Is there a link between CEO gender and accounting fraud s there a link between CEO gender and accounting frauds there a link between CEO gender and accounting fraud s there a link between CEO gender and accounting fraud

(1)

1

Amsterdam Business School

Is there a link between CEO gender and accounting fraud?

Name: Xin, Shuhe

Student number: 11089148

Thesis supervisor: Alexandros Sikalidis Date: 19 June 2016

Word count: 12,524

MSc Accountancy & Control, Accountancy

(2)

2 Statement of Originality

This document is written by Shuhe,Xin 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.

(3)

3 Abstract

In my study I investigate the correlation between Chief Executive Officer (CEO) gender and accounting fraud. Accounting fraud can be determined by accrual-based earnings management and real activities manipulation. Previous research shows that accounting fraud is closely related ethics and will lead to not only legal and financial consequences but also social consequences. There are 1015 observations in my sample, with 269 female observations and 746 male observations from year 2007 to 2013. After conducting the descriptive analysis and multiple element regression, I find that there is no significant correlation between CEO gender and accounting fraud, so my first hypothesis is rejected. For the second hypothesis, I cannot objectively measure it since the dummy variable is insignificant.

(4)

4

In recent decades, more and more accounting scandals have emerged which caught the public’s eyes. And so as researchers, they pay more attention to topics about accounting fraud and the mechanism behind the fraud. A typical scandal is the downfall of the American energy company Enron, which is regarded as the most serious audit failure that causes huge consequences. One of the recent accounting scandals is Tesco’s fraudulent reporting. Despite that Tesco is one of the most popular grocery and retail megastore, the financial report was still being manipulated with incomes overstated and costs understated to strive to compete in the market. Among many accounting frauds, like Enron scandal in 2001, WorldCom scandal in 2002, American International Group (AIG) Scandal in 2005 and Lehman Brothers Scandal in 2008, CEOs are the ones that play the main part. Behind the accounting scandals, the characteristics of CEOs are starting to catch researches’ attention, especially the gender of CEO.

Many parties have defined what fraud is. According to Statement on Auditing Standards No. 99, fraud is defined as “an intentional act that results in a material misstatement in financial statements”. Besides, the statement classifies two types of fraud: misstatements arising from fraudulent financial reporting, like falsification of accounting records and misstatements arising from misappropriation of assets like theft of assets or fraudulent expenditures. Wells (2009) suggests that fraud is not an accidental thing and all types of fraud share a common basis with four elements: a material false statement, knowledge of the falsity of the statement as intent, reliance on the false statement by the victim, and damages as a result. Three categories of fraud are defined under Wells (2009): asset misappropriation, corruption, and financial statement fraud.

(6)

6

Besides legal and financial consequences, accounting fraud also has impact on the whole society. For executives of firms undergoing legal affair of accounting fraud, their reputation drop to a great extent even if they do not engage in the accounting fraud, and the board memberships decrease as well after the lawsuit, according to perpetrators. Fich and Shivdasani (2007). For investors and users of financial reporting, fraudulent financial reporting or misstatements arising from misappropriation of assets could negatively influence the decision usefulness. With the financial information being manipulated, users cannot get a faithful representation of the real economic condition of the company and hence cannot make the right decision as they should make when the report is faithfully represented. The figures are being manipulated by executives in order to achieve a certain result, so the financial reporting is not useful for investors or users to make a decision. How useful the financial information could be depends on how good the quality of earnings is, as suggested by Dechow, Ge and Schrand (2010). So in this sense, accounting fraud is about to what extent the earnings are being managed and how good is the earnings quality.

Prior studies like Kim, Park and Wier (2012) have concluded that earnings management is about ethics. Firms that focus on trust and social responsibility, i.e., ethical firms, are less involved in earnings management by manipulating discretionary accruals and real activities. They tend to be more conservative when prepare financial information for their investors and users so that the financial reports are more transparent and will increase decision usefulness. There are studies suggesting that women tend to be more ethical than men, like Nguyen et al. (2008) and Heilman et al. (2004). Like the research conducted by Eagly (1987), women has an orientation towards relationship and more sensitive to the society. However, men tend to be more self-centered, aggressive and eager for success. Female CEOs are more risk averse and more sensitive to ethics, which in turn results in more conservative and more transparent financial reporting, as suggested by Ho et al. (2015).

(7)

7

With more and more female executives appearing on the high level of companies, it is valuable to investigate whether gender plays a part in affecting firm accounting perspective, in affecting quality of earnings and whether female CEOs who are considered to be more ethical would less likely to commit accounting fraud. There have been many studies about how executive gender would influence business running, however the area for accounting fraud is quite new, and that is what motivates me to conduct my research.

My research question is:

“Is there a link between CEO gender and accounting fraud?”

Conducting my research would be valuable for different parties. If there is a link between CEO gender and accounting fraud, then for firms when a new CEO need to be selected, they firm can take the gender of the candidates into account. For investors, if female CEOs will increase earnings quality so that less likely to commit accounting fraud, then investors will have more consideration about the firms they want to invest. If firms with female CEOs have higher earnings quality and accounting fraud, then the financial information provided to public would be more faithfully presented and more transparent, which increase the decision usefulness.

The rest of my paper is organized as follows: in chapter 2 I will discuss and analyze the previous studies in depth which helps me develop my hypotheses ; in chapter 3, I will introduce how my samples are selected and what my methodology is and regression models will be introduced in this section as well; in chapter 4, I will analyze and discuss the results; in the last chapter, I will summarize my results, answer the research question, list the limitations of my study and show the direction for future research in this field.

(8)

8

2. Literature review and hypothesis development

In this section, I will introduce the main theoretical background of my research. Firstly, I will explain what accounting fraud actually is and address factors that have influence on accounting fraud, which will be discussed in depth later. After that, I am going to tell the differences between male executives and female executives. I will address the firm performance difference and conservatism difference first, and then show some other differences. Based on the above, I will develop my hypotheses of the effect of CEO gender on accounting fraud.

2.1 Accounting fraud

The term fraud is defined by SAS 99 as the intentional activity which will lead to material misstatement in financial statements. There are two classifications of misstatement: misstatements arising from fraudulent financial reporting and misstatements arising from misappropriation of funds. Rubin (2007) suggests that besides fraud is an intentional activity, people commit fraud in order to achieve not fair or not legal benefit or gain. In this sense, accounting fraud is about manipulating financial statements to make its appearance look good for financial statement users, in other words, accounting fraud is fraudulent reporting. In short, intention is what that drives fraud. What is interesting is that, different from what people generally believe, fraudsters are just like other common people with similar education background, way to live, things for fun, according to the research of Ratley (2011). Besides being greedy and afraid of failure, Balaciu & Cosmina (2008) suggest that the

(9)

9

conflict of interest appears to be the core element that makes people have the intention to commit fraud. According to Baxevani and Mylonas, personal financial gain is the most effective motive of fraud and the possibility of immediate imprisonment of perpetrator is the most efficient deterrent. Other motivators, based on what have been found by Baxevani and Mylonas, include achievement of goals, reduction of company taxes and the ability to raise capital with low cost, preservation of current job status, possibility of fraud not to be detected and committing fraud for greater good. The authors also notify that there is a difference between men and women in the ranking of fraud motivation. Females put preservation of job status at the second place and females have greatly higher grade in personal financial gain and job status preservation than males, which suggests that these two factors are more efficient drivers for women than they are for men. Five deterrents have been investigated by Baxevani and Mylonas, including immediate imprisonment of the perpetrator, existence of strong internal control mechanisms, economic sanctions imposed as part of the penalty to fraud, reduced employment capabilities and the social impact. It has also been showed that only strong internal control mechanisms are related with the sex of respondent.

Several studies have been conducted in recent years. The study of“Fraudulent Financial Reporting: 1998–2007 ”shows that fraudulent reporting often involve top executives like CEO and CFO. However, lower level employees are mostly involved in conducting the practice of fraud schemes. Another research, “Report to the Nations: 2010 Global Fraud Study ” reveals that almost 14% of fraud cases involving executives or top management are fraudulent financial reporting cases which means that fraudulent financial reporting cases are concentrated among executive and upper-management perpetrators. A survey done by KPMG (2011) point out that people who conduct fraud are usually the ones that are accessible to sensitive information and able to exert controls to commit fraud. Olsen et al. (2013) shows that it is executive officers that usually be the main part in the process of fraud. As for CFOs, since they act in the interest of CEOs, if they are trying to manipulate earnings then it is usually

(10)

10

not for their own benefit, according to Feng et al. (2011). When the CEO compensation or bonus is more closely related with equity, they will exert their personal power on the company and try to manipulate earnings even if they do not have CFO around to be assisted. Accounting fraud is a big concern for investors since it may cause substantial losses and reduce investors’ confidence. Research done by Jonathan M. Karpoff, D. Scott Lee, and Gerald Martin (2008) shows that firms that commit fraud will be seriously punished by market. It is not just financial or legal influence, accounting fraud will also have social impact.

Accounting fraud is determined by two factors: accrual quality and real activities manipulation. These two factors will be discussed in the first two sub-paragraph. In the last sub-paragraph, I will discuss how CEO can affect accounting fraud.

2.1.1 Accrual quality

There are two classifications of earnings management, accruals management and real activities manipulation. According to Dechow and Skinner 2000, accruals management is about obscuring and masking true economic performance through accounting choices when following the accounting principles. Accrual management is conducted by manipulating accounting method choice used to represent those activities. Schipper defined earnings management as a purposeful intervention in the external financial reporting process, with the intention of obtaining some private gain.

It is suggested that financial reporting reliability is a function of firm internal control effectiveness, based on the research of Donaldson 2005. Ashbaugh-Skaife et al. (2007) indicate that when the firm has weak internal control, then managers are less capable of determining reliable amounts of accruals, resulting in financial information being

(11)

11

more noisy and less reliable. Besides, the weak internal control gives managers opportunity to override controls and purposefully prepare biased accrual estimates to help to meet their opportunistic financial reporting goals. Effective internal control helps investors by decreasing intentional and unintentional misstatements, including recording, measuring, and processing financial information, which leads to more reliable financial statements. The authors posit that firms with internal control deficiencies have lower quality accruals, and those firms have greatly bigger positive and negative abnormal accruals compared to firms with good control. According to Dechow and Dichev (2002), accruals bear the responsibility to shift or adjust recognition of cash flows over time in order to better measure firm performance through adjusted numbers, i.e., earnings. They hold the idea that accruals are temporary adjustments that resolve timing problems in underlying cash flows at the cost of making assumptions and estimates. In this sense, good estimates means that current and past accruals, present and future cash flows match well. However, incorrect or inaccurate estimates have negative effect on benefits that accruals would bring. Based on this sense, Dechow and Dichev (2002) define accrual quality as the extent to which accruals map into cash flow realizations. Research done by Dechow (1994) suggests that one of accounting accruals role is to provide a measure of shore-term performance that more closely reflects expected cash flow than do realize cash flows. Based on Dechow (1994), over short measurement intervals earnings have more tight association with stock returns than realized cash flow. What’s more, earnings are more greatly related with stock returns than realized cash flows in firms that undergo big changes in working capital requirements and investment and financing activities.

There is a trend for the quality of accruals and earnings to decrease when there is estimation error in accruals. There are both intentional and unintentional errors in regard to estimation errors. When the estimation error is caused by management manipulation, then it is intentional error. If it is caused by unintentional act, say, complicated calculation, then it would be unintentional error. Besides, what has found out by Dechow and Dichev (2002) is that firm and industry characteristics will be

(12)

12

systematically associated with accrual quality, even if there is no existence of intentional earnings management. They argue that the accrual quality will be lower if the operating cycle is longer, the firm is smaller, sales volatility has greater magnitude, cash flow volatility has greater magnitude, greater magnitude of accrual volatility, earnings volatility has greater magnitude, more frequent negative earnings reporting and greater magnitude of accruals.

2.1.2 Real activities manipulation

Another category of earnings management is real activities manipulation. When the manager takes actions that could alter timing or structuring of an operation, investment, and/or financing transaction in an effort to influence the output of the accounting system, RM could occur, as suggested by Gunny (2010).

Roychowdhury (2006) defines real activities manipulation as departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations. Different from accrual management, real activities manipulation is accomplished through changing the firm’s underlying operations in an effort to boost current-period earnings. Activities like overproduction to reduce cost of goods sold and reducing R&D expenses to increase current period earnings are real activities manipulation. The existence of real activities manipulation has been proved by prior research, like Roychowdhury (2006), Baber, Fairfield, and Haggard (1991), Bartov (1993) and Bens, Nagar, and Wong (2002).

Graham, Harvey, and Rajgopal (2005) claim that almost 80 percent of the interviewed executives would like to manipulate reporting perceptions even if to sacrifice

(13)

13

economic value. Based on research of Gunny (2010), there are several reasons why managers want to engage in RM instead of accruals management. For example, it’s more risky to manipulate accounting choices in regard of accruals because of Securities and Exchange Commission (SEC) scrutiny and class action litigation. What’s more, there are more limitations for firm to manipulate accruals. Gunny (2010) finds out that when size, performance and market-to-book is under control, there is a positive relation between RM and firms just meeting earnings benchmarks, which indicates that earnings management through real activities manipulation is in line with managers obtaining benefits which lead to better future performance. Prior study, like Roychowdhury (2006), found evidence also suggests that managers engaging in real activities to prevent financial reporting losses. Healy and Wahlen (1999) suggest demonstrate that earnings management will exist if managers put their judgment in accounting reports and build transactions to change reports in order to let stakeholders misinterpret the performance of company in the way that managers want, or just to manipulate contractual outcomes which rely on the financial reporting.

2.1.3 Chief Executive Officer impact on earnings quality

In this part I will illustrate what factors affect quality of accruals and real activities manipulation. According to Gunny (2010), earnings management has two classifications, accrual-based earnings management and real activities management. I will discuss these two later. Besides, I will analyze how chief executive officer can have impact on earnings quality. If the CEO has capability of influencing earnings, then CEO will affect earnings quality as well, because earnings management can influence earnings quality.

(14)

14

fundamental performance and measurement of performance. According to Statement of Financial Accounting Concepts No. 1 (SFAC No. 1), accounting reporting should reflect information of a company’s financial performance over a period of time. So in this sense, Dechow, P.M., W. Ge and C. Schrand (2010) state that high earnings quality is positively related with useful information about a company’s financial performance which has relevance in financial reporting users making decisions. There is no meaning of earnings quality unless it is defined under specific decision model, which means that earnings quality needs to be linked with decision-relevance of information. Besides, earnings quality is associated with the fact that whether the firm financial performance is informative to users. What’s more, the quality of reported earnings is determined by both financial performance relevance with decision-making and financial system capability to measure firm performance. Company’s financial performance and the accounting system which measures the performance together decide the company’s earnings quality. However, these two factors influence earnings quality to a different extent. Accruals have both beneficial and negative effects (Dechow and Dichev, 2002). For example, putting a receivable on the account increases the speed of future cash flow recognition in earnings, which betters maps the accounting recognition into future economic benefit and provides more useful information to decision makers. On the other hand, since accruals are usually determined by assumptions and estimates, if assumptions and estimates are wrong, then corrections are needed in the future. Still using receivable as an example, if the estimates of the original receivable is bigger than the assumed one, then the cash received and estimation correction need to be recorded in the subsequent entry. The estimation errors and following corrections are very noisy which can decrease the benefit of accruals. In this sense, with less accrual estimation errors, there will be higher accrual and earnings quality. A better match between working capital and cash flow from operations implies higher quality of accruals and earnings. Accurate estimations means that the current, past and future cash flows match well, while the inaccurate estimations decrease the benefits that accruals can bring. There are both intentional errors and unintentional errors. The unintentional errors are more

(15)

15

observable than intentional ones, because the intentional errors are created by management to manipulate earnings in order to achieve certain result that could benefit them own. The earnings management by managers reduce the decision usefulness for users of financial reporting since the information cannot fully reflect the true activities of the company. But the observable estimation errors are very common among companies. The quality of accruals is naturally influenced by those observable estimates errors and inherent company and industry characteristics. For example, the volatility of operations has direct influence on the occurring of estimation errors.

Real activities manipulation can also affect quality of earnings in a negative way. In order to meet specific profit goals or avoid losses on annual report, managers have incentives to manipulate real activities, which have influence on cash flows or accruals in certain aspect, according to Roychowdhury (2006). Roychowdhury (2006) defines real activities manipulation as management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds. For example, the manager could delay asset write-offs to present a higher income or reduce provisions for bad debts expenses. However, the real activities manipulation only brings benefit to the firm in short term and in the long term it will not increase benefit. This is also partly due to the fact that users and investors are becoming more experienced now and the real activities cannot be manipulated to fool them all the time. Real activities manipulation is more likely to occur when it is closing to debt covenants, when there are receivables, stock of inventories and growing chances. Gunny (2010) states that when managers try to alter the timing or structure of an activity in order to achieve certain outcome from financial system, then real activities manipulation emerge. Different from changing accounting methods of operating activities, real activities manipulation alters the operations in order to increase short-term profit.

In the above part I discussed that quality of earnings can be affected by both accrual-based earnings management and real activities manipulation. In the following section

(16)

16

I will illustrate to what extent the Chief Executive Officer is capable of influencing earnings quality.

According to Bergstresser and Philippon (2006), CEO has incentives to manage earnings when the compensation is related to stock price and option holdings. Besides, CEOs exercise uncommonly huge amount of options in the years of high accruals. Meanwhile, CEOs and other inside traders sell huge numbers of shares. It is said that exposing CEO to stock prices more helps to link top management incentives with shareholders’ interests, but this could cause problems. For example, when the compensation is closely related with options, then managers will have incentives to manipulate earnings in order to get higher compensation. Managers may manipulate accruals to change reported outcomes, since accruals is one of the factors that determine earnings management. Accruals are part of the earnings which are not shown in current cash flows and need a lot of discretion to estimate. Since the reported income contains both cash flows and earnings not shown in current cash flows, managers therefore have opportunities to manage earnings. Healy (1985) indicates that executives usually manage earnings when their bonus is linked to reported income. For example, when it is in the middle of the year, managers tend to accelerate speed of increasing earnings in order to make higher profit and higher bonus for this year. While when it is close to year end, they tend to slow down the speed in order to make higher profit and get higher bonus for next year. Beneish and Vargus (2002) find out that during the time when managers or their firms try to sell shares to the capital market, managers also have incentives to manage earnings. High levels of accruals are followed by insiders selling shares.

Besides CEOs have incentives to manipulate earnings, CFOs also have incentives to do so and even more. According to Jiang, Petroni and Wang (2010), since chief financial officers mainly focus on preparing financial reporting, their equity incentives are more closely related with earnings management compared to chief executive officers. CEOs try to manipulate accruals and meet or beat goals so that they can get higher bonus,

(17)

17

and so do CFOs, who show a stronger willing to do so. Some previous research (like McAnally, Weaver, and Srivastava, 2008, p. 194) states that since CEO equity incentives are way bigger than CFO’s, the CEO equity incentives are supposed to be the dominant one. Besides, it has been said CFOs correspond with CEOs equity incentives so that they cannot react to their own equity incentives in a direct way, due to the fact that CFOs represent CEOs and the latter can remove the former that do not follow their wishes (Graham and Harvey 2001, Mian 2001, Fee and Hadlock, 2004) . However, there are other studies show that CFOs act independently on financial reporting. Geiger and North (2006) find that around the period of naming a new CFO, discretionary accruals drop to a great extent and that is not because of the CEO nomination at the same time. What’s more, when the last CFO fails to meet the earnings forecast, then the CFO turnover will increase, according to Mergenthaler, Rajgopal, and Srinivasan (2008). Jiang, Petroni and Wang (2010) allege that the level of accruals corresponding with CFO equity incentives is much higher than that of CEO equity incentives. CFO and CEO both need to meet or beat earnings forecast and CFO equity incentive ratios are much higher than that of CEO. CFO equity incentives act independently from whole company’s earnings management activities and CFO equity incentives even play a greater role than CEO equity incentives.

In conclusion, accrual-based earnings management and real activities manipulation together jeopardize the quality of earnings. Besides, CEOs have incentives to engage in earnings management in order to achieve higher compensation or meet or beat earnings forecast. What’s more, CFO equity incentives are even bigger than that of CEO.

(18)

18

In this section, I am going to introduce how ethics have effect on earnings management. I will discuss whether more ethical firms will engage in less earnings management activities.

The ethics play a role in earnings management. The necessity to recognize acceptable and unacceptable actions drives the forming of organizational control systems and professional code of conduct, according to Merchant and Rockness (1994). However, it may be hard to distinguish what kind of behavior is socially acceptable and what is not. Managers conduct earnings management to achieve certain result but the manipulated financial report will mislead users, and the actions could be illegal. Merchant and Rockness (1994) conducted their research through a questionnaire which contains 13 doubtful earnings management activities for samples to judge the acceptance. Actually many earnings management activities are legal and are not conducted against accounting principles, like GAAP. But the problem is whether it is right for managers to manipulate earnings. If the activities are not conducted in a socially acceptable way then laws or code of conduct are needed to rule those actions. Almost every fraudulent reporting case begins with little ethical problem, according to The National Commission on Fraudulent Financial Reporting (1987). Merchant and Rockness (1994) indicate that the consideration of whether earnings management activities are acceptable or not depends on type, size, timing and purpose of the activities.

Kim, Park and Wier (2012) find out that firms that are more ethical engage in less earnings management activities. They suggest that socially responsible firms have less likelihood to engage in earnings management by manipulating discretionary accruals, more likely to not manipulate real activities from manipulation. They adopt discretionary accruals, real activities manipulation and the incidence of Accounting and Auditing Enforcement Releases (AAERs) as proxies for earnings management. Jones (1995) suggests that firms focus on trust and cooperation in their business tend to commit to ethical behaviors. Firms that are socially responsible have a tendency to

(19)

19

make efforts in conducting CSR practices in order to meet stakeholders’ expectations. Besides, they usually try to limit earnings management activities so as to provide more faithfully presented financial information for investors. As for the three proxies Kim, Park and Wier (2012) use, they conclude that CSR firms, which are more ethical, tend to engage in less earnings management activities. Less earnings management through discretionary accruals are conducted in CSR firms and they manipulate less real activities from operations. The CEOs and CFOs of those firms are also less likely to be investigated for violating GAAP by SEC.

From the above analysis I can conclude that earnings management is an ethical problem, and both accrual-based earnings management and real activities manipulation are unethical behaviors. Ethical firms will engage in less earnings management activities and try to provide more faithful financial information for investors.

2.2 Difference between male and female CEO

In this part, I will show the differences between male and female Chief Executive Officer. I will address three differences. Firstly, firms with female CEO in charge have better performance. Secondly, female CEOs are more conservative in accounting compared to male CEOs. Finally, I will address some other differences between male and female CEOs.

Carter et al. (2003) argue that a board with management diversity will generate more choices when making a decision than board with less management diversity, since more management diversity, more evaluation options.Management diversity can be defined as the proportion of women among the highest-ranking CEOs in firms and on

(20)

20

boards of directors. Because of different experiences from life, females may have experiences that males don’t have and women may comprehend things better than men in perspective of some elements of firm’s market place, which could have positive effect on the quality and creativity of board decision-making process. So based on the above research, I find it is very important to include my flowing demonstration.

Steffensmeier et al. (2013) allege that females usually have the least engagement in the process of committing fraud and they generally do not take the leading part. Under general conditions, when women engage in fraud it is because of their personal relationships with the person who commits the fraud or the necessity to take part in the fraud in order to achieve success for women who are in the high level of the company. The research indicates that with more women in the industry, there will be more women that commit fraud. In other words, it is proportionate between female fraudsters and number of females in the industry. In this sense, females are males are not different in the likelihood to commit fraud.

2.2.1 Better firm performance with female CEO in charge

According to Krishnan and Parsons (2008), firms will have higher earnings quality when there is gender diversity in senior management. Besides, firms with higher proportion of female executives appears to be more profitable and enjoy higher stock returns compared to lower proportion of female executives. Results from Erhardt, Werbel and Shrader (2003) show that profitability is positively associated with the number of women in senior management. Smith, Smith and Verner (2006) suggest that when observed factors such as firms’ age, size, sector and export orientation have been controlled, the percentage of females among top executives and on boards of directors will positively influence firm performance, and what’s more, a major part of the effect

(21)

21

is due to good education background which better qualifies females. Dezso and Ross (2012) also find out that with females in top management, firms will have better performance to the extent that it is focused on innovation as part of the strategy.

In the paper of Khan and Vieito (2013), it addresses that when the firm is with female CEO in charge, then there will be better firm performance compared with male CEO in charge. The firm risk level is smaller for firm with female CEO in charge than male CEO. Besides, when boards design compensation packages, they seem not consider risk aversion differences of gender. An explanation of this is that boards want to encourage female CEOs to take more risks when they award same amount of risk to female CEOs as to male CEOs. They find out that when the firm is managed by a female CEO, the ROA will be higher than a firm managed by a male CEO, which means that female executives tend to provide higher returns to equity and shareholders. There is a tendency for firms to appoint a female CEO in order to have lower risk, since female CEOs are generally considered to be less risky than male CEOs. However, firms with female CEOs are usually smaller in size than those with male CEOs, and firm size is negatively related with firm risk level, which means that smaller firms have higher risk level, so small firms tend to be more risky even if it is managed by a female CEO. Besides, firm risk level is positively related with female CEO’s firm stock ownership, which suggests that higher the stock ownership of female CEOs, higher the firm risk level. Khan and Vieito (2013) conclude that female CEOs will usually lead to an improvement in firm performance and the risk level of firm tend to be smaller with a female CEO in charge.

2.2.2 Differences in accounting conservatism

(22)

22

principles of accounting and it reflects manager’s opinion of risk. Watts (2003a) demonstrates that accounting conservatism can reduce conflicts between management and other related parties. Biddle, Ma, and Song (2010) suggest that default risk is determined in the same time with accounting conservatism. Prior studies like Barua, Davidson, Rama, and Thiruvadi (2010) have showed that lower absolute discretionary accruals or higher income-decreasing discretionary accruals will occur when firms is with female CFOs in charge. Generally, business decisions made including financing, investing, merging and acquisitions are very different between male and female CEOs, as research conducted by Mohan and Chen (2004), Levi, Li, and Zhang (2008) and Huang and Kisgen (2013) shows. Risk aversion and ethical sensitivity are two characteristics of the leadership characteristics of female executives that are closely related with conservatism in financial reporting and fighting against fraud, according to Ho et al. 2014. CEO gender is positively related with accounting conservatism, with female CEOs report more conservative earnings. Specifically, if the litigation and takeover risks of the firms are higher, the relation would be stronger. Also, based on Ho et al, gender effect is more obvious in firms with smaller size and stronger corporate governance.

Francis et al. (2015) have found the difference between genders in accounting conservatism. According to the research, the appointment of a female CFO greatly raise the level of accounting conservatism compared to the appointment of a male CFO. Female CFOs are positively related with accounting conservatism only under the condition that companies have higher risk of litigation, system or management turnover. One difference between male and female CFOs is that female CFOs are more likely to not adopt equity-based compensation. Another difference is that female CFOs tend to choose more tangible assets as company’s investment pattern, instead of intangible assets. What’s more, there is a greater possibility for female CFOs to decrease dividend payouts. All these findings imply the increase in accounting conservatism after the appointment of a female CFO. Even if Francis et al. (2015) only investigate the correlation between CFO gender and accounting conservatism, the

(23)

23

study also gives suggestion for CEO gender.

2.2.3 Other differences

Another difference is that male CEOs are more overconfident than female CEOs. Huang and Kisgen (2012) find out that female executives undertake less acquisitions and issue debt less often than male executives. What’s more, acquisitions made by companies with female executives have a bit higher announcement returns than with male executives. Collectively, it suggests that male CEOs are overconfident in decision making relative to female CEOs.

There have been a lot of literatures in the area of sociology, psychology, and economics showing that females are usually more risk averse than males, like research done by Eckel and Grossman (2004) and Croson and Gneezy (2009). The research conducted by Byrnes, Miller and Schafer (1999) suggests that women are more cautious and risk-averse when making decisions and business consideration compared to men. In this sense, females will need more evidence to recognize revenues.

2.3 Hypotheses

In conclusion, accounting fraud is determined by accrual quality and real activities manipulation. Accruals bear the responsibility to shift or adjust recognition of cash flows over time in order to better measure firm performance through adjusted numbers, and that is earnings. The better the accruals map into cash flow realizations, the higher the accrual quality is. Real activities management could occur when the

(24)

24

manager takes actions that could alter timing or structuring of an operation, investment, and/or financing transaction in an effort to influence the output of the accounting system. If the firm is with female CEO in charge, then it will have better firm performance relative to firm with male CEO in charge, since risk level of the firm is smaller. It is suggested that female CEO is more conservative when it comes to report earnings. In addition, male CEOs are overconfident relative to females and would undergo more acquisitions and issue more debts. When females make business decisions, they tend to be more risk-averse and cautious. Based on these arguments, I predict that there will be less accounting fraud with female CEO in charge of the firm. Here are my hypotheses:

H1: There will be higher accrual quality and less real activities manipulation when female CEO are in charge.

What’s more, when I look at the research done by Merchant and Rockness (1994), in which they investigate the prevailing morality in respect of earnings management, I realize that there could be a difference between real activities manipulation and accrual-based earnings management in the extent to be effected by CEO gender. According to Merchant and Rockness (1994), accrual-based earnings management is perceived to be more unethical than real activities management, which leads me to assume that female CEO influence accrual-based earnings management to a greater extent than real earnings management, since ethics have bigger impact on accrual-based earnings management. So, here is my second hypothesis:

H2: A female CEO will influence accrual-based earnings management to a greater extent than real activities manipulation.

(25)

25

3. Data and method

I’m going to conduct a quantitative research by using database. I find my data from Fundamental Annual database, Simplified Financial Statement Extract database and Execucomp database from Compustat-Capital IQ in Wharton wrds. I will explain how I choose my data in this chapter and introduce the models to be used for my research question. After that, I will discuss how accounting fraud is measured.

3.1 Sample selection

I choose companies through a NASDAQ list from year 2007 to 2013. There is a practical reason for doing that: the United States is the only country with accessible executive data in the Execucomp database. At first, there are 3111 companies in the NASDAQ company list, but in the end only 145 companies left with all the available information I need for the research. From the beginning I wanted to adopt data from 2005 to 2015, because I thought with bigger sample, the result would be more representative. But since so much information was lost for some years, 2007—2013 would be a better period for me to find data.

After my first collection of data, I found that data for research and development and advertising expenses are the ones with most lost information, so I started to delete data from these two. Then again and again, after deleting all the companies without enough information from 2007 to 2013, I can only find 145 companies with all the available data, which means there are 1015 samples left for my study.

(26)

26

3.2 Research design

To conduct my research, I learn from prior studies (Jones 1991, Subramanyam 1996, DeFond and Subramanyam 1998, Kothari et al. 2005, Roychowdhury 2006, Cohen et al. 2008, Cohen and Zarowin 2010, Badertscher 2011 and Zang 2012) and build a regression model. I adopt different proxies for accrual quality and real activities manipulation to measure earnings management in order to detect whether there is a relation between CEO gender and accounting fraud. The measures will be discussed in 3.2.1. The regression model will be discussed in section 3.2.2.

3.2.1 Proxies for fraud

For my research, I use accrual quality and real activities manipulation as measures for fraud. I use the amount of discretionary accruals to measure accrual quality. And for real activities manipulation, I use three individual proxies and one combined proxy.

3.2.1.1 Accrual quality

There has been many researches that adopt discretionary accruals as proxies for earnings quality and earnings management, for example, Jones 1991, Subramanyam 1996, DeFond and Subramanyam 1998 and Kothari et al. 2005. Based on prior studies, I want to use discretionary accruals as the first proxy for fraud. Just like DeFond and Subramanyam 1998, I want to use the modified Jones model in a cross-sectional version, because it has better specification and broader data requirement. As in Kothari et al. 2005, I want to make use of return on asset (ROA) in prior year to act as

(27)

27

a regressor in model. I do so in order to put performance effect on measured discretionary accruals under control. I adopt the absolute value of discretionary accruals (ABS_DA) for the main analyses, since earnings management include either income-increasing or income-decreasing accruals. If the outcome is in line with the transparent financial reporting (opportunistic financial reporting) hypothesis, then a positive (negative) relation can be expected between CEO gender and the absolute discretionary accruals.

TAit/A 1=α0(1/A 1)+ α1(ΔREVΔRECit)/ A 1+α2PPEit/ A 1+α3IBXI1/ A it-1+εit (1)

Where:

TAit= total accruals for a firm i at year t;

ΔREVit= change in net revenues in year t from year t-1; ΔRECit= change in net receivables;

PPEit= gross property, plant, and equipment;

IBXIit-1= income before extraordinary items at year t-1; and A it-1= lagged total assets.

3.2.1.2 Real activities manipulation

I make use of past research to form the proxies for real activities manipulation, like Roychowdhury 2006, Cohen et al. 2008, Cohen and Zarowin 2010, Badertscher 2011 and Zang 2012. Following Kim et al (2012), the following four measures are adopted to detect real activities manipulation: (1) abnormal levels of operating cash flows (AB_CFO), (2) abnormal production costs (AB_PROD), (3) abnormal discretionary expenses (AB_EXP), and (4) a combined measure of real activities manipulation. Three

(28)

28

individual proxies and a combined one are used, borrowing from Cohen et al. (2008) (Cohen, D., A. Dey, and T. Lys. 2008. Real and accrual-based earnings management in the pre- and post-Sarbanes-Oxley periods. The Accounting Review 83 (3): 757–787.). Due to the expected discretions of the first three variables, COMBINED_RAM as AB_CFO -AB_PROD + AB_EXP is being calculated. If the consequence is in line with transparent financial reporting (opportunistic financial reporting) hypothesis, then a positive (negative) association can be expected between CEO gender and AB_CFO, AB_EXP, and COMBINED_RAM, and a negative (positive) association with AB_PROD.

I make use of model from Roychowdhury’s (2006) to estimate the normal level of operating cash flows:

CFOt/At-1=α0+α1(1/At-1)+ β1(St/At-1)+ β2(ΔSt/At-1) +εit (2)

Where:

CFOt= cash flow from operations in year t; A= total assets;

S= net sales; and ΔS= St – St-1

Abnormal production costs is another proxy to measure real activities manipulation. Following the research done by Roychowdhury (2006), Cohen et al. (2008), Badertscher (2011) and Zang (2012), the model below is to be estimated for normal cost of goods sold (COGS):

COGSt/At-1=α0+α1(1/At-1)+ β(St/At-1) +εit (3)

(29)

29

Where COGSt= the cost of goods sold in year t. Model for normal inventory growth is estimated in a similar way:

ΔINVt/At-1=α0+α1(1/At-1)+ β1(St/At-1) +β2(St-1/At-1)+εit (4)

Where ΔINVt = change in inventory in year t. Based on Roychowdhury (2006), Cohen et al. (2008), Badertscher (2011), and Zang (2012), production cost can be defined as PRODt= COGSt +ΔINVt.. Making use of equation (3) and (4), I want to estimate normal production costs through the equation below:

PRODt/At-1=α0+α1(1/At-1) + β1(St/At-1) +β2(ΔSt/At-1)+ β3(ΔSt-1/At-1)+εit (5)

Abnormal production cost (AB_PROD) is the residual from the model.

Abnormal discretionary expenses is the third measure for real activities manipulation. Making use of Roychowdhury (2006), Cohen et al. (2008), Badertscher (2011), and Zang (2012), the following model is used to estimate the normal level of discretionary expenses:

DISEXPt/At-1=α0+α1(1/At-1) + β(St-1/At-1) +εit (6)

Where DISEXPt= discretionary expenses in year t. It is the total of R&D, Advertising, and SG&A expenses. For every firm-year, abnormal discretionary expenditure (AB_EXP) is the residual from the model.

Learning from Cohen et al. (2008), a combined proxy of real activities manipulation is built by combining those individual measures, AB_CFO, AB_PROD, and AB_EXP. The combined proxy COMBINED_RAM is calculated as AB_CFO - AB_PROD + AB_EXP, taking

(30)

30

the direction of each real activities manipulation element in to consideration.

3.2.2 Regression models

To investigate whether there is an association between CEO gender and accounting fraud, I want to estimate the model below:

ABS_DAit=α0+α1FEMALECEOit+α2COMBINED_RAMit+α3SIZEit-1+α4ADJ¬_ROAit-1 +α5LEVit-1+α6RD_INTit+α7AD_IND_INTit+εit (1) RAM_PROXYit=α0+α1FEMALECEOit+α2ABS_DAit+α3SIZEit-1+α4ADJ-_ROAit-1+α5LEVit-1+α6RD_INTit+α7AD_IND_INTit+εit (2) Where:

ABS_DA=absolute value of discretionary accruals (signed discretionary accruals), where discretionary accruals are computed through the cross-sectional modified Jones model adjusted for performance;

RAM_PROXY=AB_CFO, AB_PROD, AB_EXP, or COMBINED_RAM: AB_CFO= the level of abnormal cash flows from operations;

AB_PROD= the level of abnormal production costs, where production costs are defined as the sum of cost of goods sold and the change in inventories;

AB_EXP= the level of abnormal discretionary expenses, where discretionary expenses are the sum of R&D expenses, advertising expenses, and SG&A expenses;

(31)

31

COMBINED_RAM= AB_CFO- AB_PROD+ AB_EXP

FEMALECEO=1 when CEO of the firm i is a female in year t and 0 otherwise. SIZE= natural logarithm of the market value of equity (MVE);

ADJ_ROA=industry-adjusted ROA, where ROA is measured as income before extraordinary items, scaled by lagged total assets;

LEV=long-term debt scaled by total assets;

RD_INT= R&D intensity (R&D expense/net sales) for the year;

AD_IND_INT=advertising intensity for the two-digit SIC code industry for the year;

I make use of multiple regressions to estimate equations (1) and (2). If firms intend to manipulate reported earnings, they will probably use a combination of discretionary accruals and real activities manipulation. In other sense, the firm can make decision between two options based on which mechanism costs less (Cohen et al. 2008; Zang 2012). According to Zang (2012), which mechanism to choose of the two earnings management options depends on the relative cost. Learning from Cohen et al. (2008), I introduce ABS_DA as a measure for accrual-based earnings management, acting as a control variable in the real activities manipulation regressions, i.e., RAM_PROXY. Besides, it is also a proxy for real activities manipulation and acts as a control variable in the accrual-based earnings management regressions and that is ABS_DA or DA.

I introduce different control variables which can have influence on financial reporting behavior and CEO gender in order to prevent issues of correlated omitted variables. The big difference in earnings management can be explained by research of Roychowdhury (2006), which indicates that each firm growth opportunity and firm size could have an influence on it. In this sense, I want to adopt proxies for firm growth opportunity (MB) and size of the firm (SIZE). Industry-adjusted ROA (ADJ_ROA) is also included in the regressions in order to avoid the influence of the ethical perspective of CEO gender on earnings management.

(32)

32

of to what extend earnings management has influence. So I introduce BIG4 as an indicator variable for the firms that adopt one of Big Four auditors in the regressions. Leverage is introduced to control leverage-related motivation for earnings management. Similarly, an indicator for the incidence of an equity offering during the following fiscal year is also introduced to control equity-offering-related incentives for earnings management. According to McWilliams and Siegel (2000), there is a positive relation between R&D intensity and advertising intensity and earnings, so I introduce RD_INT and AD_IND_INT as controls for firm R&D expenditure and the advertising intensity.

Since corporate governance have influence on financial reporting behaviors, I introduce a net score of KLD’s corporate governance ratings (GOVERNANCE) in the regression model to control corporate governance. FIRM_AGE is adopted to control potential influence through various development phases of business since financial reporting behavior can chance in accordance with the maturing of a firm. There is a positive association between firm reputation and its earnings performance, according to Musteen et al. (2009). ADMIRED is being introduced to control for the possibility of KLD’s evaluation affected by firm reputation.

4. Results

In this chapter I’m going to illustrate the results of my research. First, I will discuss the descriptive statistics of the study. Then, I will analyze the main part of my research. At the end, there will be additional test I conducted to further explain the research question.

(33)

33

4.1 Descriptive statistics

The descriptive statistics of the full sample are listed in Table 1. The median of absolute value of discretionary accruals is 0.43 and mean is 0.80, which is higher than what previous research has concluded, like Kim et al. (2012) and Barua et al. (2010). I think of a reason which can be that the sample in my study is relatively small, with 145 firms from year 2007 to 2013, i.e. 1015 samples in total. Due to the demands of each company having data for every variable from 2007 to 2013, it is understandable that only 1015 samples left after removing unqualified ones. For other dependent variables AB_CFO, AB_PROD, AB_EXP and COMBINED_RAM, the means are 0.26, -0.08, -0.06 and 0.81 respectively. Meanwhile, for control variables, the means of ADJ_ROA, LEV, RD_INT, AD_IND_INT are 0.07, 0.13, 0.13 and 0.02 respectively, which means that discretionary accruals are not normally manipulated through return on assets, long-term debt scaled by assets, research and development expense intensity and advertising intensity. However, means for COMBINED_RAM and SIZE are -0.54 and 7.63 respectively, which suggests that they are usually manipulated for discretionary accruals. The median and standard deviation of variables can be found in Table 1.

(34)

34

TABLE 1

Descriptive Statistics full sample

Full sample N= 1015

VARIABLES _MEAN _MEDIAN _{STD. DEVIATION}

ABS_DA 0.80 0.43 0.39 AB_CFO 0.26 0.13 0.48 AB_PROD -0.08 -0.07 0.19 AB_EXP -0.06 -0.03 0.40 COMBINED_RAM -0.54 -0.24 0.81 SIZE 7.63 6.38 0.43 ADJ_ROA 0.07 0.06 0.13 LEV 0.13 0.11 0.24 RD_INT 0.13 0.09 0.10 AD_IND_INT 0.02 0.02 0.03

(35)

35

TABLE 2

Descriptive Statistics of Female CEO firms versus Male CEO firms

Female CEO Male CEO N= 269 N= 746

VARIABLES

MEAN MEDIAN MEAN MEDIAN

P-VALUES (MEAN/MEDIAN) ABS_DA _0.81 _0.42 _0.85 _0.45 _0.87/0.89 AB_CFO _0.36 _0.17 _0.34 _0.15 _0.19/0.54 AB_PROD _-0.08 _-0.06 _-0.09 _-0.07 _0.67/0.71 AB_EXP _-0.07 _-0.04 _-0.08 _-0.05 _0.24/0.78 COMBINED_RAM _-0.49 _-0.12 _-0.38 _-0.09 _0.42/0.79 SIZE _7.36 _7.00 _7.81 _7.00 _0.03/0.09 ADJ_ROA _0.06 _0.06 _0.06 _0.05 _0.35/0.78 LEV _0.16 _0.12 _0.11 _0.09 _0.09/0.04 RD_INT _0.14 _0.09 _0.13 _0.11 _0.89/0.13 AD_IND_INT _0.03 _0.02 _0.03 _0.02 _0.05/0.01

In Table 2 I divided the samples into two groups, male CEOs and female CEOs, in order to have a more clear comparison between the CEO gender effect on accounting fraud. The p-values show the test of difference between male and female mean or median. I am going to illustrate my results in two parts, significant results and insignificant results. Firstly, I will discuss the insignificant variables between male and female CEOs. The dependent variables ABS_DA, AB_CFO, AB_PROD, AB_EXP and COMBINED_RAM all turn out to be insignificant, with AB_CFO closest to be significant, the p-value of which is 0.19. In this sense, just based on the descriptive statistics, there is no evidence

(36)

36

to support the prediction that firms with female chief executive officer have higher earnings quality than those with male chief executive officer. As for control variables, the differences are not significant for ADJ_ROA and RD_INT, which means that I cannot conclude whether firms with female CEOs have higher return on asset or invest more in research and development than firms with male CEOs.

The differences between male and female CEOs are significant for control variables SIZE, LEV and AD_IND_INT. The means of SIZE for firms with female CEOs is lower than those with male CEOs, with 6.94 compared to 7.81 and p-value is 0.03. This suggests that firms with female CEOs is usually smaller than those with male CEOs. However, what is different from my expectation is that the results show that firms with female CEOs tend to be more leveraged than firms with male CEOs, which is 0.16 compared to 0.11, with p-value equals 0.04. Perhaps this can be explained by firm size. As explained earlier, firms with female CEOs are usually smaller, so it would be easier for chief executive officer to manipulate leverage. As for advertising expense intensity, the means are 0.04 and 0.03 for firms with female and male CEOs, with p-value equals 0.04. The result indicates that firms with female CEOs tend to invest more in advertising expenses.

4.2 CEO gender influence on accrual quality

In Table 3 I will show the results of regression tests and make analysis about them. I used absolute value of discretionary accruals (ABS_DA) as proxy for accrual-based earnings management. FEMALECEO is used as the dummy variable for the regression model, and if the CEO is female then FEMALECEO=1, otherwise 0. Higher the absolute value of discretionary accruals, higher the levels of earnings management, which implies accrual quality and earnings quality will be lower. I expect that with female

(37)

37

CEO on board, there will be higher accrual quality and earnings quality, so the correlation of FEMALECEO would be negative.

After conducting the regression tests, I found that the coefficient of FEMALECEO to ABS_DA is negative, -0.005, but the significance is more than 0.05, which means that the result is far from significant. As for COMBINED _RAM, the coefficient is negative, with p-value more than 0.05. ADJ_ROA is negatively related with ABS_DA, but the p- value is more than 0.05, which means that the result is not significant, and so is RD_INT. LEV is positively related with ABS_DA and p value lies between 0.01 and 0.05. There is negative relation between AD_IND_INT and ABS_DA, with p-value bigger than 0.01 and less than 0.05. SIZE is negatively correlated with ABS_DA, with p-value less than 0.001.The results show that if the CEO is female, then firms are more likely to be leveraged. Besides, those firms tend to have higher advertising expense, and this result is significant. The figures show that firms with female CEO are usually smaller in size, which is significant.

From the analysis above, different from my expectation, I conclude that female CEOs do not manipulate less discretionary accruals than male CEOs, which means that they do not necessarily engage in less accrual-based earnings management. The relatively small sample size can be the reason for not finding a significant correlation between CEO gender and accrual-based earnings management. Besides that reason, there are 269 females and 746 males in total samples, with male CEOs outnumbered than female CEOs, which can contribute for the insignificance.

(38)

38

TABLE 3

Accrual-based Earnings Management and Real Activities Manipulation

N= 1,015

VARIABLES ABS_DA AB_PROD AB_CFO AB_EXP COMBINED_RAM COEFFICIENT FEMALECEO -.005 -.034* -.045 -.045 .056 SIZE -.189*** .567*** .678*** .789*** .350*** ADJ_ROA -.025 .067* .345*** .034** .056** LEV .043** .045 -.057 -.005 -.078* RD_INT -.067 -.456*** .023 .045** .369*** AD_IND_INT -.076** .056 .024 .224*** .256*** ABS_DA -.243*** -.456*** -.568*** -.356*** AB_PROD -.243*** .667*** .698*** -.588*** AB_CFO -.468*** .550*** .799*** .256*** AB_EXP -.357*** .679*** .678*** .267*** COMBINED_RAM -.332*** -.456*** .234*** .345***

SIG. (1-TAILED) FEMALECEO .876 .076 .115 .298 .179

SIZE .000 .000 .000 .000 .000 ADJ_ROA .365 .091 .000 .045 .020 LEV .041 .433 .125 .564 .059 RD_INT .167 .000 .440 .034 .000 AD_IND_INT .045 .298 .245 .000 .000 ABS_DA .000 .000 .000 .000 AB_PROD .000 .000 .000 .000 AB_CFO .000 .000 .000 .000 AB_EXP .000 .000 .000 .000 COMBINED_RAM .000 .000 .000 .000

*,** and *** represent significance levels of 10%, 5% and 1%

4.3 CEO gender influence on real activities manipulation

In table 3, the results of regression of real activities manipulation are listed as well. I used the level of abnormal cash flows from operations (AB_CFO), level of abnormal

(39)

39

production costs (AB_PROD), the level of abnormal discretionary expenses (AB_EXP) and combined real activities manipulation (COMBINED_RAM) as proxies for real activities manipulation. There will be negative AB_CFO, AB_EXP, COMBINED_RAM and positive AB_PROD if the real activities are being manipulated. In this sense, I predicted that the FEMALECEO will be positively correlated with AB_CFO, AB_EXP, COMBINED_RAM and negatively correlated with AB_PROD.

The coefficient of FEMALECEO to AB_PROD is -0.034, which is the same as my prediction of negative correlation. The p-value is more than 0.05, so the result is not significant. Other variables, including CPMBINED_RAM, SIZE, RD_INT, AB_CFO and AB_EXP, are significant in influencing AB_PROD. In the AB_CFO regression, the coefficient is -0.045 of FEMALECEO, which is beyond expectation. The p-value is more than 0.05 so FEMALECEO is not significant in affecting AB_CFO. There are other variables significant for AB_CFO, like SIZE, ADJ_ROA, AD_PROD and AB_EXP.

As for AB_EXP, the coefficient of FEMALECEO is -0.045, which is also different from what I expected. The p-value is bigger than 0.05, which means that there is no significance. Except for LEV, other variables are all significant for AB_EXP. The FEMALECEO coefficient in the COMBINED_RAM regression is 0.056, with p-value more than 0.05, so it is not significant. Except for FEMALECEO and LEV, other variables are significant in influencing COMBINED_RAM.

So, after conducting the regressions for each proxy and analyzing the results, it is easy to conclude that there is no significant relation between CEO gender and real activities manipulation, let alone the expectation that with female CEO on board there will be less real activities manipulation. In this sense, the first hypothesis can be rejected since there is no significant correlation between gender of executives and accrual quality and real activities manipulation.

(40)

40

female executive samples could be the reason contributing to the insignificance between CEO gender and real activities manipulation, with 269 compared to 746 of male executives samples. However, another reason for the insignificance can be that CEO is not the person to engage in real activities manipulation directly. The manipulation can be conducted by other officers who are closer to the day to day operations while CEOs are more involved in making long-term decisions.

4.4 Different CEO gender impact on accrual quality and real activities manipulation

I hypothesized that a female CEO would affect accrual-based earnings management to a greater extent than real activities manipulation, since accrual-based earnings management are regarded as more unethically than real activities management due to the research of Merchant and Rockness (1994). According to Merchant and Rockness (1994), earnings management practice have impact on ethical judgments, and the research shows that real activities manipulation is less unethical, so in this sense, I predicted that female CEO would have more influence on accrual-based earnings management than real activities manipulation. In my research, I hypothesized that the FEMALECEO correlation would be bigger in discretionary accrual regression than in real activities manipulation regression.

The results show that the correlation in discretionary accrual regression is actually smaller than real activities manipulation, but most correlations are not significant, so in this sense I am not able to measure the differences in an objective way. Therefore, I cannot accept or reject the second hypothesis due to the insignificant results.

(41)

41

5. Conclusion and Discussion

Almost every year there will be large public-well-known companies being reported as committing accounting fraud. The numerous accounting scandals have caught the public eyes and crashed public trust at the same time. CEOs are usually the ones that hide behind the screen and try to manipulate the numbers in order to reach certain goals, so that the financial information provided by firms are not transparent and users will be misled in a way that the CEOs want them to be. CEOs play a big part in accounting fraud and the question of whether the gender of CEO plays a part in accounting fraud needs to be answered.

Accrual quality and real activities manipulation together determines earnings quality, and if quality of earnings is high, then less accounting fraud will exist. Previous research shows that earnings management will impair earnings quality. Accruals capture the extent to which current estimates map into the future cash flow so as to have better prediction for future firm performance. Accounting earnings and cash flows can be used as the measurement for company performance. Nevertheless, since accruals acquire estimations, estimations can be wrong, either by intent or not. So this gives the room for earnings management: executives may manipulate earnings to achieve what they want. Real activities manipulation is detached from ordinary operating activities which can be used by executives to achieve certain result, either for their own benefit or for reaching financial goals. Executives alter the activities from operations in an attempt to increase short-term profit, unlike manipulating accruals to change future cash flow. CEOs have stronger intention to be involved in earnings management when their compensation or bonus is linked with stock price or option holdings. They have opportunity to engage in accrual-based earnings management or real activities manipulation, since they are at the top level of the company and have power and more access. Ethics play a big part in affecting earnings quality and more

Is there a link between CEO gender and accounting fraud s there a link between CEO gender and accounting frauds there a link between CEO gender and accounting fraud s there a link between CEO gender and accounting fraud

Amsterdam Business School