnings management will decrease and real earnings management will take its place.
H1a: Long-term auditor tenure is associated with lower levels of accrual-based earnings management.
H1b: Long-term auditor tenure there is associated with higher levels of real earnings management.
Big N auditors, long-term auditor tenure and increased regulation (SOx Act) have all shown to be a factor in the use of real earnings management (Chi et al., 2011). This appears to be due to such factors increasing the cost of engaging in accrual-based earnings management by making it more difficult or outright not possible and therefore, more costly. Prior research indicates that the employment of a Big N auditor significantly increases audit quality; therefore, I expect when a Big N auditor is employed the degree of accrual-based earnings management will be lower, and that real earnings management will be greater when compared to non-Bin N audit clients.
H2a: Big N audit clients have lower levels of accruals-based earnings management compared to non-Big N clients.
H2b: Big N audit clients have higher levels of real earnings management compared to non-Big N clients.
4.0 Research methodology
4.1 Accrual-based earnings management
!
To compute total discretionary accruals I will use the modified Jones model (1991). It is one of the most frequently used models to calculate accrual-based earnings management. It works by calculating total accruals as, non-discretionary accruals plus discretionary accruals. Then it generates an industry average based on two digit SIC code for non-discretionary accruals. The residual acts as a proxy and is deemed to be a firm’s discretionary accruals. These discretionary accruals are compared to the year and industry average.
Equation 1:
NOA / ATA = β0 + β1(1/ATA) + β2(ΔSales – ΔRec / ATA) + β3(GPPE / ATA ) + ε
Where:
NOA = Net operating accruals ATA = Average total assets ΔSales = Change in sales
ΔRec= Change in accounts receivable GPPE = Gross PPE
4.2 Real earnings management
!
This research will follow the examples of Roychowdhury (2006), Cohen et al. (2008), and Chi et al. (2011) in order to define proxies for calculating real earnings management. As done by Roychowdhury and Cohen et al., I will calculate abnormal levels of cash flows from operations, discretionary expenses, and production costs; to examine real earnings management.
The rational behind this technique is due to the manner in which firms may engage in real earnings management. Firms may adopt real earnings management by increasing their sales, which would be achieved through the use of price reductions or more lenient credit terms. By
introducing terms more favourable to the customer sales are increased, however, as these terms are less favourable to the firm although earnings in the current period will be higher, margins will be lower. This tends to lead to abnormally high production costs (Roychowdhury, 2006). Another technique that may be employed is lowering cost of goods sold by increasing production. If production is increased above normal levels, fixed costs will be spread out over more units. This increases the margin on products, however, has many negative effects as a firm now has more inventory that required leading to additional holding costs and therefore a reduction in cash flows from operations (Roychowdhury, 2006).
The final method outlined by Roychowdhury (2006) is to alter levels of discretionary expenditures such as R&D, SGA, and advertising. By reducing these expenses in the current period a firm’s current period cash flows will be increased. Nevertheless, as these expenses are necessary for the continued growth of a firm it will negatively affect future cash flows (Roychowdhury, 2006).
In order to calculate real earnings management I will follow Roychowdhury (2006) and express ‘normal’ cash flow from operations as a linear function of sales and change in sales in a year. I will run a cross sectional analysis for every year and industry estimate this model. As for accrual-based earnings management, the ‘normal’ values of the proxy categories will be compared to actual levels and the residual will be the proxy for real earnings management. Equation 2:
CFOt/At-1=α0+α1(1/At-1)+β1(St/At-1)+β2(ΔSt/At-1)+εt,
Where:
CFO = Cash flow from operations
At = the total assets at the end of period t
St = the sales during period t ΔSt = St−St−1
Abnormal cash flows from operations equal actual CFO – estimated CFO.
Roychowdhury expresses expenses as a linear function of contemporaneous sales. Following him the model for ‘normal’ COGS is:
Equation 3:
COGSt/At-1=α0+α1(1/At-1)+β(St/At-1)+εt.
As with expenses, I will estimate ‘normal’ inventory growth following Rochowdhury’s regression:
Equation 4:
ΔINVt/At-1=α0+α1(1/At-1)+β1(ΔSt/At-1)+β2(ΔSt-1/At-1)+εt,
Where:
ΔINVt = the change in inventory in period t.
As done by Rochowdhury I will estimate ‘normal’ production costs as PRODt = COGSt + ΔINV,
from the following regression: Equation 5:
PRODt/At-1=α0+α1(1/At-1)+β1(St/At-1)+β2(ΔSt/At-1)+β3(ΔSt-1/At-1)+εt.
The final proxy for real earnings management is discretionary expense; this will also be expressed as linear function of contemporaneous sales. The equation to find ‘normal’ discretionary expenses taken from Rochowdhury is:
Equation 6:
DISEXPt/At-1=α0+α1(1/At-1)+β(St-1/At-1)+εt.
DISEXPt = discretionary expenses in period t.
It is my intention to then compare these expected ‘normal’ values to the actual values within my sample.
4.3 Sample
!
The sample consists of firms on Compustat from the years 1987-2013. I am leaving out firms pre-1987, as they have no information for cash flow from operations; this follows Roychowdhury (2006) and Davis (2009). In accordance with prior research I will leave out banks and financial institutions (SIC code 6000-6500) and firms in regulated industries (SIC code 4400-5000) (Myers, 2003; Roychowdhury, 2006; and, Davis, 2009). This is due to increased regulation of these firms, which may falsely skew data.
4.4 Estimation models for accrual-based earnings management and real earnings management
4.4.1 Proxies of audit quality
!
In order to test the hypothesis outlined previously, two proxies of audit quality will be included: auditor type and audit firm tenure. Auditor size will be represented by a dummy variable, as was done by Becker et al. (1998). When the auditor is a Big N auditor, which is defined as KPMG, EY, PWC, Deloitte, Arthur Anderson, or any firm that merged into these, this value will be given a value of 1. Any other auditing firm and the variable will be given a value of 0. In accordance with Chi’s (2009) findings, I expect a big N auditor presence to be negatively (-) associated with accrual-based earnings management and positively (+) associated with real earnings management.
Johnson et al (2002), Myers et al. 2003, and Davis et al. 2009 all had varying results regarding the outcome of long-term auditor tenure. In this analysis auditor tenure will be measured according to number of years of continuous service as stated by Compustat. I will start measuring auditor tenure from 1974, as Compustat does not identify auditors prior to this. Like
Davis (2009) I will treat an auditor change due to a merger as a continuation of auditor tenure. Davis’s definition of long-term auditor tenure (15 years) is far higher than that of Johnson (2002) or Myers (2003), this is considerably higher than my sample average. Therefore, I predict my results to mimic those of Johnson (2002) and Myers (2003) and expect auditor tenure to be associated positively (+) with real earnings management and negatively (-) with accrual-based earnings management.
4.4.2 Control variables
!
As done by Roychowdhury (2006), I will use Size, as measured by the logarithm of the firm’s market value at the beginning of the year, as a control variable. As expressed by Watts and Zimmerman (1990), the managers of firms in financial distress have greater incentive to engage in earnings management, therefore, I predict Size to be negatively (-) associated with accrual-based and real earnings management. Cash flow will represent the second control variable; it is to be expressed as a firm’s cash flow from operations divided by average total assets. This is in line with Jones (1991); the intention is to use this model as a further predictor of firms in financial distress as a greater cash flow to debt ratio signals financial well-being. Based on Watts and Zimmerman’s (1990) positive accounting theory I predict cash flow to be negatively (-) associated with accrual-based and real earnings management. Trend represents a control for time. It is intended to control for years of particular financial distress, which may have raised the inventive for earnings management across the board. I have no prediction (-/+) for trend. As done by Roychowdhury (2006), I will include market to book ratio (MTB) as defined by market value of equity to the book value of equity. The intention of this is to factor in the growth potential of firms as this may also affect earnings management incentives. Firms with a greater growth potential should have less need for earnings management; therefore, I am expecting MTB to be negatively (-) associated with accrual-based and real earnings management. Net income (NI) is a further control variable taken from the Roychowdhury (2006) paper, the intention of which is to control for abnormal values within the estimation model that may contain errors correlated to performance. Firms with a higher net income should present lower incentives for earnings management, thus, I expect there to be a negative (-) association with earnings management. As Davis et al. (2009), found auditor tenure to be associated with reduced audit
quality before SOx but not after it, I will use a dummy variable to control for the affect this may have. If a firm year is pre-2002 the variable will be given a value of 0, if the a firm year is 2002 or later it will receive a value of 1. I have no prediction (-/+) for SOx.
4.4.3 Regression Models
!
The first estimation model has been adapted from Myers (2003) by including control variables from Rochowdhury (2006). To test accruals, discretionary/current accruals will be the independent variable. Tenure and auditor type will be the proxies for audit quality; while size, cash flow, trend, MTB, NI, and SOx will represent the control variables.
Accruals = a + + β1Tenure + β2AuditorType + β3Size + β4Cashflow + β5Trend + β6MTB + β7NI + β8SOx + !
Accruals = Discretionary Accruals or Current Accruals
Tenure = Number of years auditor has been working on client Auditor Type = 1 if auditor is Big N firm, 0 if it is not
Size = Logarithm of the market value at the beginning of year
Cash Flow = The firm’s cash flow from operations divided by average total assets Trend = A control for calendar year
MTB = Market to book ration, market value of equity to the book value of equity
NI = Net income
SOx = 1 if post-SOx, 0 if pre-SOx
The second estimation model has been adapted from Rochowdhury (2006) by including control variables from Myers (2003). To test real earnings management the REM index will be the independent variable. This is made up of abnormal production, abnormal discretionary expenses, and abnormal cash flows. As with the first estimation model tenure and auditor type will be the proxies for audit quality; while size, cash flow, trend, MTB, NI, and SOx will represent the control variables.
REM = a + β1Tenure + β2AuditorType + β3Size + β4Cashflow + β5Trend + β6MTB + β7NI + β8SOx + !
REM = Real earnings management
Abnormal production of inventory - Abnormal discretionary expenses - Abnormal cash flows
Tenure = Number of years auditor has been working on client Auditor Type = 1 if auditor is Big N firm, 0 if it is not
Size = Logarithm of the market value at the beginning of year
Cash Flow = The firm’s cash flow from operations divided by average total assets Trend = A control for calendar year
MTB = Market to book ration, market value of equity to the book value of equity
NI = Net income
SOx = 1 if post-SOx, 0 if pre-SOx
4.4.4 Expectations
Table 1 – Expectations of REM regression Table 2 – Expectations of Accruals regression
Variable Prediction Variable Prediction
REM + Accruals +
Ab_Prod + Tenure -
Ab_disexp - Auditor Type -
Ab_cfo - Size -
Tenure + Cash flow -
Auditor Type + Trend (+/-)
Size - MTB - Cash flow - NI - Trend (+/-) SOx (+/-) MTB - NI - SOx (+/-)
5.0 Empirical Findings
5.1 Descriptive Statistics
!
From the descriptive statistics we can see that the real earnings management index has a mean of 0.089 and a standard deviation of 1.253. This is considerably higher than those of Zang (2012) and is mostly related to the low quantity of assets. Accrual-based earnings management has a mean of -0.45, this is contrary to what I would have expected, it has been affected by a large standard deviation of 3.574. This indicates many firms are within the expected range, however, there is more variance that I had imagined. The auditor type dummy has a mean of 0.88 indicating that Big N auditors audited 88% of firms involved in sample. The mean of trend is 2002.9, showing that roughly half of the sample is pre-SOx and half post-SOx. This is further shown by the SOx dummy having a mean of 0.539, showing a very slight skew to post SOx data values. Size of 5.2 shows that a fair proportion of the sample are larger than the corresponding firm year average. A market to book ratio of 3.489 shows that most firms had a higher than average market value.
Table 3 – Descriptive statistics of sample
Descriptive statistics
Variable Obs Mean Std. Dev.
REM 16652 0.0888222 1.253049
Ab. Dis. Exp. 16652 -0.0323322 1.169712
Ab. Prod. 16652 1.612374 1.492346 Ab. CFO 16652 1.555884 1.335921 Accruals 16642 -0.1536521 3.574368 Tenure 16652 8.085335 5.586828 Auditor Type 16650 0.8899231 0.3129949 Cash Flow 16652 -0.0097432 0.9223515 Trend 16652 2002.983 6.908907 Size 16652 5.203008 2.653611 MTB 16645 3.489379 77.7301 NI 16652 171.9053 1127.459 SOx 16652 0.5394547 0.4984559
Table 4 – Sample by two-digit SIC codes
! SIC!CODE! Frequency! Percent! Cum.!Percent!
Metal!Mining! 10! 4! 0.02! 0.02! Food!&!Kindred!Products! 20! 420! 2.52! 2.55! Tobacco!Products! 21! 13! 0.08! 2.62! Apparel!&!Other!Textile!Products! 23! 54! 0.32! 2.95! Lumber!&!Wood!Products! 24! 7! 0.04! 2.99! Furniture!&!Fixtures! 25! 92! 0.55! 3.54! Paper!&!Allied!Products! 26! 64! 0.38! 3.93! Chemical!&!Allied!Products! 28! 1,762! 10.59! 14.6! Rubber!&!Miscellaneous!Plastic!Products! 30! 251! 1.51! 16.02! Leather!&!Leather!Products! 31! 47! 0.28! 16.3! Stone,!Clay,!&!Glass!Products! 32! 59! 0.35! 16.65! Primary!Metal!Industries! 33! 16! 0.1! 16.75! Fabricated!Metal!Products! 34! 252! 1.51! 18.26! Industrial!Machinery!&!Equipment! 35! 1,731! 10.4! 28.66! Electronic!&!Other!Electric!Equipment! 36! 2,451! 14.72! 43.38! Transportation!Equipment! 37! 371! 2.23! 45.6! Instruments!&!Related!Products! 38! 1,968! 11.82! 57.42! Miscellaneous!Manufacturing!Industries! 39! 380! 2.28! 59.7! Wholesale!Trade!T!Durable!Goods! 50! 266! 1.6! 61.3! Wholesale!Trade!TNondurable!Goods! 51! 130! 0.78! 62.08! Building!Materials!&!Gardening!Supplies! 52! 62! 0.37! 62.45! General!Merchandise!Stores! 53! 314! 1.89! 64.34! Food!Stores! 54! 150! 0.9! 65.24! Automotive!Dealers!&!Service!Stations! 55! 165! 0.99! 66.23! Apparel!&!Accessory!Stores! 56! 592! 3.56! 69.79! Furniture!&!Home!furnishings!Stores! 57! 300! 1.8! 71.59! Eating!&!Drinking!Places! 58! 716! 4.3! 75.89! Miscellaneous!Retail! 59! 684! 4.11! 80! Real!Estate! 65! 4! 0.02! 80.02! Holding!&!Other!Investment!Offices! 67! 95! 0.57! 80.59! Hotels!&!Other!Lodging!Places! 70! 7! 0.04! 80.63! Business!Services! 73! 2,782! 16.71! 97.34! Amusement!&!Recreation!Services! 79! 149! 0.89! 98.23! Health!Services! 80! 102! 0.61! 98.85! Engineering!&!Management!Services! 87! 87! 0.52! 99.37! NonTClassifiable!Establishments! 99! 105! 0.63! 100! ! Total! 16,652! 100! !
The industry sample contains thirty-six different industries and a total of 16,652 firm years. The largest representative within the sample is industry 73 (business services) with 2,780 firm years. Industry 65 (real estate) and 10 (metal mining) contain the fewest firm years, both only comprising of four firm years. As a whole manufacturing firms make up the largest group of the sample, including sic codes 20-39. This part of the sample consists of 9,938 firm years, which is equal to 59.6% of the entire sample. The second largest segment is that of sic codes 50-59, this segment is composed of firms in retail trade. It consists of 3,379 firm years, which represents 20.3%. Together these groups consist of 79.9% of the total sample. This is a very significant number and indicates that any results found by this analysis will be strongly directed towards manufacturing and retail firms.
5.2 Correlation
!
Table 5 & 6 represent the Pearson correlation between real earnings management, accrual-based earnings management, and related variables. For a Pearson correlation a large dataset is necessary for effective results, with 16652 firm years this data set will have sufficient data. If any variables are correlated that strongly their mutual existence in the regression is deemed to be ineffective. Within this matrix we see that only the REM index, abnormal cash flow from operations, abnormal discretionary expenses, abnormal production are linked beyond an acceptable point. Nevertheless, it is to be expected that there be strong correlation between the REM index and those values of which it consists. Furthermore, the strong relationship between abnormal, production and cash flow from operations is also expected, as a firm that is over producing to lower cost of goods sold will thereby be increasing abnormal cash flow from operations.
I had predicted that abnormal cash flow would be negatively correlated with the REM index but the reverse is true in these results. We see a further strong negative correlation between the SOx dummy and abnormal cash flows from operations. This may indicate a restricting effect caused by the SOx legislation and may be the cause of the unexpected direction. Abnormal, discretionary expenses and production, are negative and positive, respectively, as was expected. Market to book ratio and net income were intended to be indicators of financial distress, it was expected that firms performing poorly were more likely to engage in earnings management, however, both are positively correlated with REM. This indicates high performing firms are in fact more likely to engage in earnings management, while this was not expected it is compatible with Watts and Zimmerman’s (1990) positive accounting theory.
Real and accrual-based earnings management have a negative relationship significant at the 0.01 percent level, which signals the existence of the trade-off discussed by Zang (2012). We also see the existence of a Big N auditor to be correlated with an increase in real earnings management. Another point of interest is the relationship between Big N auditors and auditor tenure. From this matrix it would appear that firms are more likely to have a longer tenure with a Big N auditor than a smaller audit firm.
Table 5 - Correlation Matrix
REM Ab. Disexp. Ab. Prod. Ab. CFO Accruals Tenure Auditor Type
REM 1 Ab. Disexp. -0.9175*** 1 Ab. Prod. 0.2086*** -0.0255*** 1 Ab. CFO 0.0984*** -0.0435*** 0.9438*** 1 Accruals -0.0204*** 0.0310*** -0.0986*** -0.1182*** 1 Tenure 0.008 -0.016** -0.0487*** -0.0479*** -0.0025 1 Auditor Type 0.0205*** -0.0121 0.0041 -0.004 -0.0049 0.1726*** 1 Cash flow 0.0095 0.0217*** -0.009 -0.0378*** 0.0747*** 0.0329*** -0.0009 Trend -0.0073 -0.014* -0.0987*** -0.0911*** -0.0197** 0.3810*** -0.1284*** Size -0.0076 0.0621*** -0.0529*** -0.1063*** 0.0410*** 0.3039*** 0.1755*** MTB 0.0211*** -0.0113 0.0177** 0.0099 0.0012 0.001 0.0017 NI 0.0165** -0.0125 -0.0218*** -0.0289*** 0.0065 0.1602*** 0.0524*** SOx 0.0016 -0.0136* -0.0741*** -0.0723*** -0.0151* 0.3013*** -0.1259***
*. Correlation is significant at the 0.1 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). ***. Correlation is significant at the 0.01 level (2-tailed).
Table 6 - Correlation Matrix Cont.
Cash flow Trend Size MTB NI SOx
REM Ab. Disexp. Ab. Prod. Ab. CFO Accruals Tenure Auditor Type Cash flow 1 Trend -0.0165** 1 Size 0.1622*** 0.2038*** 1 MTB 0.0119 0.0027 0.0142* 1 NI 0.0305*** 0.0785*** 0.3435*** 0.0065 1 SOx -0.0135* 0.8441*** 0.1800*** 0 0.0709*** 1
*. Correlation is significant at the 0.1 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). ***. Correlation is significant at the 0.01 level (2-tailed).
5.3 Regression Results
In this regression the dependent variable is the proxy for accrual-based earnings management. The regression coefficients are estimated using the following model:
Accruals = a + + β1Tenure + β2AuditorType + β3Size + β4Cashflow + β5Trend + β6MTB + β7NI + β8SOx + !
In this regression accrual-based earnings management is the dependent variable. We see that auditor tenure has a negative effect on accrual-based earnings management as was predicted, however, it is not significant. The presence of a Big N auditor is also not significant, however is negative as predicted. It appears that the control for trend is showing a strong and significant positive link to accrual-based earnings management. This indicates that something is causing accrual-based earnings management to become more prevalent throughout time. Cash flow is operating in a positive manner, which is opposite to what was expected and is significant. Size is acting as a negative and is also significant.
The indicator of audit tenure supports Myers’s (2005) assertions that extended audit tenure reduces accrual-based earnings management. The indicator of auditor type is consistent with the results of Chi et al. (2011) and indicates that higher audit quality reduces accrual-based earnings management.
Table 7 – Accruals regression results
Accruals Regression
Accruals Coef. Std. Err t P>|t| [95% Conf. Interval]
Auditor Tenure -0.0020675 0.0056553 -0.37 0.715 -0.0131524 0.0090174 Auditor Type -0.1664233 0.0930275 -1.79 0.074 -0.348767 0.0159205 Cash flow 0.2628724 0.0304189 8.64 0 0.2032481 0.3224967 Size -0.0155474 0.0077526 -2.01 0.045 -0.0307433 -0.0003516 Trend 0.0552497 0.011886 4.65 0 0.0319519 0.0785475 MTB -3.08E-06 0.0003554 -0.01 0.993 -0.0006996 0.0006935 NI -0.0000199 0.0000261 -0.76 0.446 -0.0000712 0.0000313 SOx 0.0250592 0.1035479 0.24 0.809 -0.1779058 0.2280242 _cons 30.85765 15.47728 1.99 0.046 0.5205209 61.19477
In this regression the dependent variable is the proxy for real earnings management. The regression coefficients are estimated using the following model:
REM = a + β1Tenure + β2AuditorType + β3Size + β4Cashflow + β5Trend + β6MTB + β7NI + β8SOx + !
In this regression real earnings management is the dependent variable. We see that auditor tenure has a positive effect on real earnings management as was predicted, however, it is not significant, 0.26. The presence of a Big N auditor is significant at the 0.009 percent level; also positive in direction, as predicted. Cash flow is operating in a positive manner, which is opposite to what was expected; nevertheless, it is not significant. Size is acting as a negative and is significant. This indicates that size is a better proxy for financial health than cash flow.
These results are consistent with those of Chi et al. (2011), the proxies for audit quality are acting positively on the REM index, thus, materialising as an incentive to favour increase real earnings management.
Table 8 – REM regression results
REM Regression
REM Coef. Std. Err t P>|t| [95% Conf. Interval]
Auditor Tenure 0.0022354 0.0019748 1.13 0.258 -0.0016355 0.0061062 Auditor Type 0.0845661 0.0324711 2.6 0.009 0.0209192 0.148213 Cash flow 0.0159371 0.010625 1.5 0.134 -0.0048891 0.0367633 Size -0.0104805 0.00415 -2.53 0.012 -0.0186149 -0.0023461 Trend -0.0052542 0.0027068 -1.94 0.052 -0.0105598 0.0000515 MTB 0.0003412 0.0001241 2.75 0.006 0.0000979 0.0005845 NI 0.0000237 9.13E-06 2.59 0.01 5.76E-06 0.0000416 SOx 0.0699594 0.0361544 1.94 0.053 -0.0009071 0.1408259 _cons 10.53064 5.403904 1.95 0.051 -0.0615845 21.12287
The control for trend has a negative indicator while SOx is positive, I had expected the two to correlate as SOx was one of most pervasive events within the last 20 years. In order to further test this, I will separately re-run the regressions for pre-SOx and post-SOx firm years.
In this regression the dependent variable is the proxy for accrual-based earnings management. It is run once for pre-SOx years and a second time for post-SOx years. The regression coefficients are estimated using the following model:
Accruals = a + + β1Tenure + β2AuditorType + β3Size + β4Cashflow + β5Trend + β6MTB + β7NI + β8SOx + !
When we look at the accruals regression we see that all significant variables have remained the same except for auditor tenure. Before SOx legislation was implemented tenure is associated with a significant positive increase in accrual-based earnings management. Nevertheless, in the post-SOx time period we see no significant relationship between tenure and abnormal accruals. This is consistent with the study conducted by David (2009). The auditor type variable is negative in both time periods, however it is not significant, just as was the case in the original regression.
Table 9 – Accruals regression pre-SOx
Accruals Regression - Pre SOx
Accruals Coef. Std. Err t P>|t| [95% Conf. Interval]
Auditor Tenure 0.0070769 0.0020215 3.5 0 0.0031142 0.0110396 Auditor Type -0.0421884 0.027976 -1.51 0.132 -0.097029 0.0126522 Cash flow 0.1893274 0.0091422 20.71 0 0.1714061 0.2072487 Trend -0.008317 0.001624 -5.12 0 -0.0115005 -0.0051336 Size 0.0080048 0.0029362 2.73 0.006 0.0022491 0.0137605 MTB -0.0000124 0.0000662 -0.19 0.851 -0.0001421 0.0001173 NI 3.32E-06 0.0000124 0.27 0.79 -0.000021 0.0000277 SOx 0 (omitted) _cons 16.46751 3.240727 5.08 0 10.11479 22.82022
Table 10 – Accruals regression post-SOx
Accruals Regression - Post SOx
Accruals Coef. Std. Err t P>|t| [95% Conf. Interval]
Auditor Tenure -0.0049071 0.0087127 -0.56 0.573 -0.021986 0.0121717 Auditor Type -0.2244343 0.1514022 -1.48 0.138 -0.5212171 0.0723486 Cash flow 0.2882656 0.0495755 5.81 0 0.1910863 0.385445 Trend -0.0326656 0.0185865 -1.76 0.079 -0.0690993 0.0037681 Size 0.0976441 0.0223341 4.37 0 0.0538642 0.141424 MTB 0.0001531 0.0010768 0.14 0.887 -0.0019577 0.0022638 NI -0.0000382 0.0000383 -1 0.318 -0.0001132 0.0000368 SOx 0 (omitted) _cons 65.10442 37.31906 1.74 0.081 -8.049462 138.2583
In this regression the dependent variable is the proxy for real earnings management. It is run once for pre-SOx years and a second time for post-SOx years. The regression coefficients are estimated using the following model:
REM = a + β1Tenure + β2AuditorType + β3Size + β4Cashflow + β5Trend + β6MTB + β7NI + β8SOx + !
When we look at the real earnings management regression split between pre and post SOx periods we see the control variables cash flow, size, MTB, and NI all behave in the same manner. In the pre-SOx period auditor tenure has a positive indicator and is significant; the auditor type variable is not significant. In the post-SOx period we see that tenure is not significant while the auditor type variable is significant and positively associated with real earnings management. This indicates that pre-SOx any additional tenure caused an increase in real earnings management, however, after more stringent regulation was enacted this no longer becomes the case. The evidence toward Big N auditors being positively associated with real earnings management is consistent with Chi et al. (2011) and Zang et al. (2012).
Table 11 – REM regression pre-SOx
REM Regression - Pre SOx
REM Coef. Std. Err t P>|t| [95% Conf. Interval]
Auditor Tenure 0.0079 0.0033851 2.33 0.02 0.0012643 0.0145358 Auditor Type -0.0224398 0.0467978 -0.48 0.632 -0.1141762 0.0692966 Cash flow 0.0103622 0.0153155 0.68 0.499 -0.0196605 0.0403848 Trend -0.0121845 0.0027193 -4.48 0 -0.0175151 -0.0068539 Size -0.0083048 0.0049156 -1.69 0.091 -0.0179408 0.0013312 MTB 0.0002805 0.0001108 2.53 0.011 0.0000632 0.0004977 NI 0.0000363 0.0000208 1.74 0.081 -4.53E-06 0.0000771 SOx 0 (omitted) _cons 24.42145 5.426494 4.5 0 13.78403 35.05886
Table 12 – REM regression post-SOx
REM Regression - Post SOx
REM Coef. Std. Err t P>|t| [95% Conf. Interval]
Auditor Tenure -0.0000313 0.0025736 -0.01 0.99 -0.0050762 0.0050136 Auditor Type 0.1364731 0.0447151 3.05 0.002 0.0488214 0.2241249 Cash flow 0.017487 0.0146471 1.19 0.233 -0.0112247 0.0461988 Trend 0.0062724 0.0054899 1.14 0.253 -0.0044891 0.0170339 Size -0.0126317 0.006598 -1.91 0.056 -0.0255653 0.0003018 MTB 0.000564 0.0003181 1.77 0.076 -0.0000596 0.0011877 NI 0.000022 0.0000113 1.94 0.052 -1.90E-07 0.0000441 SOx 0 (omitted) _cons -12.55961 11.02299 -1.14 0.255 -34.16718 9.047958
Table 13 – Hypothesis conclusions
Hypothesis Whole Period Pre-SOx Post-SOx
H1a Reject Reject Reject
H1b Reject Accept Reject
H2a Reject Reject Reject
Table 14 – Observed direction of variables*
Observed direction of
variables Whole Period Pre-SOx Period Post-SOx Period
Variables Accruals REM Accruals REM Accruals REM
Tenure - + + + - - Auditor Type - + - - - + Cash flow + + + + + + Size - - - + Trend + - + - + - MTB - + - + + + NI - + + + - +
SOx + + N/A N/A N/A N/A
*Significant variables in red
The results show that there are significant differences in how auditor tenure and auditor type influence earnings management in pre and post-SOx periods. When looking at the whole period together we only find significant results to suggest that the presence of a Big N auditor increases the prevalence of real earnings management. Although the directions support the hypotheses, none of the other results concerning the whole period are significant.
In the pre-SOx period the results indicate that longer auditor tenure increased both real earnings management and accrual-based earnings management. This is contrary to my expectations but supports the results of Myers (2009). The presence of a Big N auditor has a negative indicator for both real and accrual-based earnings management, yet neither is significant.
In the post-SOx period the results suggest that tenure reduces both accrual-based and real earnings management, these are however also not significant. The auditor type variable is negative for abnormal accruals and positive for REM, nevertheless only the latter is significant. !
6.0 Conclusion
This paper examines the effect of auditor size and tenure on accrual-based and real earnings management. More specifically, how auditor size and tenure affect the trade-off between accrual-based and real earnings management. Furthermore, the potential mitigating effect the Sarbanes-Oxley Act 2002 has had on earnings management is examined.
Evidence is found that suggests SOx significantly altered the sphere of earnings management. Within the separate periods vastly different results were seen. In the pre-SOx period the results indicate that longer auditor tenure is positively associated with both forms of earnings management. While the presence of a Big N auditor is negatively associated with accrual-based earnings management and positively associated with real earnings management. In the post-SOx period auditor tenure is negatively associated with both forms of earnings management. While the presence of a Big N auditor is negatively associated with accrual-based earnings management and positive with real earnings management.
Overall the results suggest that prior to the implementation of SOx long term auditor tenure resulted in higher levels of earnings management while after the opposite is true. This suggests that increased regulation has increased the necessity for auditors to detect earnings management and the increase knowledge of a firm that comes with increased tenure is leading to higher audit quality. Furthermore, the presence of a Big N auditor is associated with less accrual-based earnings management and positively with real earnings management. This is consistent with Zang’s (2012) theory that firms trade-off between the two forms depending on their inherent cost. It is possible that due to Big N’s greater resources and increased audit quality accrual-based earnings management is not possible and firms therefore resort to real earnings management. Nevertheless, due to limitations regarding the significance of results only limited conclusions can be drawn from this research.
The estimation models are a significant limitation to this research. There are many factors that affect the use of accrual-based and real earnings management and it is only possible to incorporate a certain number of them within the estimation models. From the lack of significant
variables and inconsistency with the direction of variables, it shows that these models are leaving out many factors of earnings management.
The scope and time restraints that were imposed on this project present a noteworthy limiting factor. Due to its nature there were constraints that resulted in the statistical analysis not being as in depth as would be desired. Due to this the results are not as consistent or significant as would otherwise be possible.
Accrual-based and real earnings management are formed of accruals choices and business decisions, both present significant challenges to being identified. It is in their nature and their very purpose to be deceptive. In this study I relied on the Jones (1991) model and the model presented by Roychowdhury (2006) to identify earnings management, therefore I am relying entirely on their ability to identify discretionary accruals and abnormal cash flows. This presents the obvious difficulty of verifying their validity.
Furthermore, due to the substitutional nature of earnings management it is particularly difficult to gauge the extent of earnings management employed and the reasoning for substituting between accrual-based and real earnings management. Real earnings management must be engaged in throughout the financial year, in contrast accrual-based earnings management must not be decided upon until the end of financial year. This means that a manager has both the knowledge of how near to his targets he is and how effective his real earnings management was and therefore uses one as a base for deciding upon the other.
A suggestion for future research would be to focus on the effect of SOx. In my results one can clearly see a distinct difference in pre and post-SOx periods, this was also observed by Myers (2009). It is clear that the SOx act significantly changed the reporting environment, business culture, and managers’ incentives; these are all likely to have a considerable effect on the earnings management trade-off.
A further suggestion for future research would be to investigate negative real earnings management. Accrual-based big bath accounting has been substantially documented and
investigated. If accrual-based earnings management is utilised for big bath accounting it would be reasonable to assume that real earnings management would also be. Nevertheless, all prior literature on real earnings management has focused on positively geared earnings management (Roychowdhury, 2006; Chi, 2011; Zang 2012).
Prior literature focuses on auditor tenure and how this affects earnings management. There have been no studies that investigate what effect auditor rotation has on the earnings management trade-off. If as stated by Zang (2012) managers switch between accrual-based and real earnings management depending on their inherent costliness it would logical that if the auditor would change so would this incentive. If this theory holds true measuring the trade-off before and after auditor rotation would provide interesting insights into the effect of auditor rotation.
7.0 Bibliography
Arens, A. A., J. R. Elder and M. Beasley (2011). Auditing and Assurance Services, and An integrated approach. Pearson Education, Limited, 14th edition.
Becker, C. L, DeFond, M.L., Jiambalvo, J., and Subramanyam K.R.(1998). The effect of audit quality on earnings management. Contemporary Accounting Research , 15 (1), 1-24.
Carey, P. J., and R. Simnett. 2006. Audit partner tenure and audit quality. The Accounting
Review 81 (3):653–676.
Cazavan-Jeny, A., Jeanjean, T., Joos, P., (2011) ‘Accounting choice and future performance: The case of R&D accounting in France’, Journal of Accounting and Public Policy, Elsevier, vol. 30(2), 145-165, March.
Chan, L. H., Chen, K. W., Tai Yuan, C., & Yangxin, Y. (2015). Substitution between Real and Accruals-Based Earnings Management after Voluntary Adoption of Compensation Clawback Provisions. Accounting Review, 90(1), 147-174.
Cohen, D.A., Dey. A., and Lys. T.Z. (2008). Real and Accrual-based earnings management in the Pre- and Post-Sarbanes-Oxley Perio. The Accounting Review , 83 (3), 757-787.
Chi, W. Lisic, L., & Pevzner, M. (2011). Is enhanced audit quality associated with greater real earnings management? Accounting Horizons , 25 (2), 315-335.
Davis, L.R., Soo, B.S., Trompeter, G.M. (2009). Auditor tenure and the ability to meet or beat earnings forecast. Contemporary Accounting Research Vol. 26 No. 2 , 517-48.
DeAngelo, L. E. (1981). Auditor size and audit quality. Journal of accounting and economics, 3(3), 183-199.
Ewert, R. and Wagenhofer., R. (2005). The effects of tightening accounting standards to restrict earnings management. 80 (4), 1101-1124.
Johnson, E., Khurana, I.K., Reynolds, J.K. (2002). Audit-Firm Tenure and the quality of Financial reports. Contemporary Accounting Research , 19 (4), 637-660.
Jones, J. (1991). Earnings Management During Import Relief Investigations. Vol. 29 No. 2, pp 193-228.
Healy, P. (1985). The effect of bonus schemes on accounting decisions. Journal of accounting and economics, 7 (1985), 85-107
Healy and Wahlen, 1999 P.M. Healy, J.M. Wahlen. A review of the earnings management literature and its implications for standard setting. Accounting Horizons, 13 (1999), pp. 365–38
Litt, B., Sharma, D. S., Simpson, T., & Tanyi, P. N. (2014). Audit Partner Rotation and Financial Reporting Quality. Auditing: A Journal Of Practice & Theory, 33(3), 59-86.
Myers, J.N., Myers, L.A., and Omer, T.C. (2003). Exploring the term of the auditor-client relationship and the quality of earnings: A case for mandatory auditor rotation? The Accounting Review , 78 (3), 779-799.
Roychowdhury, S. 2006. Earnings management through real activities manipulation. Journal of Accounting and Economics 42 (2006) 335-370.
Tan, Jason. 2013. Real earnings management: the impact of audit quality and PCAOB.
University of Rotterdam Thesis Repository.
Wang, H. Davidson. W.N., & Wang, X. (2010). The Sarbanes-Oxley Act and CEO turnover, and risk aversion. . The Quarterly Review of Economics and Finance, 50, 367-376.
Watts, R.L., and Zimmerman, J.L. (1978). Towards a positive theory of the determination of accounting standards. The Accounting Review, 53 (1), 112-134.
Watts, R. L., & Zimmerman, J.L. (1990). Positive Accounting Theory: a ten-year perspective.
The Accounting Review, 65, 131-156.
Zang, A.Y. (2012). Evidence on the trade-off between real activities manipulation and accrual-based earnings management. The Accounting Review, 87 (2), 675-703.
