The impact of female CEOs on earnings management
Student: Eline van der Plas Student number: 10003550
Date final version: 22nd june 2015 Words: 12593
MSc Accountancy & Control, variant Accountancy
Statement of Originality
This document is written by student Eline van der Plas 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.
Abstract
This study examines the effect of a female CEO on earnings management. Specifically, it is hypothesized that a female CEO decreases earnings management and that this effect is smaller for CEOs who are in their last year of working at the firm. The paper extends existing literature on CEO gender and provides useful information about earnings management for users of financial statements.
The research was done using different proxies for earnings management. For accrual based earnings management the proxies accrual estimation errors and discretionary accruals were used. For real earnings management proxies for sales manipulation, discretionary expenses and overproduction were used. Two different regressions were formulated to test the effect of a female CEO on accrual based earnings management and the effect on real earnings management. Subsequently two hypotheses were formulated to test whether CEO tenure has an effect on the two different types of earnings management. The sample consists of S&P500 firms in the period of 2002-‐2013.
It is found that a female CEO only has an effect on earnings management though overproduction and through the reduction of discretionary expenses. The effect on discretionary expenses however is positive which is against expectations. The other types of earnings management showed no significant results suggesting there is no difference in earnings management between a firm with a female CEO and firms with a male CEO. There were no significant results regarding a CEO being in her last years at the firm, suggesting CEOs do not increase earnings management in their last years.
The study has several limitations. First, the results might not be applicable to other countries or smaller companies as the sample consisted of big US firms. Second, the sample contained a small amount of female CEOs, which shows there are still a lot more males in this position.
Table of contents
1 Introduction ...6
2 Literature and hypotheses development ...10
2.1 Gender differences... 10
2.1.1 Ethical differences ... 10
2.1.2 Other differences... 11
2.2 Earnings management... 12
2.2.1 Definition... 12
2.2.2 Accrual based earnings management and real earnings management. 13 2.2.2.1Accrual based earnings management ... 13
2.2.2.2 Real earnings management... 14
2.2.3 Incentives for earnings management... 15
2.2.3.1 Different incentives ... 15
2.2.3.2 Positive accounting theory... 16
2.2.4 Ethics and earnings management... 16
2.2.5 The board of directors and earnings management ... 17
2.2.6 CEO tenure and earnings management... 18
2.3 Hypotheses ... 19
3 Research design ...20
3.1 Data and sample selection ... 20
3.2 Proxies for earnings management ... 21
3.2.1 Accrual based earnings management ... 21
3.2.2 Real earnings management ... 22
3.3 Regressions ... 24
4 Results ...28
4.1 Descriptive statistics... 28
4.1.1 Total sample... 28
4.1.2 Female versus male CEOs... 29
4.2 The relation between a female CEO and earnings management ... 31
4.3 The relation between a female CEO in her last year and earnings management... 33 4.4 Additional analysis... 35 4.4.1 Combined RAM... 36 4.4.2 Absolute accruals ... 36 4.4.3 Audit fees... 36 5 Conclusion...38 References ...40 Appendix 1...43 Appendix 2...44 Appendix 3...45
1 Introduction
From research done by the Bureau of Labor statistics (2015) it is clear that the amount of females who work has increased from 1970 to 2012. The labor force participation rate was 43,9% in 1972 and increased to 57,5% in 2012. The amount of females on the board of directors is also increasing. A report from Institutional Shareholder Services (2014) states that the percentage of women on the board of directors in S&P500 companies was 15,2% in 2008 and increased to 18,7% in 2014.
Women and men differ in their way of working at a firm. Research shows that having women in a group improves group-‐decision making but on the other hand also increases conflicts. Firm performance seems to be higher in firms where there is gender diversity on the board than when this is not the case. This means having women on the board will positively influence firm performance (Erhardt et al., 2003). Research also states that women are more risk averse than men when looking at financial decision-‐ making (Jianakoplos and Bernasek, 1998).
Women act differently and gender diversity causes a difference in firm performance. Earnings management is often not seen as an ethical activity and it does not give the best representation of a firms’ earnings (Burns and Merchant, 1990; Elias, 2002). Because of this, it is likely that women will reduce earnings management. This research examines whether having a female CEO will affect earnings management.
There have been multiple studies on the effect of having women in certain positions in a firm and the effect on financial performance. Shrader et al. (1997) found that firms with more female managers had higher ROA, ROI and ROE, which indicates a higher firm performance. They could not find this same relationship for top management or board directors. The authors think that probably this is the case because there were very few women in those positions in their sample.
Carter et al. (2003) did find a relationship between female board members and firm performance in their study. They performed a research on Fortune 1000 companies and found there are more women on boards in bigger firms than in small firms and the number of women decreases as the number of inside board members increases. They
also found that when there are women on the board, firm performance is higher than when there are not. Moreover, Adams and Ferreira (2009) conclude that gender diversity only improves firm performance in firms that otherwise have weak governance. If firms have strong governance, their results show that gender diversity might even decrease shareholder value.
Women also seem to have an effect on the quality of earnings. Dichev et al. (2013) conclude that CFOs believe earnings are of high quality when they are sustainable and repeatable. More specifically, reporting choices should be and should not include one-‐ time items and long-‐term estimates.
Krishnan and Parsons (2008) find that earnings quality is positively associated with gender diversity in senior management. They performed a research on Fortune 500 companies for the period of 1996 to 2000. They used different measures of earnings quality like asymmetric timeliness, conservatism, earnings smoothness and persistence. From their results it appears that firms where gender diversity is present perform better in terms of stock returns and return on equity. Using different measures of conservatism they found earnings to be more conservative in the high diversity groups than in the low diversity group. They conclude that the fact that firms perform better when they have more females is not because of these same firms engaging more in earnings management than the worse performing firms.
Shrinidhi et al. (2011) performed a research on firms with information on gender in the Corporate Library Board Director database for the period 2001-‐2007. They measure the quality of earnings with different measures like accruals quality and meeting or beating benchmarks. The conclusion of the research is that earnings are of higher quality when there are females on the board, because this improves the oversight function of the board.
Barua et al. (2010) examined the effect of CFO gender on accruals quality. They perform the research for the years 2004 and 2005 using the Corporate Library Database. The different measures of accruals they use are performance matched abnormal accruals, which are estimated using the Modified Jones Model and accrual estimation errors, which are estimated using the Dichow Dichev model. From their results it also seems that in companies with female CFOs the level of absolute abnormal
accruals is lower and the level of estimation error is lower. This suggest the quality of accruals is higher in firms with female CFOs.
Finally, Peni and Vähämaa (2010) examined the effect of female CEOs and CFOs on earnings management for the period of 2003-‐2007. They found no relation between a female CEO and earnings management. A female CFO though, is according to their findings, associated with income-‐decreasing discretionary accruals. This suggests firms with female CFOs use more conservative financial reporting.
Earnings management is a very important ethical issue regarding the accounting profession (Merchant and Rockness, 1994, p.92). A research on ethics was done using a questionnaire answered by 100 people in two different firms. From the results it can be concluded that the ethical climates in the two corporations were significantly different. Furthermore, between the respondents there was disagreement on most of the topics, not only between the two firms but also within the same firm and in similar positions. This indicates there is no agreement about what is right and what is wrong. This disagreement causes problems, as none of the actions in the firm will be transparent to financial statement users (Merchant and Rockness, 1994).
This paper examines whether there is an impact on earnings management when the CEO of the firm is a female. Therefore the research question is:
What is the impact of a female CEO on earnings management?
There has been a lot of research done on the effect of gender diversity on firm performance and earnings quality but not on the effect of gender diversity in the board on earnings management. Females in the board is also becoming a more common phenomenon than it was before, which makes the research more interesting.
Peni and Vähämaa (2010) did a similar research to this research for the period of 2003-‐2007. Because of the short period of 5 years examined and because of the low number of female executives in that period they are not able to conclude whether having female executives would improve earnings quality. This research therefore uses a longer period to examine the situation. As the research is done including later years than Peni and Vähämaa (2010) it is probable that the amount of female executives in the sample will have increased.
The contribution of the paper will be the extension of knowledge on the relation of female CEOs with earnings management.
The rest of the paper is structured as follows: section 2 gives an overview of important theory for this paper including gender differences and earnings management. In this section the hypotheses are also formulated. In section 3 the research methodology is discussed, which contains the sample selection, proxies for earnings management and the regression models. In section 4 the results will be discussed. The last section, section 5 is the conclusion.
2 Literature and hypotheses development
2.1 Gender differences
In this paragraph the differences between males and females are explained. First, ethical differences between the two are explained and second other differences are explained.
2.1.1 Ethical differences
The research question suggests that there is a difference in behavior or influence in a firm between men and women. In this paragraph some of these differences are explained. The study of Betz et al. (1989) examines differences between males and females. They first explain two different existing theories that are used to explain gender differences. The first theory is the structural approach that states the differences between men and women exist because of early socialization and role requirements. This approach states that men and women in the same work environment will respond similarly because these differences will be overridden by the rewards and costs, which are present in a work environment. The gender socialization approach states that the values of men and women differ, which causes them to act differently. Men are more focused on competitive success while women are more focuses on having good relationships and doing things well. The study showed that men are more concerned with income, organizational status and leadership than women. Also females are more willing to help people. Lastly, the results showed that men are more likely to engage in unethical behavior. Overall, the gender socialization role was more supported (Betz et al., 1989).
Under male and female accounting students, female students find unethical behavior more severe than male students which means females are less likely to engage in unethical behavior than men. Also male students appear to be more cynical than female students. This again supports the gender socialization approach (Ameen, 1996).
A study done by Nguyen et al. (2008) focuses on gender differences in ethical judgment. The research was performed testing three different dimensions of ethics: moral equity, relativism and contractualism. Moral equity is what is right and wrong in
the broadest sense and is an individual perception, relativism is the perception of what is right versus what is wrong based on perceptions of a social/cultural system and contractualism is what is right versus what is wrong based on concepts of an implied contract between organizations and society. A six point scale was used from one being most unethical to six being least unethical. Results showed that for most of the moral issues and ethics theories, females were more ethical than males. The authors did an additional analysis in which they controlled for age but this did not change their results.
2.1.2 Other differences
When looking at men and women there are more differences than the ethical differences mentioned before. One of them is that female directors have less attendance problems than men. On a board with more gender diversity, male directors also have less attendance problems than men on a board with less diversity. Females are also more likely to join monitoring committees than males (Adams and Ferreira, 2009).
Toussaint and Webb (2005) performed a research on differences in empathy and forgiveness between men and women with a questionnaire that was answered by 127 people. The results showed that women were more empathic than men. For men there was a relationship between empathy and forgiving cognition, behavior and affect. This was not the case for women.
Another difference between the two genders is that women are, overall, more risk averse than men. Byrnes et al. (1999) performed a research on this topic. They examined different types of risk and whether there was a difference in taking those risks between men and women. For nearly all the types of risk it appeared that men were more willing to take risks. The gender differences also depended on age. For some types of risk, the gender differences would be bigger while people get older while for other risks the difference would get smaller.
Chaness and Gneezy (2012) performed a research on a more specific type of risk, which is financial risk. They did this with data from previous studies which all included an investment game. Their results show that women invest less in risky assets, which makes them more risk averse.
2.2 Earnings management
In this paragraph earnings management is explained. First a definition of earnings management is given. After that the difference between accrual based earnings management and real earnings management is explained. The next section describes incentives to engage in earnings management. Finally, the relation of earnings management and ethics, and the relation of earnings management and the board of directors are explained.
2.2.1 Definition
Previous research has shown that firms use earnings management to avoid earnings decreases and losses (Burghstaler and Dichev, 1997). “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 numbers.” This is a definition of earnings management given by Healy and Wahlen (1999). The judgment can be exercised in many different ways. It is used to estimate future events, choose between accounting methods, in working capital management which affect cost allocations and revenues, to choose in making or deferring expenditures and in choosing how to structure corporate transactions. If earnings management is used to mislead shareholders than managers assume that shareholders do not undo earnings management or shareholders do not have access to the same information as managers do. Managers can also use judgment to make financial information more informative.
Beneish (2001) give two other definitions of earnings management. The first is “The process of taking deliberate steps within the constraints of generally accepted accounting principles to bring about a desired level of reported earnings.” This definition was given by Davidson, Stickney and Weil in 1987. Another definition given in Beneish’s article is one that was given by Schipper in 1989. This definition is as follows: “A purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain (as opposed to say, merely facilitating the neutral operation of the process).”… “A minor extension would encompass ‘real’ earnings management,
accomplished by timing investment of financial decisions to alter reported earnings or some subset of it.”
Scott (2009, p.403) gives a broader definition of earnings management which is: “Earnings management is the choice by a manager of accounting policies, or actions effecting earnings, so as to achieve some specific reported earnings objective.”
As shown above there are a lot of different definitions of earnings management. The definition used by Scott is the most neutral and the broadest definition. This definition will be used in this research.
2.2.2 Accrual based earnings management and real earnings management
Earnings can be managed in two different ways. The first is through manipulation of accruals. This type of earnings management does not affect cash flows. The second way of earnings management is real earnings management where activities are manipulated. This type of earnings management affects cash flows and can also affect accruals (Roychowdhury, 2006). In the next two paragraphs the types of earnings management are further explained.
2.2.2.1 Accrual based earnings management
One way of earnings management is the use of accruals to increase or decrease income. For the accounting of accruals a large amount of managerial discretion is used. As mentioned before accruals do not affect cash flows. Accruals capture the difference between cash flows and income (Bergstresser and Philippon, 2004).
According to Badertscher (2011) the opportunity of accrual based earnings management is created because of the flexibility of GAAP. An advantage of accrual based earnings management for managers are that this type of earnings management is less likely to damage the long-‐term value of the firm as it does not affect cash flows. Another advantage is that this earnings management can be done at the end of the period when the amount of earnings before managing them is already known. A disadvantage of accrual based earnings management is that accruals reverse over time. When the accruals reverse the firm might not have the possibility to engage in this type of earnings management anymore.
2.2.2.2 Real earnings management
Real earnings management is defined as “management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds” (Roychowdhury, 2006, p.336). It is done to mislead stakeholders so they believe goals have been met with the normal operations of the firm. Earnings manipulation occurs when the activities performed are more than the activities would normally be in the same economic situation. Real earnings management can cause a negative effect on future cash flows which can reduce the firm value (Roychowdhury, 2006). Roychowdhury (2006) describes three different types of real earnings management, which are sales manipulation, reduction of discretionary expenditures and overproduction. These types are described below.
Sales manipulation is temporarily increasing sales with the use of discounts or certain credit terms. When firms use limited-‐time discounts on their prices the sales volume will increase. When the price goes back to the price it was before the discount, the increased volume will disappear. The margins of sales with discounts are lower which causes production costs to be very high relative to sales and which causes the cash inflow per sale to decline. Another way of sales manipulation is using more lenient credit terms. An example for this is when firms offer lower interest rates. Using these lenient credit terms, the cash inflow declines.
Reduction of discretionary expenditures like for example advertising or R&D is another way of real earnings management. These expenditures are usually expensed in the same period that they are incurred. By reducing these expenditures the expenses decrease and the earnings increase. If these expenditures are managed that will result in an unusually low amount of them.
The last method of real earnings management described is overproduction. A higher production causes the fixed overhead costs to be spread over more goods than before. This results in lower total costs per good. Overproduction will lead to high production costs relatives to sales.
Cohen , Dey and Lys (2008) find that after the implementation of SOX the level of real earnings management increased and the level of accrual based earning management decreased. This means SOX caused firms to switch from one type of earnings management to the other. The reason for this switch is that real earnings
management is harder to detect. This type of earnings management is also more costly. Similarly Chi, Lisic and Pevzner (2011) find that real earnings management increases with high quality audits. Audit quality is measured with auditor industry expertise and the auditor being a big N firm. The reason for this finding is that high quality auditors constrain the possibility of accrual based earnings management, which again forces them to switch to real earnings management.
2.2.3 Incentives for earnings management 2.2.3.1 Different incentives
There are different incentives for earnings management. Healy and Wahlen (1999) describe 3 different incentives. The first is capital market motivations, which refers to the incentive that arises from investors and analysts who use the financial information to value stock. Manipulating earnings can influence the short-‐term stock price. The second incentive is one that is there because of contracting. Accounting data is used in contracts and can cause an incentive for managers to manipulate earnings. These contracts can be lending contracts or compensation contracts. The last incentives arise from regulatory reasons, which can be industry regulations and anti-‐trust or other regulations. Earnings management can be used to circumvent the regulations.
Beneish (2001) describes 4 different incentives. The first is debt contracts, managers want to avoid covenant default and they use earnings management to do so. The second is compensation agreements, which is also called the bonus hypotheses. The idea is that managers get a higher bonus if they change earnings through earnings management. The third is equity offerings, which causes earnings management because of the information asymmetry between owners-‐managers and investors. The last one is insider trading where income is increased to mislead investors.
Dichev et al. (2013) did research on the incentives for managers to engage in earnings management from the perspective of a CEO. They conclude that 20% of firms manage earnings in any of the periods they examined. From this earnings management, 60% relates to income increasing activities and 40% relates to income decreasing activities. The most important reasons for a CEO to manage earnings are to influence the stock price, when there is outside pressure to hit earnings benchmarks and to influence executive compensation (Dichev et al. 2013, p.26).
2.2.3.2 Positive accounting theory
The positive accounting theory describes the relation between the accounting choice of a firm and other firm variables. The three hypotheses that are mostly tested are the bonus plan hypothesis, the debt/equity hypothesis and the political cost hypothesis. These hypotheses are explained below.
The bonus plan hypothesis states that managers are more likely to use accounting that will increase income when they have bonus plans than when they do not have bonus plans. If the bonus is not likely to be paid this year as a result of not reaching the target, managers might use accounting which will decrease income. Doing this they increase the profits and their bonuses for the next years.
The debt/equity hypothesis is that managers are more likely to use income increasing accounting when the debt/equity ratio is higher. When the debt/equity ratio rises the firms get closer to constraints in the debt covenants. As the constraint gets tighter the probability of a covenant violation and the costs this creates get greater.
The political costs hypothesis predicts that firms with higher political costs are more likely to use income decreasing accounting techniques. Size is used as a proxy for political attention so the bigger the firm, the more they tend to use income decreasing techniques (Watts and Zimmerman, 1990).
2.2.4 Ethics and earnings management
Burns and Merchant (1990) performed a research regarding earnings management on 649 managers. The results show there is a large disagreement among managers on the opinion on what is ethical behavior, what is questionable behavior and what is unethical behavior. Because of these different opinions it is very hard to determine the quality of reported earnings. Also it shows that most managers engage in managing earnings. Although the methods that are used are legal, there is no consistency with a strict ethical framework. The managers questioned that were using earnings management did not believe they were doing something wrong. Burns and Merchant (1990) describe that to achieve an ethical approach to management there should be a balance between individual interests and stakeholders of any kind. According to them, managers are acting unethically when they use earnings management as they might be acting in their
own interest or even in interest of the firm, but they are not acting in the interest of stakeholders.
Another study done on ethics and earnings management is one done by Elias (2002). He examined different determinants of ethics among 763 accounting practitioners, faculty and students. The research provided different findings. First, the respondents in general classified accounting manipulations as minor to serious ethical infractions. Second, moral philosophies have an influence on the perception of what is ethical and what is not. Idealists were found to judge earnings management as more unethical than relativists. Finally, he found a relationship between the opinion about the perceived role of ethics and social responsibility and the perception of earnings management. If respondents believed the role of social responsibility was of a big importance they rated earnings management to be more unethical.
2.2.5 The board of directors and earnings management
In this paragraph the influence of the board of directors on earnings management is explained. Xie et al. (2003) examine the role of different committees and the board of directors on earnings management. From their research it seems that board composition does have an effect on earnings management. Their results show this for different characteristics. First, the number of board meetings is negatively correlated to earnings management. The reason they give for this finding is that active board may be better monitoring than boards that do not meet so often. The second finding is one regarding independent outside directors. If there are a lot of outside directors this provides a better monitoring board so this variable is also negatively correlated to earnings management. If the outside directors have a corporate background this also gives a negative relation as those directors are assumed to be financially sophisticated. Another is that if the board is bigger, earnings management is smaller. This is not what would be first expected because smaller boards are better monitors, but they explain the finding by the possibility of having more experienced directors on a board if the board is larger.
Klein (2002) also performed a study on the effect of board characteristics on earnings management. The measure she uses for earnings management is the adjusted abnormal accruals. The results show that firms which have boards where less than half
of the members is an independent director have higher levels of adjusted abnormal accruals. Consequently, in firms that change from board structure where the old structure was with a majority of independent board members and the new structure is with a minority of independent board members, the adjusted abnormal accruals increase a lot. Another test she performs is the effect of having a CEO who sits on the nominating or compensation committee. The effect of a CEO in the nominating committee was insignificant but for the compensation committee this was positively significant. This means having a CEO on the compensation committee increases earnings management.
Bergstresser and Phillipon (2004) also find that CEOs can influence earning management. Similar to the results above they find that in firms where compensation of the CEO is linked more to the value of stock, accrual based earnings management increases.
2.2.6 CEO tenure and earnings management
CEO tenure has an influence on the level of earnings management in the firm. This paragraph explains how the CEO tenure has an effect. Dechow and Sloan (1991) show that R&D expenditures of firms are less in the final years of a CEO. The reason they give for this is that the retirement benefits are based on the compensation that the executive received in final years. When the CEO is going to leave the firm, this increases incentives to manage earnings so their compensation based on earnings will be increased.
Davidson et al. (2007) find this same relation between a CEO leaving the firm and earnings management. They explain this with a slightly different reason. According to them CEOs are less concerned with the long-‐term if they are not going to be with the firm in the long-‐term. If the CEOs are not worried about firm performance, they will be more concerned about their own compensation, which gives them incentives to engage in earnings management. CEOs who plan to continue with the firm know that earnings management will reverse over time, so they are less likely to engage in earnings management.
Ali and Zhang (2015) also find that CEOs increase earnings management in the last year they work for the firm. They only found this result after controlling for the first
years of the firm in which CEOs also increase earnings management to make their reputation better.
2.3. Hypotheses
In this paragraph the hypotheses are formulated with which the research question will be answered. As shown above, there are differences between men and women in general and also in ethical behavior (Betz et al., 1989; Nguyen et al., 2008; Adams and Ferreira, 2009; Byrnes et al., 1999). It is said that earnings management can be seen as unethical behavior (Burns and Merchant, 1990; Elias, 2002). Literature also shows that the board of directors can influence earnings management (Xie et al., 2003; Klein, 2002; Bergstresser and Phillipon, 2004). As mentioned before women in general show more ethical behavior and also women are more risk averse than men (Nguyen et al., 2008; Ameen, 1996; Byrnes et al., 1999; Chaness and Gneezy, 2012). It is expected that when a firm has women in the board of directors, earnings management will decrease. The first hypothesis will test whether a female CEO in a firm has an influence on earnings management so this will be:
H1: Having a female CEO will reduce earnings management.
The second hypothesis tests whether CEO tenure has an effect on earnings management. As explained in the literature section 2.2.6, it is said that CEOs increase earnings management in their last year with the firm to increase their retirement profits and to increase their current profits as they are less concerned with the long-‐term profits of the firm. Assuming this the following hypothesis is formulated:
H2: The effect of female CEOs on earnings management will be smaller if the CEO is in his or her last year of being with the firm.
3 Research design
This paragraph first explains the data used and the selection of the sample. Second, the proxies for earnings management are described which is divided in a section for accrual based earnings management and a section for real earnings management. Last, the regressions used to test the hypotheses are given and the variables are described.
3.1 Data and sample selection
The research will be done using existing data from the Compustat database. The sample that will be used is S&P500 firms in the period 2002-‐2013. This is a period after the implementation of SOX so that this cannot have any influence on the sample. First, data was extracted and merged from the Compustat fundamentals annual database and from the Compustat Execucomp database. This created a dataset with information on CEO gender and tenure and gave a sample of 4390 observations.
Following the research of Kim, Park and Wier (2012) financial institutions are excluded because their accruals have different characteristics, this reduced the sample with 569 to 3821 observations. Deleting variables which are needed to calculate the proxies for earnings management reduced the sample to 909 observations. Last, the observations with missing control variables were removed which left the final sample with 887 observations. The table below shows how the sample was created.
TABLE 1 Sample selection
Firm-Year obs.
CEO gender and tenure coverage in Compustat from 2002-‐2013 4390
Financial institutions (569)
Missing variables for proxies (2912)
Missing control variables (22)
887
3.2 Proxies for earnings management
To measure earnings management different proxies were used. For accrual based earnings management two different proxies have been used and for real earnings management three proxies have been used. All the proxies are calculated with separate regressions. The paragraphs below describe the different proxies and how they are calculated.
3.2.1 Accrual based earnings management
For the amount of accrual based earnings management, two different models have been used to better test the hypotheses. The first proxy that will be used is the extent of estimation errors, which was introduced by Dechow and Dichev (2002). The idea of using estimation errors to proxy for earnings management is that when there is a good match between current accruals and past accruals there have been precise estimates. When the estimates were not precise, current and past accruals will not match. Therefore, if accruals are of good quality the accruals will map into cash flow realizations, as accruals are not included in cash flows. The regression used to measure the quality of earnings is:
ΔWCt/ 𝐴= β 0 + β 1 (CFOt-1 / 𝐴) + β 2 (CFOt / 𝐴) + β 3 (CFOt+1 / 𝐴) + ε1 (1)
Where:
∆WC = the difference in working capital between year t-‐1 and year t
𝐴 Is the average of the assets in year 1 and t-‐1 calculated as (Ait + Ait-1 )/2
CFO = the cash flow from operations β 0, β 1, β 2, β 3 = regression coefficients
ε = the error term which measures earnings quality because this captures the extent to which accruals map into cash flows.
The second model that will be used is the modified Jones model (Dechow, Sloan and Sweeney, 1995). In this model discretionary accruals are used to estimate earnings
management. The normal accruals are estimated using revenues, receivables and property, plant and equipment. The difference between the normal accruals and the total accruals are the discretionary accruals. The regression used is:
TAit/ Ait-1 = β0 + β1 (1/ Ait-1) + β2 ((ΔREVit - ΔRECit)/ Ait-1) + β3 (PPEit/ Ait-1) + εit (2)
Where:
TAit = total accruals for firm i in year t, which is measured by earnings before
extraordinary items minus operating cash flows following prior research (Ali and Zhang, 2015, Barua et al., 2010).
Ait-‐1 = the total assets of firm i at the beginning of the year t
∆REVt = the difference in revenues between year t-‐1 and year t
∆RECt = the difference in receivables between year t-‐1 and year t
PPEt = gross property plant and equipment in year t
β 1, β 2, β 3 are firm specific parameters
εit = residual of firm i in year t, which represent the discretionary accruals and which
measures earnings management.
3.2.2 Real earnings management
As described before, Roychowdhury (2006) gives three ways to engage in real earnings management, which are the manipulation of sales, the reduction of discretionary expenses and overproduction. He also gives different ways to measure these types. This research uses the same models.
The first model is used to calculate real earnings management through the manipulation of sales. The difference between the actual cash flow from operations and the normal cash flow from operations, which is calculated with sales and assets, is the abnormal cash flow from operations. This is calculated as follows:
CFOit/Ait-1 = β0 + β1(1/ Ait-1) + β2(Sit/ Ait-1) + β3(ΔSit/ Ait-1) + εit (3)
Where:
CFOit = total cash flow from operation for firm i in year t.
Ait-‐1 = total assets for firm i in year t
Sit = sales for firm i during period t
ΔSit = difference in sales for firm i between year t-‐1 and year t
β 1, β 2, β 3 are firm specific parameters
ε = the error term which measures earnings quality as it measures the abnormal cash flows
The second model calculates real earnings management through the reduction of discretionary expenses. This is calculated as a function of sales and assets:
DISEXPit/Ait-1 = β0 + β1(1/ Ait-1) + β2(Sit-1 / Ait-1) + εit (4)
Where:
DISEXPit = discretionary expenses for firm i in year t which is the sum of advertising
expenses, R&D expenses and selling, general and administrative expenses. Ait-‐1 = total assets for firm i in year t
Sit = sales for firm i during period t
β 1, β 2, β 3 are firm specific parameters
ε = the error term which measures earnings quality as it measures the abnormal discretionary expenses
The last model is used to calculate real earnings management through overproduction. The difference between the normal production costs and the actual production costs is the abnormal part of the production costs, which captures earnings management. This is calculated as follows:
PRODit/ Ait-1 = β0 + β1(1/ Ait-1) + β2(Sit / Ait-1) + β3(Δ Sit / Ait-1) + β4(Sit-1 / Ait-1) + εit (5)
Where:
PRODit = The production costs for firm i in year t. Calculated by COGS + ΔINV
Ait-‐1 = total assets for firm i in year t
Sit = sales for firm i during period t
ΔSit = difference in sales for firm i between year t-‐1 and year t
β 1, β 2, β 3 are firm specific parameters
ε = the error term which measures earnings quality as it measures the abnormal production costs
3.3 Regressions
To test the hypotheses a regression analysis is done. The regressions are simplified models of Kim, Park and Wier (2012) combined with variables used in Chi, Lisiz and Pevzner (2011) and Ali and Zhang (2015). The first regression is used to test accrual based earnings management and the second is used to test real earnings management. Both regressions are used to test hypothesis one. The regressions are the following:
AEMit = β0 + β1FEMALECEOit + β2SIZEit-1 + β3ΔE it-1 + β4BIG4it + β5LEVit-1 + β6RD_INTit +
Β7AD_INTit + β8ROAit-1 + β9AB_CFOit + β10AB_PRODit + β11AB_DISEXPit (6)
REMit = β0 + β1FEMALECEOit + β2SIZEit-1 + β3ΔE it-1 + β4BIG4it + β5LEVit-1 + β6RD_INTit
+Β7AD_INTit + β8ROAit-1 + β9AB_DAit (7)
Hypothesis two is also tested for accrual based earnings management and real based earnings management. Following the research of Ali and Zhang (2015), to test whether earnings management is more in the last year of the CEO a variable for the final year but also for the early years are included:
AEMit = β0 + β1FEMALECEOit + β2FINALYEARit + β3SIZEit-1 + β4ΔE it-1 + β5BIG4it + β6LEVit-1
+ β7RD_INTit + Β8AD_INTit + β9ROAit-1 + β10EARLYYEARSit + β11AB_CFOit + β12AB_PRODit +
β13AB_DISEXPit (8)
REMit = β0 + β1FEMALECEOit + β2FINALYEARit + β3SIZEit-1 + β4ΔE it-1 + β5BIG4it + β6LEVit-1 +
β7RD_INTit + Β8AD_INTit + β9ROAit-1 + β10EARLYYEARSit + β11AB_DAit (9)