MSc Accountancy & Control, Accountancy track Faculty of Economics and Business, University of Amsterdam
Types of earnings management and the effect of corporate
governance practices
Do firms conduct efficient or opportunistic earnings management and what are
the effects of corporate governance practices on those types of earnings
management?
Final version
Tim Vervaat (5941474) 16th of June 2015
Supervisor: Georgios Georgakopoulos
Statement of Originality
This document is written by student Tim Vervaat 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.
The faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
1. Introduction ... 5
1.1 Background ... 5
1.2 Research question ... 6
1.3 Motivation ... 6
2. Literature review and hypotheses ... 7
2.1 Opportunistic earnings management ... 7
2.2 Efficient earnings management ... 8
2.3 Corporate governance... 9 2.3.1 Audit quality ... 10 2.3.2 Audit committee ... 10 2.3.3 Board independence ... 11 2.4 Hypotheses ... 12 3. Research methodology ... 14 3.1 Research model ... 14 3.2 Definition of variables ... 19 3.2.1 Future profitability ... 19 3.2.2 Earnings management ... 20
3.2.3 Corporate governance practices ... 22
3.2.3.1 Auditor’s size ... 22
3.2.3.2 Board independence ... 22
3.2.3.4 Audit committee ... 22
4. Sample selection ... 23
5. Results ... 25
5.1 Opportunistic versus efficient earnings management ... 25
5.2 Corporate governance practices ... 28
6. Conclusion and suggestions for future research ... 32
6.1 Conclusion ... 32
6.2 Implications ... 32
6.3 Limitations of the study ... 33
6.4 Suggestions for future research ... 33
7. References ... 34
8. Appendix ... 37
Abstract
The purpose of this study is to investigate whether publicly listed companies on the S&P 500 index conduct efficient or opportunistic earnings management and to examine the effects of the corporate governance practices board independence, audit quality and audit committee on those types of earnings management.
Using two separate regression models, this study finds evidence that American firms listed on the S&P 500 index tend to conduct efficient earnings management. These findings are consistent with most views on earnings management in the United States that earnings management tends to be efficient. Firms with higher audit fees as a proxy of higher audit quality do not use more efficient earnings management than firms with small audit fees, a proxy of lower audit quality. Firms with a higher percentage of inside directors in the audit committee will use more opportunistic earnings management and, in turn, the governance practice of more outside directors in audit committees will lead to more efficient earnings management.
This study finds inconsistent evidence in that a higher proportion of independent members on the board of directors will lead to a better, more significant, effect of discretionary accruals on future profitability.
1. Introduction 1.1 Background
According to Siregar and Utama (2008) there are two types of earnings management: efficient earnings management (i.e., to improve earnings informativeness in communicating private information) and opportunistic earnings management (i.e., management reports earnings opportunistically to maximize his/her utility). Multiple studies, for example Burgstahler and Dichev (1997), Balsam et al. (2002) and Jiraporn et al. (2008) provide
evidence that is consistent with the opportunistic perspective. Jiraporn et al. (2008) report that “The scandals at Enron, Worldcom and elsewhere have generated a public perception that earnings management is utilized opportunistically by firm managers for their own private benefits rather than for the benefits of the stockholders. Reinforcing this negative public perception on earnings management is the fact that regulators have devised a number of measures for the purpose of combating earnings management. For instance, the Sarbanes–Oxley Act, which requires certain board members to possess financial sophistication, is the result of an ongoing attempt by Congress to mitigate earnings management. Even the NASDAQ has issued new guidelines requiring its listed firms to have financially literate audit committees” (Jiraporn et al., 2008: 622-623).
Furthermore, Jiraporn et al. (2008) mention that the misalignment of managers’ and shareholders’ incentives might encourage managers to use the flexibility by the Generally Accepted Accounting Principles (GAAP) to manage their income opportunistically. This would create a distortion in the reported earnings. Thus, there seems to be a prevalent perception that earnings management is opportunistic in nature. Opportunistic earnings management is harmful to stockholders and the public because management chooses to maximize their utility over informing stakeholders.
On the other hand, Subramanyam (1996) and Krishnan (2003) find evidence that the behavior of discretionary accruals (a proxy for earnings management) is consistent with the perspective of efficient earnings management because discretionary accruals have a positive relationship with future profitability (Siregar and Utama, 2008). Furthermore Jiraporn et al. (2008) state that a number of academic studies including Subramanyam (1996) and Krishnan (2003) have argued that earnings management might be efficient and beneficial because it
potentially enhances the information value of earnings. Managers can implement discretion over earnings to communicate private information to stockholders and the public. This implies that earnings management does not have to be damaging to stockholders and the public and can be done in an efficient and beneficial manner.
Consequently, Jiraporn et al (2008) state that “earnings management can be viewed as either opportunistic or beneficial” (Jiraporn et al., 2008, p.623). Siregar and Utama (2008) use the word efficient instead of beneficial but the connotation is identical.
As cited by Siregar and Utama (2008), Krishnan (2003) finds evidence that the corporate governance proxy external auditing plays an important role in constraining
opportunistic earnings management. Other corporate governance practices are for example the proportion of an independent board and the existence of an audit committee (Siregar and Utama, 2008). Therefore corporate governance mechanisms could also influence the type of earnings management a firm uses.
1.2 Research question
This study will investigate whether firms conduct efficient or opportunistic earnings management. The underlying principle of this study is to make a distinction between the opportunistic and efficient utilizations of earnings management. In addition, the research will study the effects of three different corporate governance proxies on those types of earnings management. Those three corporate governance proxies consist of audit quality (i.e. Big Four auditors versus non Big Four auditors), the proportion of an independent board and the existence of an audit committee.
1.3 Motivation
There has been widespread research on earnings management; however regarding the type of earnings management there is little to no research. Siregar and Utama (2008) argue that the existing literature on earnings management only observes the effect of institutional investors and corporate governance practices on the scale of earnings management. This study will focus on the effects of corporate governance practices on the type of earnings
management. Furthermore, there is a gap in the literature concerning the corporate governance factors that manipulate the choice of the earnings management types. For
example, Siregar and Utama (2008) claim that there still has not been a perfect proxy for audit quality.
Siregar and Utama (2008) conducted a similar research on the effect of corporate governance practices on types of earnings management regarding Indonesia but that was a relatively small sample size limited to Indonesian companies. Jiraporn et al. (2008) also attempted to distinguish between opportunistic and beneficial uses of earnings management. However, that study focused on the relation between the severity of agency costs and the extent of earnings management rather than the effects of corporate governance proxies on the types of earnings management.
This study would contribute to existing literature in several ways. First, it will provide a large scale analysis of the effects of corporate governance practices on both opportunistic and efficient earnings management, a study that hasn’t been done before in a first world country and without restrictions regarding overall scale. Siregar and Utama (2008) claim that a suggestion for future research could be to research the type of earnings
management on a larger scale and longer period to see how the corporate governance practices influence earnings management. Second, I will hope to find consistent results on whether corporate governance practices provide monitoring mechanisms that constrain
opportunistic earnings management since the existing literature has ambiguous results. On the one hand, studies by Burgstahler and Dichev (1997) and Balsam et al. (2002) claim that opportunistic earnings management occurs whereas on the other hand according to
Subramanyam (1996), Gul et al. (2000) and Krishnan (2003), efficient earnings management occurs in firms.
2. Literature review and hypotheses 2.1 Opportunistic earnings management
As cited by Siregar and Utama (2008), William Scott (1997) states that management reports earnings opportunistically to maximize his or her utility in opportunistic earnings management. According to Siregar and Utama (2008), several studies find evidence consistent with the opportunistic perspective of earnings management. Burgstahler and Dichev (1997) find that management engages in earnings management to avoid reporting losses or a decline in earnings. Furthermore, Balsam et al. (2002) find a negative relationship between
unexpected discretionary accruals and stock returns around the earnings announcement date. This result indicates that the market views discretionary accruals as an opportunistic type of earnings management (Siregar and Utama, 2008). Healy (1985), while researching the impact of bonus schemes on the selection of accounting principles, finds sufficient proof that
confirms the hypothesis that executives manage earnings downwards when their bonuses are at maximum (Healy, 1985). In addition, DeAngelo and DeAngelo (1989) find that managers who face the possibility of losing their jobs often use their accounting judgment to give the perception of a better own performance to voting stockholders. Also, Dechow and Sloan (1991) find evidence that current CEOs who tend to be in their final years of employment reduce R&D spending in order to increase reported earnings.
Another way of earnings management is when firms need to meet expectations set by capital markets. According to Jiraporn et al. (2008), Teoh et al. (1998) report that income-increasing abnormal accruals can be identified shortly before initial public offerings.
Furthermore, there is high informational asymmetry between the investors and issuers at the time of the offering (Jiraporn et al., 2008). Another area where opportunistic earnings management can be detected is stock-for-stock mergers (Jiraporn et al., 2008). As cited by Jiraporn et al. (2008), Ericksen and Wong (1999) find that in the quarters leading up to a merger, acquiring firms manage earnings upward in an attempt to increase their stock prices.
Concluding, it is evident that managers use opportunistic earnings management
throughout the years. Opportunistic earnings management is notable is several areas including capital markets and stock-for-stock mergers.
2.2 Efficient earnings management
As cited by Siregar and Utama (2008), Scott (1997) states that efficient earnings management occurs when management desires to improve earnings informativeness by communicating private information. Furthermore, according to Jiraporn et al. (2008)
“Some studies argue that managers exercise discretion over earnings to enhance earnings' information by allowing communication of private information (Healy and Palepu, 1993), (Holthausen, 1990), (Watts and Zimmerman, 1986) for example. This argument gains empirical support in Subramanyam (1996), who examines whether the stock market prices discretionary accruals” (Jiraporn et al., 2008, p.624).
Subramanyam (1996) concludes that discretionary accruals are efficient because they have a positive, significant relationship with future profitability. This positive relationship describes the ability that discretionary accruals have to communicate information about a firm's future profitability to the public. Gul et al. (2000) and Krishnan (2003), following Subramanyam (1996), also find consistent evidence of the positive relationship between discretionary accruals and future profitability.
In addition to the studies of Subramanyam (1996), Gul et al. (2000) and Krishnan (2003), Louis and Robinson (2005) also support the theory that earnings management adds information value: “Prior studies suggest that managers use their reporting discretion to signal private information. However, because managers are often assumed to use their discretion to mislead investors, discretionary accruals might be regarded as opportunistic. We posit that combining the accrual signal with other signals maybe an effective means of communicating private information” (Louis and Robinson, 2005, p.361).
Combining the literature on both opportunistic and efficient earnings management, it is prevalent that the empirical evidence in the existing literature is rather ambiguous. Research has shown that both opportunistic, for example in the studies by Burgstahler and Dichev (1997) and Balsam et al. (2002), and efficient earnings management, as mentioned by Subramanyam (1996), Gul et al. (2000) and Krishnan (2003), occurs in firms.
2.3 Corporate governance
Corporate governance is defined as “The framework of rules and practices by which a board of directors ensures accountability, fairness, and transparency in a company’s
relationship with its all stakeholders (financiers, customers, management, employees, government, and the community)” (Business Dictionary, 2015).
Sarkar et al. (2008) state that corporate governance can be classified into two categories: internal and external mechanisms. The internal mechanisms are those related to board
structure, executive compensation, and ownership structure and monitoring by large shareholders (Sarkar et al., 2008). External mechanisms relate to the market for corporate control, for example the takeover market and protection offered to shareholders via the legal system in which the firm operates (Sarkar et al., 2008). Furthermore, Sarkar et al. (2008) state that “In widely held corporations with separation of ownership and control, the primary objective of governance mechanisms, both internal and external, is to align the interests of managers with that of shareholders. Managers undertake earnings management for several
reasons, including capital markets, contracting and regulatory motivations” (Sarkar et al., 2008, p.5).
2.3.1 Audit quality
There are several studies examining the relation between earnings management and audit quality. Siregar and Utama (2008) mention that Becker et al. (1998) and Francis et al. (1999) find that discretionary accruals in firms audited by then Big Six auditors are less than those in firms audited by non-Big Six auditors. Becker et al. (1998) state that
“Also, consistent with earnings management, we find that the mean and median of the absolute value of discretionary accruals are greater for firms with non-Big Six auditors. This result also indicates that lower audit quality is associated with more "accounting flexibility" “(Becker et al, 1998, p.1).
Nowadays this proxy should be examined by testing Big Four auditors versus non Big Four auditors. Krishnan (2003) finds that discretionary accruals in firms audited by then Big Six auditors have a higher positive relationship with future profitability than that of firms audited by non-Big Six auditors. Non-Big Six auditors, in fact, have a negative relationship with future profitability. This implies that opportunistic earnings management occurs in firms with lower audit quality (Krishnan, 2003). At the moment this proxy should be studied by testing Big Four auditors versus non Big Four auditors. Krishnan (2003) furthermore states that auditing has a significant role in constraining opportunistic earnings management.
However, according to Siregar and Utama (2008), Sandra and Kusuma (2004) find no significant evidence that audit quality has a moderating effect on the relationship between earnings management and stock returns in Indonesia. This could indicate that auditor size (Big Four versus non Big Four) does not have to be a good proxy for audit quality in Indonesia.
2.3.2 Audit committee
Following Siregar and Utama (2008),
“An audit committee should have at least three members, one of whom must be an independent board member, who will act as chairman of the audit committee, and the other members should be independent external parties, at least one of whom has accounting and/or finance skills” (Siregar and Utama, 2008, p.5).
According to Sarkar et al. (2008), Dechow et al. (1996) find that firms with higher proportion of independent directors, smaller boards, and with an audit committee have lower earnings manipulation. Additionally, Peasnell et al. (2005) provide evidence of independent directors reducing earnings manipulation and their effectiveness in doing so increases when the board appoints an audit committee (Sarkar et al., 2008). According to Siregar and Utama (2008), Chtourou et al. (2001) find that income-increasing earnings management has a negative relationship with an audit committee composed only of independent directors that meet more than twice a year. Furthermore, Klein (2002) finds that an independent audit committee and active audit committee are associated with lower levels of discretionary accruals for U.S. firms (Siregar and Utama, 2008).
Furthermore, Sarkar et al. (2008) mention that Klein (2002) and Xie et al. (2003) “have found that the level of earnings management is inversely related to the extent of audit
committee independence” (Sarkar et al., 2008, p.7). Klein states that independent audit committees have a constraining effect on earnings manipulation particularly when a majority of directors are independent (Sarkar et al., 2008). Xie et al. (2003) find that more active rather than independent audit committees reduce the extent of earnings management in a firm (Sarkar et al., 2008).
2.3.3 Board independence
According to Siregar and Utama (2008),
“The number of independent board members must be ≥30% of the board size, and the independent board members should be:
– Individuals who have no affiliated relationship with controlling shareholders in related firms.
– Individuals who have no affiliated relation with company managers and/or board members of related listed firms.
– Individuals who are not engaged as officers in other firms affiliated with related listed firms.
– Individuals who understand stock exchanges rules” (Siregar and Utama, 2008, p.5).
Dechow et al. (1996) find that firms where the CEO is also the chairman of the board of directors are more likely to be subject to accounting enforcement actions by the SEC for alleged GAAP violation (Siregar and Utama, 2008). Peasnell et al. (2000) find that the probability of income-increasing accruals decreases as the proportion of outside-directors increases for a sample of U.K. firms (Siregar and Utama, 2008). Chtourou et al. (2001) find that an independent board is constraining earnings management activity (Siregar and Utama, 2008). Finally, Sarkar et al. (2008) conclude that
“Focusing specifically on the impact of board characteristics on the incentives to manage earnings, one finds that corporate governance codes around the world emphasize the fiduciary role of the board of directors in curbing opportunistic earnings manipulation and in ensuring that earnings figures convey true information about firm operations (Young, 2000). In this regard, the existing literature largely supports the contention that board structure matters in earnings management; independence of the board of directors as well as the independence of audit committees which are created with a subset of directors on the board with the primary responsibility of monitoring the financial reporting process of a firm, are found to constrain earnings management” (Sarkar et al., 2008, p.6).
Thus, the consensus with respect to board independence effectiveness in earnings management is a stark contrast to the absence of any agreement regarding the role of an independent board in corporate governance and firm performance (Sarkar et al., 2008).
2.4 Hypotheses
As cited by Siregar and Utama (2008), Subramanyam (1996) states that “there are two types of earnings management: efficient and opportunistic. Earnings management is efficient if managers use their discretion to communicate private information about firm profitability, which is yet to be reflected in the historical cost-based earnings, while it is opportunistic if managers use their discretion to maximize their utility, thereby garbling earnings” (Siregar and Utama, 2008, p.8). According to Subramanyam (1996) we can therefore test whether earnings management is efficient or opportunistic by examining the discretionary accruals’ ability to signal future profitability. If earnings management is efficient, then discretionary accruals (earnings management proxy) will have a significant positive relationship with future profitability. On the other hand, if earnings management is opportunistic, discretionary
accruals will have a significant negative relationship or an insignificant relationship with future profitability (Siregar and Utama, 2008). Since, the relationship can go either way, positive if earnings management is efficient and negative if opportunistic, the first hypothesis is non-directional. Following Subramanyam (1996):
Hypothesis 1: There is a relationship between discretionary accruals and future profitability.
As shown by the literature on the existence of corporate governance in firms, the
consensus is that corporate governance practices suppress opportunistic earnings management (Siregar and Utama, 2008). As quoted by Siregar and Utama (2008), DeAngelo (1981)
indicates that audit quality is a “market-assessed joint probability that a given auditor will both (a) discover a breach in the client's accounting system, and (b) report the breach. This suggests that high audit quality should limit opportunistic earnings management” (Siregar and Utama, 2008, p.10). Furthermore, as mentioned before Becker et al. (1998) state that
“Also, consistent with earnings management, we find that the mean and median of the absolute value of discretionary accruals are greater for firms with non-Big Six auditors. This result also indicates that lower audit quality is associated with more "accounting flexibility"” (Becker et al, 1998, p.1). Based on these studies it is expected that a higher audit quality will lead to less opportunistic earnings management and in addition will increase the positive effect of discretionary accruals on future profitability. Following this:
Hypothesis 2: The positive effect of discretionary accruals on future profitability is higher in firms audited by Big Four auditors with higher audit fees compared to firms audited by Big Four auditors with lower audit fees.
According to Siregar and Utama (2008),
“The board of commissioners of a public firm is required to establish an audit committee. Overseeing the firm's financial-reporting process is one of the audit committee's tasks. It meets regularly with the external and internal auditors to review the corporation's financial statements, audit processes, and internal controls. Therefore, its existence will ultimately induce the company to produce more accurate financial statements or, in other words, less opportunistic earnings management” (Siregar and Utama, 2008, p.10-11).
Furthermore, Klein (2002) finds that the less independent an audit committee is, the more the discretionary accruals in a firm will increase. This indicates that the existence of an audit committee decreases the opportunistic earnings management. Less opportunistic earnings management means a higher effect of discretionary accruals on future profitability. Following Klein:
Hypothesis 3: The positive effect of discretionary accruals on future profitability is higher when firms have a lower percentage of inside directors in the audit committee compared to firms with a higher percentage of inside directors in the audit committee.
Finally, Peasnell et al. (2000) find that the probability of income-increasing accruals decreases as the proportion of outside-directors increases for a sample of U.K. firms (Siregar and Utama, 2008). Furthermore, Chtourou et al. (2001) find that an independent board constrains earnings management activity. Based on these two studies it is expected that an increase in the proportion of independent members on the board of directors will lead to a decrease in opportunistic earnings management. As mentioned before, it is expected that a decrease in opportunistic earnings management will lead to an increase in the positive effect of discretionary accruals on future profitability (Siregar and Utama, 2008). Following this, the final hypothesis will be:
Hypothesis 4: The higher the proportion of independent members on the board of directors, the higher the positive effect of discretionary accruals on future profitability.
3. Research methodology 3.1 Research model
To test hypothesis 1 I will use a modified version of the model Subramanyam (1996) used in his study for future profitability. The model by Subramanyam (1996) is as follows:
Xit+1=β0+ β1CFOit+ β2NDACit+ β3DACit+ β4DFAMit+ β5INSTit+ β6DSIZEit+
β7AUDITit+ β8BODit+ β9AUDCOMit+ β10D99i+ β11D00i+ β12D01i+ε
Where dependent variable Xit+1 is future profitability, measured by three different
CFOt+1 = cash flows from operation one-year-ahead
NDNIt+1 = non-discretionary net income one-year-ahead, calculated by taking net income
before extraordinary items minus the discretionary accruals.
ΔEARNt+1 = change in earnings one-year-ahead. Earnings is measured as net income before
extraordinary items.
All dependent variables are scaled by beginning total assets. Independent variables are: CFO = cash flows from operation
NDAC = non-discretionary accruals, calculated by taking the total accruals from the chosen earnings management model minus the discretionary accruals.
DAC = discretionary accruals, calculated by taking the residuals of the chosen earnings management model regression.
DFAM= one if a firm has a proportion of family ownership>50% and does not belong to business groups and zero otherwise.
INST= proportion of institutional ownership, measured by taking the amount of firms with institutional ownership and dividing it by the total amount of firms.
DSIZE= one if a firm has market capitalization above mean and zero otherwise. AUDIT= one if a firm is audited by Big 4 auditors and zero otherwise.
BOD= proportion of independent board members, measured by dividing the amount of independent outside board members by the amount of total board members.
AUDCOM= one if a firm has an audit committee in accordance with JSE1 rules and zero otherwise.
I have used this model as a base for the model I have actually used for my research. However, since my research focuses specifically on the corporate governance practices of board independence, audit quality and audit committee there are certain redundant variables for my research. These redundant variables are DFAM, INST, DSIZE and the dummy
variables for 1999 through 2001. DFAM, INST and DSIZE are redundant because they are no corporate governance practices and the dummy variables D99, D00 and D01 are redundant since they cover a period (1999-2001) that is not in my research. I have modified AUDIT for the audit quality proxy as a dummy variable where the dummy variable takes a value of one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. Furthermore, I have renamed AUDIT to AUDQUA. I have also renamed BOD to INDEP as a proxy for board
independence. Additionally, I have used a modified version of AUDCOM for the audit committee proxy since the model by Subramanyam uses the norm of JSE rules instead of the US rules. The new AUDCOM variable consists of the proportion of outsiders in the audit committee. Ultimately, I have used the dummy time variable DCRISIS where the variable takes one if the year is 2008-2013 and zero if the year is 2004-2007. This variable is used as a control variable to find out whether the financial crisis had any influence on future
profitability.
Earnings are decomposed into three variables: CFO, non-discretionary accruals
(NDAC), and discretionary accruals (DAC). DAC is the variable of interest and if the type of earnings management is efficient, the coefficient (β3) will be positive. Otherwise, it will be
either zero or negative.
Thus, the final version of the model I have used is the following:
Xit+1=β0+ β1CFOit+ β2NDACit+ β3DACit+ β4INDEPit+ β5AUDCOMit+
β6AUDQUAit+ β7DCRISISi+ε
Where dependent variable Xit+1 is future profitability, measured by three different
proxies:
CFOt+1 = cash flows from operation one-year-ahead
NDNIt+1 = non-discretionary net income one-year-ahead, calculated by taking net income
before extraordinary items minus the discretionary accruals.
ΔEARNt+1 = change in earnings one-year-ahead. Earnings is measured as net income before
extraordinary items.
All dependent variables are scaled by beginning total assets. Independent variables are: CFO = cash flows from operation
NDAC = non-discretionary accruals, calculated by taking the total accruals from the modified-Jones model (1991) minus the discretionary accruals.
DAC = discretionary accruals, calculated by taking the residuals of the modified-Jones model (1991) regression.
INDEP = proportion of independent board members, calculated by dividing the amount of independent, outside board members by the total amount of board members.
AUDCOM = proportion of insiders in audit committee, calculated by dividing the total amount of inside members in the audit committee by the total amount of members in the audit committee.
AUDQUA = one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. Calculated by ranking total audit fees and generating a dummy variable for the top 50%. DCRISIS = year dummy for the period from 2007 onwards.
According to by Siregar and Utama (2008), Bushee (2001) mentions that
“The proportion of firm value reflected in future earnings is positively associated with institutional investors. This indicates that a large capital investment in a firm will force institutional investors to monitor management behavior to be sure they pursue long-run profitability. Year dummy is used to control for the variation of future profitability over time” (Siregar and Utama, 2008, p.12)
For hypotheses 2 through 4 I have additionally modified the model used by
Subramanyam (1996). These hypotheses imply that the DAC coefficient (β3) is influenced by
proxies for audit quality, board independence and audit committee. In order for the DAC coefficient to vary, the model used by Subramanyam (1996) will be as follows according to Siregar and Utama (2008):
Xit+1=β0+ β1CFOit+ β2NDACit+ β3DACit+ β4DACit x DFAMit+ β5DACit x
INSTit+ β6DACit x DSIZEit+ β7DACit x AUDITit+ β8DACit x BODit+ β9DACit x
AUDCOMit+ β10 DFAMit+ β11INSTit+ β12DSIZEit+ β13AUDITit+ β14BODit +
β15AUDCOMit+ β16D99i+ β17D00i+ β18D01i+ε
Again, since my research focuses specifically on the corporate governance practices of board independence, audit quality and audit committee there are certain redundant variables for my research. These redundant variables are DFAM, INST, DSIZE and the dummy variables for 1999 through 2001. Also redundant are the interacting variables between DFAM, INST and DSIZE on the one hand and DAC, the discretionary accruals, on the other hand. DFAM, INST and DSIZE are redundant because they are no corporate governance practices and the dummy variables D99, D00 and D01 are redundant since they cover a period (1999-2001) that is not in my research. Again, I have modified AUDIT for the audit quality proxy as a dummy variable where the dummy variable takes a value of one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. Furthermore, I have renamed AUDIT to AUDQUA. I have also renamed BOD to INDEP as a proxy for board
independence. Additionally, I have used a modified version of AUDCOM for the audit committee proxy since the model by Subramanyam uses the norm of JSE rules instead of the US rules. The new AUDCOM variable consists of the proportion of outsiders in the audit committee. Next, the interacting variables between audit quality, board independence and audit committee on the one hand and discretionary accruals on the other hand are renamed as well. Ultimately, I have used the dummy time variable DCRISIS where the variable takes one if the year is 2008-2013 and zero if the year is 2004-2007. This variable is used as a control variable to find out whether the financial crisis had any influence on future profitability.
The model this study has used to answer hypotheses 2 through 4 was:
Xit+1=β0+β1CFOit+ β2NDACit+ β3DACit+ β4DACit x INDEPit+ β5DACit x
AUDCOMit+ β6DACit x AUDQUAit+ β7INDEPit+ β8AUDCOMit + β9AUDQUAit+
β10DCRISISi +ε
Where dependent variable Xit+1 is future profitability, measured by three different
proxies:
CFOt+1 = cash flows from operation one-year-ahead
NDNIt+1 = non-discretionary net income one-year-ahead, calculated by taking net income
before extraordinary items minus the discretionary accruals.
ΔEARNt+1 = change in earnings one-year-ahead. Earnings is measured as net income before
extraordinary items.
All dependent variables are scaled by beginning total assets. Independent variables are: CFO = cash flows from operation
NDAC = non-discretionary accruals, calculated by taking the total accruals from the modified-Jones model (1991) minus the discretionary accruals.
DAC = discretionary accruals, calculated by taking the residuals of the modified-Jones model (1991) regression.
INDEP = proportion of independent board members, calculated by dividing the amount of independent, outside board members by the total amount of board members.
AUDCOM = proportion of insiders in audit committee, calculated by dividing the total amount of inside members in the audit committee by the total amount of members in the audit committee.
AUDQUA = one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. Calculated by ranking total audit fees and generating a dummy variable for the top 50%. DCRISIS = year dummy for the period from 2007 onwards.
We expect β6>0 for hypothesis 2, β5>0 for hypothesis 3 and β4>0 for hypothesis 4. The
effect of the discretionary accruals DAC on future profitability is moderated by the
AUDQUA, INDEP and AUDCOM variables. Those variables interact with DAC because the coefficient results will show the percentage increase of each variable on the relationship between discretionary accruals and future profitability (Siregar and Utama, 2008). For example, if AUDQUA is zero then the effect of discretionary accruals on future profitability is β3. If AUDQUA is one then the effect of discretionary accruals on future profitability is β3+
β6. β6 is the difference between the interacting variables coefficient of AUDQUA. According
to Siregar and Utama (2008), each of the three variables AUDQUA, INDEP and AUDCOM are also included as independent variables to control the possibility that each variable has a direct influence on future profitability.
3.2 Definition of variables 3.2.1 Future profitability
Future profitability is measured by the following variables: 1. CFOt+1 = cash flows from operation one year ahead.
2. NDNIt+1 = non-discretionary net income (measured by the variables EARN-DAC) one
year ahead.
3. ∆EARNt+1 = change in earnings (EARNt+1-EARNt) one year ahead.
These measures are based on a modification by Siregar and Utama (2008) derived from previous research (Subramanyam, 1996; Krishnan, 2003). Subramanyam (1996) and Krishnan (2003) use EARN as a measure of future profitability but Siregar and Utama (2008) argue that ∆EARNt+1 is more suitable for the model since EARN has inherent weaknesses. A potential
weakness of the EARN variable is that Earnings has discretionary accruals in it, so a positive and significant relationship between discretionary accruals in year t and earnings in year t+1 could be due to management creating other discretionary accruals in year t+1 rather than being an indication of efficient earnings management (Siregar and Utama, 2008). Therefore,
as a substitute for traditional earnings measures, the change in earnings measured by ∆EARNt+1 is used as a measure of future profitability. Siregar and Utama (2008) claim that
because earnings and discretionary accruals tend to have a stationary nature, the use of change in earnings will control for the stationary nature of discretionary accruals. CFOt+1 as cash
flows from operation one year ahead and NDNIt+1 as the non-discretionary net income
(measured by the variables EARN-DAC) one year ahead do not have the discretionary accrual component hence they do not have the inherent problems of earnings.
Table 1
Correlation between discretionary accruals in year t and year t+1
DACt
DACt+1 0.1525
0.000***
Value in bold is the p value of the coefficient correlation between DACt and DACt+1. ***Significant at 1%.
Table 1 indicates that there is a positive and significant correlation between discretionary accruals in year t and year t+1. This supports the theory by Siregar and Utama (2008) that a positive and significant relationship between discretionary accruals in year t and earnings in year t+1 could be due to management creating other discretionary accruals in year t+1 rather than being an indication of efficient earnings management (Siregar and Utama, 2008).
3.2.2 Earnings management
According to Siregar and Utama (2008)
“Total accruals (ACCR) is calculated as the difference between earnings and cash flows from operation (ACCR=EARN-CFO). Earnings (EARN) is defined as net income before extraordinary items and cash flows from operation (CFO) is net cash flows from operating activities reported in the Statement of Cash Flows” (Siregar and Utama, 2008, p.13).
Based on the adjusted R2 calculated from the following three models, one of those models will be selected to decompose total accruals into discretionary and non-discretionary accrual components (Siregar and Utama, 2008).
The first model used is the Jones (1991) model: ACCR = α + α ∆REV + α PPE + ε .
1. ACCR = total accruals
2. ∆REV = change in revenue from year t-1 to year t (REVt – REVt-1)
3. PPE = gross property, plant and equipment in year t With all variables scaled by beginning total assets.
The second model used is the Dechow et al. (1995) model: ACCRit = α0 + α1 [∆REVit - ∆RECit]+ α2PPEit + εit.
1. ACCR = total accruals
2. ∆REV = change in revenue from year t-1 to year t (REVt – REVt-1)
3. ∆REC = change in net accounts receivables from year t-1 to year t (RECt –
RECt-1)
4. PPE = gross property, plant and equipment in year t With all variables scaled by beginning total assets.
The third and final model used is a modification of the first model, the Modified-Jones model (1991):
ACCRit = α0 + α1 (1/Assetsit-1) + α2 [∆REVit - ∆RECit]+ α3PPEit + εit.
1. ACCR = total accruals 2. 1/Assets = inverse assets
3. ∆REV = change in revenue from year t-1 to year t (REVt – REVt-1)
4. ∆REC = change in net accounts receivables from year t-1 to year t (RECt –
RECt-1)
5. PPE = gross property, plant and equipment in year t Where all variables are scaled by lagged total assets.
Both non-discretionary accruals (NDAC) and discretionary accruals (DAC) are derived from the above models. NDAC are the fitted values and DAC are defined as the residuals according to Siregar and Utama (2008). Following Subramanyam (1996), a cross-sectional method is used to estimate each model separately for every combination of calendar year and firm group (Siregar and Utama, 2008).
3.2.3 Corporate governance practices 3.2.3.1 Auditor’s size
The auditor’s size is used to measure audit quality via a dummy variable where one equals high audit quality (firms audited by Big Four auditors) and zero equals low audit quality (firms audited by non-Big Four auditors). As cited by Becker et al. (1998):
“Numerous studies have investigated the notion that Big Six auditors provide higher-quality audits than non-Big six auditors. Theoretical support for such a quality differentiation is provided in DeAngelo 1981, who demonstrates analytically that larger audit firms have greater incentives to detect and reveal management misreporting. Because Big Six firms are larger than their competitors, it follows from DeAngelo's analysis that they are of higher quality” (Becker et al., 1998, p.1).
Thus concluding from existing literature, it can be assumed that Big N firms provide higher quality audits. However, since every listed firm on the S&P 500 is being audited by a Big N firm, this study uses the size of audit fees as a proxy for audit quality. Siregar and Utama (2008) state that audit fees are another proxy for audit fees so this study uses a dummy variable for audit quality where the value 1 equals the top 50% of ranked audit fees and a value of 0 equals the bottom 50% of ranked audit fees.
3.2.3.2 Board independence
Siregar and Utama (2008) state that the proportion of independent board members can be derived from dividing the number of independent commissioners by the total number of commissioners on the board.
3.2.3.4 Audit committee
The existence of an audit committee variable is measured as a dummy variable. The variable takes the value of one for firms with an audit committee and zero for firms without an audit committee. Since every listed firm on the S&P 500 index is obligated to have an audit committee, this study proposes to use the percentage of inside directors on the audit committee as a proxy for the governance practice of an audit committee. This study proposes that a lower proportion of inside directors will lead to more efficient earnings management.
4. Sample selection
The sample of this research consisted of S&P 500 firms for the years 2004 to 2013. This sample and these firms provided me with a large enough data to test for audit quality, audit committee and board independence. According to Kawaller et al. (1987), the S&P 500 stock index represents the market value of all outstanding common shares of 500 selected firms (Kawaller et al., 1987, p.1310). The 500 companies on the S&P 500 index were leading companies and captured 80% of the coverage of the available market capitalization according to the official S&P 500 website (Spindicescom, 2015).
I have used the database research as a research method for my research. I have gathered my data on the S&P 500 index firms from the Wharton Research Data Services (WRDS) database. Basic financial statement data was gathered from COMPUSTAT using the ticker abbreviations of the firms I was going to select. Furthermore, my data on audit committee and board characteristics was gathered from the ISS database, formerly known as RiskMetrics. Finally, my data on audit fees and audit quality was gathered from the AuditAnalytics database. I have excluded financial firms, real estate firms and telecommunication companies in order to compare my results with the results of the Siregar and Utama (2008) paper. Last of all, I have selected firms with a fiscal year-end of December 31 in order to increase comparability between firms.
In summary, in order to select firms for the sample the following criteria will be used:
1. The firm is listed on the S&P 500 index from 2004-2013.
2. Financial, real estate and telecommunication firms are excluded. 3. Firms that delisted during the timeframe are excluded.
4. Firms have a December 31 fiscal year-end
5. Firm-level data is available for each year in the sample period. 6. Outliers are corrected for by trimming and winsorizing the data.
Finally, I will use the Jones (1991), Dechow et al. (1995) and Modified-Jones (1991) models to calculate total accruals as the difference between earnings and cash flows from operation. The models are used to decompose total accruals into discretionary and non-discretionary accruals. Based on the adjusted R2, I will select the best model to decompose the total accruals into discretionary accruals and
non-discretionary accruals. The non-non-discretionary accruals will be the fitted values of the three models and the residuals will be the discretionary accruals.
Table 2
Sample selection procedure
Total number of firms listed on the S&P 500 as of December 31, 2004 500
Firms in financial, real estate and telecommunication industries (92)
Delisting during period 2004-2013 (123)
Firms with non-December 31 fiscal year-end (80)
Incomplete data and data trimming (49)
Total sample firms 156
The procedure for sample selection is shown in Table 2. After the application of the criteria, there are 156 unique firms left in our sample. Every firm has 10 years of data so the total firm-year observations are 1560 (N=1560). Furthermore, industry distribution for the sample selected is shown in Table 3.
Table 3
Observations per industry
Industry Industry classification code Number of firms
Mining 1000-1499 9
Construction 1500-1799 1
Manufacturing 2000-3999 78
Transportation, Communications,
Electric, Gas and Sanitary service 4000-4999 31
Wholesale Trade 5000-5199 2 Retail Trade 5200-5999 4 Services 7000-8999 30 Public Administration 9100-9729 1 Total 156 Table 4
Evaluation of earnings management measurement models Part A: adjusted R2
Measurement Model adjusted R2
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Mean Jones 0.013 0.023 0.002 0.047 0.004 0.203 0.128 0.196 0.128 0.31 0.06 Dechow 0.01 0.023 0.005 0.055 0.002 0.198 0.131 0.195 0.13 0.309 0.058 Modified- 0.031 0.065 0.009 0.453 -0.001 0.169 0.181 0.192 0.183 0.281 0.10 Jones
Part B: predicted sign Jones (1991) ΔREV (+) PPE (−) Dechow et al. (1995) ΔREV-ΔREC (+) PPE (−) Modified-Jones (1991) 1/TA (−) ΔREV-ΔREC (−) PPE (−)
To decide which model to use in the main analysis, we have to evaluate the adjusted R2 of each model. The adjusted R2 explains the explanatory power of each model according to Siregar and Utama (2008). The higher the mean of the adjusted R2 for a model, the higher the explanatory power. I will use the model with the highest mean adjusted R2 for my main analysis. The results for each of the three models are presented in Table 4. The Modified-Jones model (1991) has the highest mean adjusted R2 so I will employ this model in my main analysis. However, I will still use the regular Jones model (1991) and Dechow et al. model (1995) in my sensitivity analysis.
Comparing these predicted signs with the predicted signs of Siregar and Utama (2008), we find that all predicted signs are identical. However, since Siregar and Utama (2008) did not use the Modified-Jones model (1991) we cannot compare those predicted signs. The sign that stands out in the Modified-Jones model (1991) is the ΔREV-ΔREC sign. This predicted sign is negative unlike the ΔREV-ΔREC sign of the Dechow et al. model (1995). An
explanation for this is that the Modified-Jones model (1991) uses lagged assets as a scale compared to beginning total assets for the Dechow et al. model (1995). Another explanation can be that the inverse assets variable is added and in turn influences the predicted sign of the ΔREV-ΔREC variable.
5. Results
5.1 Opportunistic versus efficient earnings management
Tables 5 and 6 display the descriptive statistics and their correlation. Table 5 shows that the sample firms on average have positive cash flows from operation one-year-ahead and positive non-discretionary net income one-year-ahead.
Table 5
Descriptive statistics
Mean Median Maximum Minimum Standard deviation
CFOt+1 0.1103 0.1005 0.5739 -0.1063 0.0622 NDNIt+1 0.0639 0.0555 0.5398 -0.1710 0.0625 ΔEARNt+1 0.0001 0.0001 0.9061 -0.9140 0.0630 CFO 0.1032 0.0943 0.5549 -0.1066 0.0587 NDAC -0.0458 -0.0401 0.2008 -0.9831 0.0527 DAC -2.72e-11 0.0076 0.2847 -0.9327 0.0534 INDEP 0.8207 0.8462 1.0000 0.2857 0.101 AUDCOM 0.3906 0.3846 0.8889 0.0000 0.1128 Proportion Proportion Dummy=1 Dummy=0 AUDQUA 50.00% 50.0%
CFOt+1 = cash flows from operation one-year-ahead, NDNIt+1 = non-discretionary net income
one-year-ahead, ΔEARNt+1 = change in earnings one-year-ahead, CFO = cash flows from operation, NDAC =
non-discretionary accruals, DAC = non-discretionary accruals, INDEP = proportion of independent board, AUDCOM = proportion of outsiders in audit committee, AUDQUA = one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%.
About the change in earnings one-year-ahead (ΔEARNt+1) we cannot say whether the
change is positive or negative since the mean is so close to zero. On average, 82.1% of board members is independent and 39.1% of the directors of the sample firms are members of the audit committee. In line with the findings of Siregar and Utama (2008), CFO has a negative and significant correlation with both discretionary (DAC) and non-discretionary (NDAC) accruals. These findings are consistent with the smoothing nature of accruals (Siregar and Utama, 2008). A significant and positive correlation between INDEP and both AUDCOM and AUDQUA shows that firms with a higher proportion of independent board members tend to have a higher percentage of directors who are members of the audit committee and tend to pay higher audit fees which in turn is a sign of higher audit quality.
Table 6 Correlation
CFOt+1 NDNIt+1 ΔEARNt+1 CFO NDAC DAC INDEP AUDCOM AUDQUA CFOt+1 1.0000 NDNIt+1 0.9788 1.0000 ΔEARNt+1 0.0404 0.0427 1.0000 CFO 0.7863 0.7590 -0.0767 1.0000 NDAC -0.3530 -0.3029 0.5181 -0.2418 1.0000 DAC -0.3488 -0.3481 0.5287 -0.2247 0.9706 1.0000 INDEP -0.0596 -0.0741 -0.0068 -0.0473 0.0432 0.0626 1.0000 AUDCOM 0.0959 0.0921 -0.0384 0.1195 -0.0394 -0.0371 0.2389 1.0000 AUDQUA -0.1741 -0.1540 -0.0013 -0.1353 0.0682 0.0480 0.1684 -0.0442 1.0000
CFOt+1 = cash flows from operation one-year-ahead, NDNIt+1 = non-discretionary net income
one-year-ahead, ΔEARNt+1 = change in earnings one-year-ahead, CFO = cash flows from operation, NDAC =
non-discretionary accruals, DAC = non-discretionary accruals, INDEP = proportion of independent board, AUDCOM = proportion of outsiders in audit committee, AUDQUA = one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. Bold = significant at 1%. Italic = significant at 5%. Underlined = significant at 10%.
Table 7 shows the findings from the regression of the research model used to answer the first hypothesis. The table shows that the coefficient for discretionary accruals (DAC) is positive and significant for CFOt+1 (cash flows from operation one-year-ahead), NDNIt+1
(non-discretionary net income one-year-ahead) and ΔEARNt+1 (change in earnings
one-year-ahead). These findings indicate that the type of earnings management tends to be efficient, which supports hypothesis 1 that there is a relationship between future profitability and discretionary accruals. This is consistent with findings by Subramanyam (1996), Gul et al. (2000) and Krishnan (2003). Those three articles show evidence that earnings management of listed firms in United States is efficient (Siregar and Utama, 2008).
Table 7
Regression of future profitability on discretionary accruals, other earnings components, and control variable.
Variable name Expected sign CFOt+1 NDNIt+1 ΔEARNt+1
Coefficient p value Coefficient p value Coefficient p value Intercept 0.0132 0.155 0.0346 0.031** 0.0213 0.122 CFO + 0.7615 0.000*** 0.0682 0.013** 0.0487 0.055* NDAC + -0.2308 0.002*** 0.3128 0.010*** 0.1185 0.292 DAC +/- 0.3000 0.000*** 0.4047 0.001*** 0.5381 0.000*** INDEP + -0.0207 0.058* -0.0240 0.199 -0.0209 0.199 AUDCOM - 0.0193 0.033** -0.0083 0.573 -0.0107 0.427 AUDQUA + 0.0014 0.493 -0.0041 0.206 -0.0024 0.410 DCRISIS 0.0069 0.002*** 0.0021 0.585 0.0012 0.710 N 1404 1248 1404 Adjusted R2 0.6266 0.3224 0.2804 F-statistic 337.40 85.76 79.12 P value (F-statistic) 0.000*** 0.000*** 0.000*** Xit+1=β0+ β1CFOit+ β2NDACit+ β3DACit+ β4INDEPit+ β5AUDCOMit+ β6AUDQUAit+ β7DCRISISi+ε Dependent variable: CFOt+1 = cash flows from operation one-year-ahead, NDNIt+1 = non-discretionary net income one-year-ahead, ΔEARNt+1 = change in earnings one-year-ahead.
Independent variables: CFO = cash flows from operation, NDAC = non-discretionary accruals, DAC = discretionary accruals, INDEP = proportion of independent board, AUDCOM = proportion of outsiders in audit committee, AUDQUA = one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. ⁎⁎⁎Significant at 1%, ⁎⁎significant at 5%, ⁎significant at 10% (two-tail).
5.2 Corporate governance practices
Table 8 shows the findings from the regression of the research model used to answer the second, third and fourth hypotheses. Before we can regress the research model we have to test for multicollinearity in the independent variables. Multicollinearity occurs when two or more independent variables in a multiple regression model have a high correlation. A small change in the model would lead to a big change in the coefficients of the variables according to Stock and Watson (2012). According to Siregar and Utama (2008), in order for multicollinearity not to occur, the variance inflation factor (VIF) of the independent variables needs to be below
10. In this regression, the VIFs are below 10 for every independent variable (CFO, NDAC, DAC, INDEP, AUDCOM and AUDQUA), so multicollinearity is not an issue.
Table 8
Regression of future profitability on discretionary accruals, discretionary accruals and corporate governance practices interaction, other earnings components and control variable.
Variable name Expected sign CFOt+1 NDNIt+1 ΔEARNt+1
Coefficient p value Coefficient p value Coefficient p value Intercept 0.0136 0.144 0.0362 0.024** 0.0228 0.100* CFO + 0.7620 0.000*** 0.0666 0.015** 0.0471 0.065* NDAC + -0.2380 0.002*** 0.3575 0.003*** 0.1457 0.198 DAC +/- -0.0284 0.873 1.0472 0.000*** 0.6771 0.011** DAC*INDEP + 0.3159 0.081* -0.3787 0.183 0.0511 0.850 DAC*AUDCOM - 0.1236 0.454 -0.8354 0.001*** -0.4790 0.051* DAC*AUDQUA + 0.0783 0.057* -0.1381 0.033** -0.0588 0.337 INDEP + -0.0219 0.046** -0.0215 0.250 -0.0200 0.219 AUDCOM - 0.0192 0.036** -0.0128 0.389 -0.0134 0.324 AUDQUA + 0.0010 0.615 -0.0032 0.322 -0.0021 0.489 DCRISIS 0.0071 0.001*** 0.0021 0.574 0.0013 0.684 N 1404 1248 1404 Adjusted R2 0.6273 0.4001 0.2811 F-statistic 237.15 84.17 55.87 P value (F-statistic) 0.000*** 0.000*** 0.000***
Xit+1=β0+β1CFOit+ β2NDACit+ β3DACit+ β4DACit x INDEPit+ β5DACit x AUDCOMit+ β6DACit x AUDQUAit+ β7INDEPit+ β8AUDCOMit + β9AUDQUAit+ β10DCRISISi +ε
Dependent variable: CFOt+1 = cash flows from operation one-year-ahead, NDNIt+1 = non-discretionary net income one-year-ahead, ΔEARNt+1 = change in earnings one-year-ahead.
Independent variables: CFO = cash flows from operation, NDAC = non-discretionary accruals, DAC = discretionary accruals, INDEP = proportion of independent board, AUDCOM = proportion of outsiders in audit committee, AUDQUA = one if audit fees are in the top 50% and zero if audit fees are in the bottom 50%. ⁎⁎⁎Significant at 1%, ⁎⁎significant at 5%, ⁎significant at 10% (two-tail).
Hypothesis 2: The positive effect of discretionary accruals on future profitability is higher in firms audited by Big Four auditors with higher audit fees compared to firms audited by Big Four auditors with lower audit fees is supported in the model where cash flows from operation one-year-ahead is the dependent variable (CFOt+1). The effect is significant at 10%.
However, the effect is significantly negative at 5% when the dependent variable is non-discretionary net income one-year-ahead (NDNIt+1). Results are inconclusive and not
significant when change in earnings one-year-ahead is the dependent variable (ΔEARNt+1).
This indicates that firms with higher audit fees as a proxy of higher audit quality do not use more efficient earnings management than firms with small audit fees, a proxy of lower audit quality. This may suggest audit fees is not a good proxy for audit quality.
Hypothesis 3: The positive effect of discretionary accruals on future profitability is higher when firms have a lower percentage of inside directors in the audit committee
compared to firms with a higher percentage of inside directors in the audit committee is also not supported in table 8. In the model where cash flows from operation one-year-ahead is the dependent variable (CFOt+1), the effect of DAC*AUDCOM is positive. However, the p value
indicates a non-significant positive effect so we cannot conclude that the positive effect of discretionary accruals on future profitability is higher when firms have a higher percentage of inside directors in the audit committee compared to firms with a lower percentage of inside directors in the audit committee. On the contrary, the other two models show that there is a significant negative effect at respectively the 1% and 5% level in the models where the dependent variable is non-discretionary net income one-year-ahead (NDNIt+1) and change in
earnings one-year-ahead (ΔEARNt+1). These findings would indicate that a higher percentage
of inside directors in the audit committee will lead to opportunistic earnings management and, in turn, the governance practice of more outside directors in audit committees will lead to more efficient earnings management.
Finally, table 8 shows that hypothesis 4: The higher the proportion of independent members on the board of directors, the higher the positive effect of discretionary accruals on future profitability is supported at the 10% level in the model where cash flows from
operation one-year-ahead (CFOt+1) is the dependent variable. The two other models with
NDNIt+1 and ΔEARNt+1 as the dependent variables give inconclusive and non-significant
results. Since only one out of three models shows that a higher proportion of independent board members will lead to more efficient earnings management we cannot accept hypothesis 4. An explanation according to Siregar and Utama could be that publicly listed firms appoint a
higher amount of independent board members just to comply with regulations. This could indicate that a high proportion of independent board members is not fully effective as a monitoring mechanism.
5.3 Sensitivity analysis
As mentioned before in section 4, the Modified-Jones model (1991) has the highest mean adjusted R2 so I employed this model in my main analysis. However, I have still used the regular Jones model (1991) and Dechow et al. model (1995) for the sensitivity analysis. In order to test the sensitivity of the results mentioned in section 5.1 and section 5.2, I have measured the discretionary accruals using the regular Jones model (1991) and Dechow et al. model (1995). As tables 9 and 10 in the appendix section, show, for both the Jones (1991) and Dechow et al. (1995) models the coefficient for discretionary accruals is positive in all three dependent variables (CFOt+1 = cash flows from operation one-year-ahead, NDNIt+1 =
non-discretionary net income one-year-ahead, ΔEARNt+1 = change in earnings one-year-ahead).
This is consistent with my main findings. This sensitivity analysis suggests that listed firms in the S&P 500 tend to conduct efficient earnings management. Furthermore, this is consistent with the aforementioned findings by Subramanyam (1996), Gul et al. (2000) and Krishnan (2003). Those three articles show evidence that earnings management of listed firms in United States is efficient (Siregar and Utama, 2008).
To test whether 50% is a good cutoff for the dummy variable for audit quality in audit fees I performed a sensitivity test using a cutoff of 25%:75% where the top 25% of audit fees are used to generate the value of 1 for the dummy variable. The lowest 75% of audit fees leave a value of 0 for the dummy variable AUDQUA. The results for the new cutoff are shown in table 11 in the appendix section. Results are significantly positive when CFOt+1 =
cash flows from operation one-year-ahead is the dependent variable. This indicates that firms with higher audit fees as a proxy of higher audit quality do use more efficient earnings management than firms with small audit fees, a proxy of lower audit quality when the cutoff is 25%:75% instead of 50%:50%. However, the other two dependent variables leave a negative significant effect. These ambiguous results are is in line with the findings mentioned in section 5.2.
Incorporated in my main analysis is the crisis period from 2008 onwards. This is also used as a control variable. To examine whether the results in this period are significantly different from the period before the crisis I examined the coefficient of the dummy variable
DCRISIS. The effect of the crisis is only significant when change in earnings one-year-ahead is the dependent variable (ΔEARNt+1). When there is just one significant effect out on the
three different dependent variables we can conclude that the evidence regarding earnings management in crisis periods does not differ from the evidence in non-crisis periods.
6. Conclusion and suggestions for future research 6.1 Conclusion
The findings in this study indicate that American firms listed on the S&P 500 index tend to conduct efficient earnings management. Although previous studies were ambiguous
whether firms conducted efficient or opportunistic earnings management, this study confirms previous findings by Subramanyam (1996), Gul et al. (2000) and Krishnan (2003). Those three articles show evidence that earnings management of listed firms in United States is efficient (Siregar and Utama, 2008).
Furthermore, this study shows evidence that firms with higher audit fees as a proxy of higher audit quality do not use more efficient earnings management than firms with small audit fees, a proxy of lower audit quality. However, this study indicates that a higher percentage of inside directors in the audit committee will lead to opportunistic earnings management and, in turn, the governance practice of more outside directors in audit committees will lead to more efficient earnings management.
As a final point, since only one out of three models shows that a higher proportion of independent board members will lead to more efficient earnings management we cannot conclude that a higher proportion of independent members on the board of directors will lead to a better, more significant, effect of discretionary accruals on future profitability.
6.2 Implications
This study finds that the effect on future profitability is inconclusive and non-significant for firms with higher audit fees as a proxy of higher audit quality compared to firms with small audit fees when the dummy variable cutoff is 50%. However, as the sensitivity analysis shows, results are significantly positive when change in earnings one-year-ahead is the dependent variable (ΔEARNt+1). This indicates that firms with higher audit fees as a proxy of
fees, a proxy of lower audit quality when the cutoff is 25%:75% instead of 50%:50%. These findings could suggest that a smaller cutoff for the dummy variable when using audit fees as a proxy for audit quality is in fact a good proxy for audit quality.
The existence of an independent board does not significantly affect the type of earnings management a firm conducts. Siregar and Utama (2008) state that most publicly traded public companies only appoint independent board members just to comply with the regulations set.
6.3 Limitations of the study
Since every firm on the S&P 500 uses a Big Four audit firm, we cannot use Big Four versus non-Big Four as a proxy for audit quality. Therefore we had to result to using audit fees as a proxy of audit quality. Getting the right cutoff proportion for the dummy variable to truly depict the effect of audit quality on future profitability is difficult. This study uses the arbitrary 50%:50% cutoff, but this cutoff did not give conclusive results.
Furthermore, since the study by Siregar and Utama (2008) was published, every listed firm on the S&P 500 is obligated to have an audit committee so this study used the proportion of outside directors on the committee as a proxy for audit committee. Results are however significant and in line with predicted signs.
6.4 Suggestions for future research
This study uses three different measures for future profitability, namely cash flows from operation one-year-ahead, non-discretionary net income one-year-ahead and change in
earnings one-year-ahead. Future research could try to calculate which of these three measures the best measure is. Future research could also attempt to find another measure for future profitability besides these three measures.
Future studies on this particular subject could try to focus on specific industries rather than an entire index. As the sensitivity analysis showed in this study, findings indicate that including financial, telecommunication and real estate industries does not significantly change the results but for example up and coming industries like Information Technology could alter results significantly.
A suggestion concerning audit quality could be to find the perfect cutoff proportion for audit fees as a proxy for audit quality. The arbitrary 50%:50% does not give significant results so future research should focus on minimalizing the proportion for the highest audit fees.
The final suggestion would be to increase the sample size. Using the S&P 1500 index rather than the smaller S&P 500 index would increase the total amount of firms and the percentage of market capitalization of those firms.
7. References
Balsam, S., Bartov, E., & Marquardt, C. (2002). Accruals management, investor
sophistication, and equity valuation: Evidence from 10-Q filings. Journal of Accounting
Research, 40(4), 987−1012.
Becker, C. L., DeFond, M. L., Jiambalvo, J., & Subramanyam, K. R. (1998). The effect of audit quality on earnings management. Contemporary Accounting Research, 15, 1−24. Burgstahler, D., & Dichev, I. (1997). Earnings management to avoid earnings decreases and
losses. Journal of Accounting and Economics, 24, 99−126.
Bushee, B. J. (2001). Do Institutional Investors Prefer Near‐Term Earnings over Long‐Run Value?*. Contemporary Accounting Research, 18(2), 207-246.
Business Dictionary. (2015) Corporate Governance. (Online, accessed 31.01.2015) URL:
http://www.businessdictionary.com/definition/corporate-governance.html
Chtourou, S., Bedard, J., & Courteau, L. (2001). Corporate governance and earnings management. Corporate Governance and Earnings Management (April 21, 2001). DeAngelo, L. (1981). Auditor size and audit quality. Journal of Accounting and Economics,
113−127.
DeAngelo, L. E. (1988). Managerial competition, information costs, and corporate
governance: The use of accounting performance measures in proxy contests. Journal of
Accounting and Economics, 10, 3−36.
DeAngelo, H., & DeAngelo, L. (1989). Proxy contests and the governance of publicly held corporations. Journal of Financial Economics, 23(1), 29-59.
Dechow, P. M., & Sloan, R. G. (1991). Executive incentives and the horizon problem: An empirical investigation. Journal of Accounting and Economics, 14, 51−89.
Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The
Accounting Review, 70, 193−225.
Dechow, P. M., R. G. Sloan and A. P. Sweeney. 1996. Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC.
