Earnings management in family firms

(1)

1 MSc Accountancy and Control, variant Accountancy

Faculty of Economics and Business, University of Amsterdam

Earnings Management in Family Firms

Master Thesis

Supervisor: Dr. Bo Qin

Co-assessor: Dr. A. Sikalidis

Version: Final Version

Name: Frank C.M. Kwakman

Student Number: 10001537

Program: MSc Accountancy

(2)

2

Earnings Management in Family Firms

ABSTRACT: This study examines the effect of a founding-family ownership structure,

hereafter family firms, on levels of earnings management for continental European firms. According to agency theory, family firms face less type 1 agency problems between management and shareholders, known as the alignment effect, but more type 2 agency problems between controlling shareholders and non-controlling shareholders, known as the entrenchment effect. The decreased type 1 and increased type 2 agency problems have effects on managerial behavior. According to the alignment effect there is no need to base management compensation on reported earnings. Therefore earnings management would be less likely in family firms. However, according to the entrenchment effect, families will extract rents from the company and hide this by managing earnings. The level of private benefits of control (rent extraction), and thereby earnings management, depends on the strength of minority protection rights. Because of the legal and institutional differences with prior literature samples, the results of prior literature are therefore unlikely to hold for continental European firms. Based on agency theory, I predict that family firms use different levels of earnings management than non-family firms. Consistent with the prediction made, the results show that family firms use less earnings management than non-family firms. In particular, I find that family firms use less accrual-based and less real activities-based earnings management. In general, the results are robust to different methods used to capture the extent of earnings management. These findings show how ownership structures and institutional characteristics influence financial reporting and managerial behavior.

Keywords: Family firm, accrual-based earnings management, real activities-based

earnings management

(3)

3

A report by the European Commission in 2009 (EC, 2009) stresses the importance of family firms for our current economy. Firms with a founding-family ownership structure, hereafter family firms, are defined as firms in which a large block of shares is owned by the family and in which family members hold at least one key management position. They make up more than 60% of all European companies, and account for 40 to 50% of employment. Besides its importance the EC also addresses the need for further research about family owned firms. In this paper I respond to the call by the EC for further research about family firms. In particular, I will focus on differences in earnings management between family firms and non-family firms in continental Europe. This specific research area has been advocated by Bhaumik and Gregoriou (2010) and Ali et al. (2007).

Based on agency theory and earnings management motivations, the specific ownership structure of family firms might have consequences for the levels of earnings management. Due to a realignment of interests between shareholders and management and increased quality of monitoring by the controlling family, there are less conflicts of interest between managers and shareholders in family firms than non-family firms (Anderson and Reed, 2003). This reduction in type 1 agency problems is known as the alignment effect of family firms (Wang, 2006). Chen (2006) argues that due to this realignment of interests between shareholders and management, there is no need to base management compensation in family firms upon earnings. If management compensation is not tied up with earnings, the bonus plan hypothesis of earnings management does not hold for family firms. In addition, earnings management is less likely in firms that have boards with sufficient expertise to effectively monitor management (Xie et al., 2003). Taken together, the reduction of type 1 agency problems in family firms should lead to less earnings management by family firms than non-family firms. On the other hand, family firms face an increase in type 2 agency problems, which is known as the entrenchment effect of family firms. These problems refer to conflicts of interest between controlling shareholders and non-controlling shareholders. In this situation the family represents the controlling shareholders and they will extract rents from the non-controlling shareholders. This occurs when voting rights exceed cash flow rights, which is a common used structure in family firms (La Porta et al., 2000). Leuz et al. (2003) argue that controlling shareholders will manage earnings to hide rent extraction behavior and rent extraction opportunities. So the increased type 2 agency problems should lead to more earnings

(5)

5

management by family firms than non-family firms. An important tool to mitigate type 2 agency problems are legal protection rights for minority shareholders. Strong minority protection rights decrease the private benefits of control (rent extraction opportunities), which in turn decrease the need to manage earnings (Ali et al., 2007). Overall, the net impact on the differences in earnings management between family and non-family firms depends on the relative decrease in type 1 agency problems and increase in type 2 agency problems.

In this study the focus will be on continental European firms, because this setting has not been studied before. Prior literature examined financial reporting quality and earnings management by family firms in the United States (Wang, 2006), Australia (Setia-Atmaja et al., 2009) and Asia (Fan and Wong, 2002). However, these studies are unlikely to hold for continental European countries because of the legal and constitutional differences (La porta et al., 1998). The legal and constitutional differences influence the private benefits of control to controlling shareholders and thereby the levels of earnings management between countries (Leuz et al., 2003). Therefore, the differences in earnings management between family firms and non-family firms might be different than the differences in earnings management between family firms and non-family firms in prior studies. I will examine both accrual-based earnings management, as real activities-based earnings management. Accrual-based earnings management will be examined based on the modified Jones model (Dechow et al., 1995). Real activities based earnings management will be examined based on the models as in Roychowdhury (2006), Zang (2012), and Chi et al. (2011). In an additional analysis, I will use 1-on-1 matched samples based on the closest ROA (Kothari et al., 2005) and the closest propensity scores. Matching based on performance (closest ROA) will control for the potential bias of extreme performance, whereas matching based on propensity scores may help to mitigate an endogeneity bias. This additional analysis should increase the reliability of the tests.

Consistent with prior studies in the United States (Jiraporn and Dadalt, 2009; Wang, 2006) and Italy (Prencipe et al., 2008), I find that family firms use less accrual-based earnings management than non-family firms. In line with these studies, this finding suggests that the benefits of providing high quality financial statements are higher than the private benefits of control for continental European firms. These findings are robust to a performance matched analysis based on Kothari et al. (2005). Furthermore, the results indicate that family firms use less real activities-based earnings management than non-family firms. Using abnormal cash flows from operations and one summary measure as a proxy, I find less real activities-based earnings management by family firms than non-family firms. In an additional analysis, using a

(6)

6

propensity score matched sample, all proxies indicate less real activities-based earnings management by family firms. This can be explained by the close link between firm value and family wealth. Since real activities-based earnings management actually destroys firm value, and thereby family wealth, families might be more reluctant to real activities-based earnings management within their firms.

This study contributes to prior literature for two reasons. First, it responds to the call for future research by Bhaumik and Gregoriou (2010), the European Commission (2009) and Ali et al. (2007). They address the necessity to study family firms in general, and in particular to examine the relationship between family firms and financial reporting quality in different countries. Whereas prior studies looked at financial reporting quality and earnings management by family firms in the United States, Australia and Asia, the relationship between earnings management and family firms has not been examined before in continental Europe. My sample of continental European firms provides an interesting setting, because of the legal and institutional differences between the US, Asia, and continental Europe. La Porta et al. (1998) states that continental European countries have insider economies, less developed stock markets and weak investor protection rights compared to the US. On the other hand, continental European firms have stronger investor protection rights than Asian countries. These factors seem to influence levels of earnings management worldwide (Leuz et al., 2003). Despite continental European countries on average have weaker investor protection rights than the US, family firms still use less accrual-based earnings management than non-family firms. This finding enhances our knowledge about how ownership structure and institutional factors influence financial reporting. Secondly, it is the first time to examine the relationship between real activities-based earnings management and ownership structure. Since accrual-based earnings management depends on the level of real activities-based earnings management realized (Zang, 2012), it is necessary to look at this form of earnings management as well. Besides its theoretical contribution, it has a societal contribution as well. These findings might be relevant to investors to gain a better understanding about the manipulation of earnings. Firms with a founding family ownership structure provide more reliable financial reports and use less real activities to manage earnings than non-family firms. This can be used in the valuation of these firms.

The remainder of this paper is structured as follows. In section 2, I will discuss prior literature and theory to develop the hypotheses. Section 3 describes the sample and methodology. Section 4 provides the results. Finally, section 5 concludes the paper.

(7)

7 2 Literature Review

2.1 Family firms

Over time and throughout literature there are many different definitions of a family firm. All definitions were guided by the common notion of ‘a firm in which a family has influence’. To overcome this problem the European Commission (EC, 2009, p. 10) established a well-defined definition for family firms in the EU. “A firm is a family business if: (1) the majority of decision-making rights is in the possession of the family which established the firm, (2) the majority of decision-making rights are indirect or direct and (3) at least one representative of the family is formally involved in the governance of the firm.” Publicly traded firms meet the definition of family business if the founding family possesses 25% of the decision rights.

Studies regarding the importance of family firms started a decade ago by Anderson and Reeb (2003) in the United States. Prior to this study family firms were perceived to be less efficient and less profitable than non-family firms due to their specific ownership structure. However, Anderson and Reeb (2003) pointed out that one third of the S&P 500 consisted out of family firms and these family firms even outperformed their non-family counterparts. Also in Europe family firms are crucial for our current economy. Studies show family firms make up 60% of all European companies, and account for 40 to 50% of employment (EC, 2009). Despite the importance of family firms, there is relatively few academic research about family firms. The European Commission stresses the importance to conduct more studies about family firms (EC, 2009).

The main difference between family firms and non-family firms is the ownership structure. In publicly listed non-family firms there is a clear separation between management and owners. This separation is less clear in family firms. In family firms the family holds a significant amount of shares and key positions within management. A common structure to retain control in family firms is the use of voting rights. This is an effective way to control an entity without contributing to the majority of capital. Under this situation, shares sold to outside investors are bundled with reduced voting rights (DeAngelo and DeAngelo, 1985), so that the family remains control over the firm. Another common structure to retain control of a firm is the use of pyramids. For example, a family owns 51% of the shares of company A, and company A owns 51% of the shares of company B, then the family controls company B even

(8)

8

though it only owns about 25% of company B. This structure is widely used in Asia and Europe, but is less common in the US and UK (La Porta et al., 1999).1 According to Anderson and Reeb (2003) these ownership structures of family firms can have severe consequences for firm performance and managerial behavior, either positive or negative. First of all, they state that families have incentives and opportunities to benefit themselves at the expense of firm performance. This is the case when voting rights exceed cash flow rights, which is a key consequence of the previous discussed structures in family firms (La Porta et al., 1999). Under these circumstances it is more profitable to the family to extract rents and enjoy the full benefits of it, than to pay out dividends and share it with the minority shareholders (Bhaumik and Gregoriou, 2010). This could be done by excessive compensation, related-party transactions or special dividends. Secondly, Singell (1997) finds that family firms are more likely to place family members in key management positions. These family members might be less capable and talented than outside managers. A final negative point is made by Barclay and Holderness (1989), who suggest that large ownership stakes reduce the likelihood of bidding by other agents, reducing the value of the firm. This refers to the case when management performs poorly. Normally, in a market for takeovers, poor performing managers will be threatened with a takeover and replacement. If there is no market for takeovers, there will not be a threat for takeovers and management continues performing poor (Bhaumik and Gregoriou, 2010). On the other hand, founding family ownership might be positively related to firm performance. First of all, family wealth is closely linked to firm value. According to Anderson and Reeb (2003), this creates a reputation effect. Those family members employed by the firm will be motivated to increase firm performance and thereby upholding their family name. Secondly, Stein (1988) shows that shareholders with long investment horizons, like founding families, suffer less managerial myopia and are thus more likely to forgo investments that boost current earnings and focus on long-term value instead.

In this paper I will examine levels of earnings management by family firms, rather than family firm performance. In the next section, the concept of earnings management and its different forms and motivations will be discussed. Thereafter, agency theory will be discussed and used to predict levels of earnings management in family firms.

1_{The use of pyramid structures is less common in the US and UK, because regulation requires that owners of}

(9)

9 2.2 Earnings management

Walker (2013, p. 446) defines earnings management as: “The use of managerial discretion over (within GAAP) accounting choices, earnings reporting choices, and real economic decisions to influence how underlying economic events are reflected in one or more measures of earnings.” Important in earnings management is the manager’s aim to achieve a specific reported objective, either through the use of accounting choices or real economic decisions. In this section I will make a distinction between earnings management based upon accounting choices, also known as accrual-based earnings management, and earnings management based upon real economic decisions, also known as real activities-based earnings management. Furthermore, I will discuss the objectives which management is trying to achieve based on three motives. This could be either contractual, capital market or political motivations.

2.2.1 Accrual-based earnings management

Accrual-based earnings management is the use of different accounting policies and choices in order to influence the reported earnings. These accruals are equal to net income minus cash flows from operations. So, net income consists out of accruals and cash flows from operations. Accruals are used to overcome the timing and matching problems of cash flows when a firm is in constant operation (Dechow, 1994). So accruals should lead to earnings better reflecting firm performance. However, as stated by Dechow (1994), managers have discretion over the use and the recognition of accruals. This discretion could be used negatively, to opportunistically manage earnings. When management uses discretion to opportunistically manage earnings, earnings become less reliable. 2_{Therefore, accruals are seen as a trade-off}

between relevance and reliability.

There are many ways to use discretionary accruals opportunistically in order to manage earnings. For instance by under- or overestimates for provisions, warranty costs, inventory values and the timing and amounts of low-persistence items such as write-offs. Regardless of the form of accrual-based earnings management, there is an “iron law” surrounding the use of accruals (Scott, 2011, p. 423). This is that accruals reverse over time. It implies that a firm which manages earnings upwards to an amount greater than can be sustained, will experience

2_{Note that earnings management doesn’t necessarily need to be bad. Managers might use accrual-based earnings}

management as a vehicle for the communication of management’s inside information to investors. Using accruals to “unblock” inside information increases the usefulness of the financial statements (Scott, 2009, p. 422).

(10)

10

that the reversal of these accruals in future periods will force future earnings downwards. This requires managers to conduct even more earnings management if the reporting of losses is to be further postponed (Scott, 2011, p. 424).

2.2.2 Real activities-based earnings management

Until 2005 the main focus of earnings management studies was about accrual-based earnings management. However, a qualitative study by Graham et al. (2005) showed that most earnings management is committed through real actions instead of accrual manipulations. In their study managers admitted they would take real actions to meet short-term benchmarks. Roychowdhury (2006, p. 337) continued to study real activities-based earnings management and defined it 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.” He further states that, opposed to accrual-based manipulation, real activities manipulation does actually reduce firm value. Current actions taken to boost earnings might have a negative impact on future cash flows. So accrual-based earnings management within GAAP is likely to do less harm to the value of the firm than real based earnings management. However, managers still prefer to use real activities-based earnings management (Graham et al., 2005). Roychowdhury (2006) provides a possible explanation by stating that accrual manipulation is more likely to attract auditor attention than real decisions. Therefore, the costs of accrual-based earnings management are higher to the manager than the costs of real activities-based earnings management. This is consistent with Zang (2012), who finds that managers trade-off the two kinds of earnings management based on their relative costs. Further, the level of accrual-based earnings management depends on the level of real activities-based earnings management realized.

According to Roychowdhury (2006) real activities-based earnings management is used by firms to avoid losses. However, there is a variety in levels of real-based earnings management by firms. He finds that firms with sophisticated investors are less likely to conduct this form of earnings management. The presence of debt, stock of inventory and receivables, and growth opportunities are positively related to the use of real-based earnings management. Furthermore, firms trying to avoid negative annual forecast errors are more likely to conduct real-based earnings management.

(11)

11

Real activities-based earnings management, as in Roychowdhury (2006, p. 340), focusses on three manipulation methods. First of all, sales manipulation. This refers to managers’ attempts to temporarily increase sales during the year by providing price discounts or more lenient credit terms. Sales manipulation will lead to lower current-period cash flow from operations (CFO) and higher production costs than what is normal given the sales level. Secondly, discretionary expenditures. These discretionary expenses refer to R&D, advertising, and selling, general and administrative (SG&A) expenses. Lowering these expenditures lowers cash outflows in the current period, possibly at the risk of lower cash inflows in future periods. Reduction of discretionary expenses leads to abnormally low discretionary expenses and to higher abnormal CFO. Finally, overproduction. With higher production levels, fixed overhead costs are spread out over a larger number of goods, lowering fixed costs per unit and thereby lowering cost of goods sold (COGS). If COGS are lower, management reports better operating margins. Overproduction will lead to abnormally high production costs and has a negative effect on abnormal CFO. The different manipulation methods and their effect on the variables, as in Roychowdhury (2006), are summarized in table 1.

TABLE 1

Real activities-based earnings management manipulation methods and effect onvariables

Note: All manipulation methods and their effects are based on Roychowdhury (2006, p. 340).

It has to be said though, that the effect of the three manipulation methods on abnormal cash flow from operations is somewhat ambiguous (Zang, 2012). Sales manipulation and overproduction lead to lower CFO, whereas discretionary expenses lead to higher CFO. This means that the three manipulation methods could balance each other out. However, since two of the three manipulation methods are negative, it still can be assumed that the overall effect on CFO should be negative as well.

Manipulation method used Influence on variables

Sales manipulation lower CFO

higher production costs

Discretionary expenses higher CFO

lower discretionary expenses

Overproduction lower CFO

(12)

12 2.2.3 Earnings management motivations

According to Walker (2013) there are three types of motives for firms to engage in earnings management. These motives are based on positive accounting theory, which “is concerned with predicting such actions as the choices of accounting policies by firm managers and how managers will respond to proposed new accounting standards” (Scott, 2011, p. 304). So this theory should provide guidance in which situations management would be more likely to manage earnings. The motives consist out of contractual motivations, motivations to influence the expectations about future cash flows of the firm and motivations to influence the set of information used by third parties to evaluate the strength of the firm.

The first motive for firms to engage in earnings management is to achieve contractual targets which are related to reported earnings (Walker, 2013). Compensation contracts and debt contracts are types of contracts which are often tied up with reported earnings. The debt-covenant hypothesis and bonus plan hypothesis, as in Scott (2011), are part of Walker’s (2013) contractual motivations. First of all, the debt-covenant hypothesis predicts that managers will manage earnings upwards when the firm is close to violating debt covenants based on earnings. Most debt contracts contain covenants which the firm must comply with. If these covenants are not met, the debtor may impose constraints on management actions. A manager will not like this and therefore he might adopt accounting policies to raise current earnings, which reduce the likelihood of violating covenants (Scott, 2011). Watts and Zimmerman (1990) find that firms with high levels of debt indeed manage their earnings upwards. The second contractual motivation is the bonus plan hypothesis. The bonus plan hypothesis predicts that management of firms with bonus plans are more likely to choose accounting policies that increase current earnings (Scott, 2011). Healy (1985) finds that US managers manage earnings to maximize their bonus contracts. A bonus scheme consists out of a minimum target (bogey) in order to meet the bonus. Bonuses would increase with performance, but only up to a certain cap. According to Healy (1985) this provides incentives to manage earnings. If earnings are just below the bogey, or between the bogey and the cap, earnings will be managed upwards in order to maximize the bonus. If earnings are above the cap, or out of reach for the bogey, earnings will be managed downwards in order to increase chances of receiving a bonus in the next year. One way to increase net income is to choose accounting policies that shift earnings from future periods to the current period. Another way to boost earnings is to manage earnings based upon real activities (Roychowdhury, 2006).

(13)

13

According to the capital market motive, managers will influence the information set used by external investors which is used to form expectations about future cash flows and/or perceptions of firm risk (Walker, 2013). In other words, managers will manage earnings in order to influence the share price of the firm. Ali et al. (2007) state that family firms have larger analyst following than non-family firms, because of their higher earnings informativeness. Another study by Yu (2008) points out that firms with larger analyst following use less earnings management. He states that the analysts serve as additional external monitors to managers, instead of putting extra pressure on them to manage earnings. In other words, larger analyst following will lead to higher quality financial reports. Because of the larger analyst following in family firms (Ali et al., 2007) it might be expected that family firms use less earnings management.

A final motive for earnings management refers to influencing the set of information used by third parties to evaluate the strength of the firm (Walker, 2013). This can be related to the political cost hypothesis as in Scott (2011). It predicts that the greater the political costs faced by a firm, the more likely earnings will be managed downwards and deferred to future periods (Scott, 2011). For example, very large firms or highly profitable firms attract attention, which might be translated into new taxes or other regulation by politicians. If a manager is able to postpone current earnings to future periods it will attract less attention. It is unlikely the effects of the political costs hypothesis will differ between family firms and non-family firms. Therefore, the political cost hypothesis will not influence differences in the levels of earnings management between family and non-family firms.

2.3 Agency Theory

As previously explained, the main difference between family firms and non-family firms is the ownership structure. In non-family firms there is a clear distinction between ownership and management, whereas under family firms this distinction is less clear and in some situations even integrated. In this section I will discuss this distinction more thoroughly based on agency theory. Further, I will examine how managerial behavior might be affected and relate these effects to managerial behavior in family firms and the earnings management motivations.

(14)

14

Agency theory is based on an agency relationship, which is defined as a contract under which one or more persons (principal) engage another person (agent) to take actions on their behalf (Jensen and Meckling, 1976, p. 4). If both the principal and the agent are utility maximizers, it might be assumed that the agent will not always take actions in the best interests of the principal. In other words, both the principal and the agent want to maximize their own value, resulting in conflicts of interests. According to Jensen and Meckling (1976), the separation of ownership and control in organizations will lead to a typical agency relationship between stakeholders (principal) and management (agent). Fama and Jensen (1983) state that the control of agency problems becomes necessary when management does not bear significant wealth effects of their decisions, which is exactly the case in separation of ownership and control. The controls are necessary to limit the conflicts of interest between both parties and will result in monitoring costs, bonding costs and residual loss (Jensen and Meckling, 1976). The principal tries to limit the divergence from his interest by monitoring the agent’s actions or establishing appropriate incentives. Monitoring can refer to directly control agent’s actions, but also indirect through auditing or restricting management budgets. The agent will be rewarded if he acts in the best interests of the principal. Managerial remuneration is a common used incentive to realign interests between principal and agent. The agent might incur bonding costs to ensure he will not take actions which would harm the principal. Finally, there will always be a residual loss because it is inefficient to control everything.

Despite agency relationships and therefore agency theory might refer to any situation, the main focus of this study is about the conflicts of interest between (a) shareholders and management and between (b) controlling shareholders and non-controlling shareholders (Ali et al., 2007). These two types of conflicts are known as type 1 and type 2 agency problems and differ significantly between family firms and non-family firms. The following two subsections will go deeper into both types of agency problems and will be linked to the alignment effect and the entrenchment effect in family firms.

2.3.1 Type 1 agency problems and alignment effect

Agency theory by Jensen and Meckling (1976) and Fama and Jensen (1983) hold for conflicts of interests between all kind of stakeholders and management. Ali et al. (2007) are more specific and made a distinction between two types of agency problems that occur in publicly traded firms. The first type of agency problems, type 1 agency problems, refer to the conflicts of interest between shareholders and management and are also known as managerial

(15)

15

agency costs. In this situation management wants to maximize personal wealth, while shareholders want to maximize firm value. Management might maximize its own wealth by not working hard, by making investment decisions which reduce firm value but increase personal wealth, also known as pet projects, or by extracting rents from the firm. Jensen and Meckling (1976) stated it might be necessary to provide incentive contracts in such a way that management’s interests are in line with shareholders’ interests. Normally, shareholders’ interests are to increase stock price. To realign interests, management’s compensation could be tied up with net income. In order to maximize compensation and thereby their own wealth, management will maximize net income, which in turn will lead to higher stock prices. Now management’s interests are realigned with shareholder’s interests. It has to be said though, that incentive contracts have to be made up very carefully in order to realign interests (Indjedjikian, 1999). It will not work out if management has the incentive to take certain actions that do not create firm value for shareholders. Furthermore, the bonus plan hypothesis predicts that incentive contracts might lead to earnings management. In order to meet the target and therefore the incentive, managers might opportunistically managing earnings (Scott, 2011).

Based on the definition of a family firm there are two roles of the founding family. First, the founding family acts as management and takes decisions on behalf of their selves and on behalf of other shareholders. Wang (2006) states that the interests of the family and other shareholders are well aligned, because of their long-term orientation and reputation protection. He further argues that, because of the better alignment, there are less type 1 agency problems in family firms. This is defined as the alignment effect of family firms. In order to be classified as a family firm, at least one key position should be represented by the family (EC, 2009). So at least one key management member will act in the best interests of the family and therefore the other shareholders. Furthermore, Singell (1997) finds that family firms are more likely to place family members in key management positions, so again this would mean less misalignment of interests between management (family) and other shareholders. Chen (2006) argues that because of better alignment between the interests of shareholders and management it is no longer necessary to realign interests based upon incentive contracts as discussed by Jensen and Meckling (1976). Therefore, it is less likely that remuneration of management in family firms depends on accounting numbers like net income. Recall from the bonus-plan hypothesis that management of firms with bonus plans are more likely to use upwards earnings management. If family firms do not use bonus plans, the bonus plan hypothesis of earnings management no longer holds for family firms. Therefore, the decrease in type 1 agency

(16)

16

problems in family firms should lead to less earnings management by family firms than non-family firms, which still use bonus plans.

On the other hand, it might also be the case that the founding family has to monitor management. In this situation the founding family acts as the majority of shareholders. Ali et al. (2007) also argue that there are less type 1 agency problems in family firms. Or in other words, there are less conflicts of interest between management and shareholders (family). They state that due to several characteristics of family firms, the family has a strong incentive to monitor management. First of all, a family has an undiversified and concentrated equity position in the firm, so they have strong incentives to monitor operations in order to be sure their concentrated equity holding is well managed (Demzets and Lehn, 1985). Secondly, a family has good knowledge about the firm’s activities and products, which provides them the opportunity to monitor management even better (Anderson and Reeb, 2003). Finally, families have longer investment horizons than other shareholders, so families help mitigate myopic investment decisions by management (James, 1999). Due to these characteristics of the founding family, monitoring of management is likely to be of higher quality in family firms than in non-family firms. Because of the better monitoring in family firms, there are less severe hidden information and hidden action agency problems in family firms compared to non-family firms (Ali et al., 2007). Prior literature finds that earnings management is less likely in firms that have boards with sufficient expertise to effectively monitor management (Xie et al., 2003). Since families also have sufficient expertise to monitor management, earnings management might be less likely in family firms.

2.3.2 Type 2 agency problems and entrenchment effect

Whereas type 1 agency problems refer to the conflicts of interest between shareholders and management, type 2 agency problems refer to the conflicts of interest between controlling and non-controlling shareholders (Ali et al., 2007). Controlling shareholders are the shareholders which own the majority of voting rights and make the decisions within the entity. In many situations, these controlling shareholders also manage the firm (Gopalan and Jayaraman, 2012). Under these circumstances the conflicts of interest between controlling and non-controlling shareholders are even worse. Controlling shareholders enjoy substantial control due to their concentrated equity holding in the firm, their domination in board of directors and voting rights exceeding cash flow rights.

(17)

17

According to Wang (2006) controlling shareholders might expropriate wealth from the remaining non-controlling shareholders to maximize their own wealth. In other words, the controlling shareholders extract value from the firm to benefit themselves. When the value of the firm decreases, this also affects the wealth of the non-controlling shareholders. So, indirectly controlling shareholders extract wealth from the non-controlling shareholders. Rent expropriation opportunities for controlling shareholders lead to private benefits of control. Private benefits of control are most prevalent in situations where there is a divergence between voting rights, or control rights, and cash flow rights (Bhaumik and Gregoriou, 2010). For instance the Wallenberg family, which has only 1% of the cash flow rights but over 40% of the voting rights of Ericsson. According to Bhaumik and Gregoriou (2010), this reduces the incentive to pay out dividends to shareholders, as it only receives 1% of these dividends. Instead, it creates incentives to use the free cash flows in ways it maximizes its own private benefits. Another interesting case study refers to Hohner AG, a firm owned by the Hohner family. The firm spent DM 11.6 million on social donations, whereas the shareholders only received DM 7.2 million in dividends (Ehrhardt and Nowark, 2001). This clearly indicates that free cash flows are used to maximize the private benefits of the family, which gained social status after the donations.

The private benefits of control can be divided into two groups: monetary and non-monetary. Non-monetary benefits of control refer the social status that comes with ownership. However, non-monetary benefits of control are unlikely to affect firm value (Mueler, 2003). Therefore, the focus will be on monetary benefits of control. These benefits can be obtained by engaging in related-party transactions, paying excessive remuneration or special dividends (Anderson and Reed, 2003). Related party transactions are known as tunneling, and is a common way to extract rents from a firm. This is described as an occurrence, whereby the controlling shareholders uses multiple ways to transfer a significant proportion of the free cash flows from a company in which they have small cash flow rights but large voting rights, to a company in which they have large cash flow rights and control (Johnson et al., 2000). If the free cash flows are transferred to the other company, the controlling shareholders can pay out dividends and are able to actually receive a significant amount of it due to their large cash flow rights. Note that the use of tunneling requires control and large cash flow rights over a second company. Tunneling can take the form of expropriation of cash flows, assets or equity. Cash flow tunneling includes for example the purchasing of firm’s inputs above the market price from a firm in which the controlling shareholders have significant cash flow rights. Asset

(18)

18

tunneling involves the transfer of an asset to a company fully owned by the controlling shareholders. One of the largest tunneling frauds refers to the Tanzi family, which controlled the Pamalat group. They tunneled out over 3 billion to companies that were directly owned by the family (Enriques and Volpin, 2007).

Despite the private benefits to the controlling shareholders, these actions will have severe negative effects on firm performance and the reported accounting numbers. Cash flow tunneling has consequences for the current earnings statement, whereas asset tunneling affects a firm’s long-term ability to generate cash flows (Bhaumik and Gregoriou, 2010). It is therefore likely that non-controlling shareholders will find out rent extraction, which in turn will lead to penalties against the controlling shareholders. Leuz et al. (2003) state that controlling shareholders will manage earnings to hide true performance and to conceal their private control benefits from outsiders. They argue that managers will use accounting choices to manage earnings upwards in order to turn a loss, for instance due to overpayment for inputs, into a small profit. This reduces the likelihood of getting caught by the non-controlling shareholders and chance of being penalized is therefore low. In an ideal world, assuming rational investors, managing earnings in order to mislead the non-controlling shareholders would not work out. This is because the investors would see through the manipulated earnings (Bhaumik and Gregoriou, 2010). However, in the real world only sophisticated investors are able to see through earnings management and price earnings management risk, whereas ordinary investors are not able to see through it (Walker, 2013, p. 474). For example, Cheung et al. (2006) finds that, in Hong Kong, outside investors (i.e. the market) do not price earnings management risk. These examples suggests that investors are not entirely rational, and controlling shareholders can mislead the non-controlling shareholders through managing earnings.

As described earlier, a common structure in family firms is the use of voting rights. According to Ali et al. (2007), this specific ownership structure of family firms causes an increase in type 2 agency problems. In case of a family firm, the family represent the controlling shareholders, while the other shareholders represent the non-controlling shareholders. As described above, type 2 agency problems might lead to rent extraction by the controlling shareholders. Because of the increased type 2 agency problems in family firms, it might be predicted that the founding family will seek their private benefits and extract rents from the non-controlling shareholders. Wang (2006) refers to this situation as the entrenchment effect of family firms. As a consequence of the entrenchment effect, earnings management is also more likely in family firms than in non-family firms (Ali et al., 2007).

(19)

19

Prior literature shows that traditional governance tools, like the take-over market and incentive compensation, to mitigate type 1 agency problems are not effective in mitigating type 2 agency problems (Setia-Atmaja et al., 2009). This is because the parties involved in type 2 agency problems are different than the parties involved in type 1 agency problems. Incentive compensation cannot realign interest between controlling and non-controlling shareholders, because incentive compensation only applies to management. Take-overs neither mitigate type 2 agency problems, because after a take-over there will be new controlling shareholders which might extract rents as well. To mitigate type 2 agency problems there are alternative governance tools and regulation. Setia-Atmaja et al. (2009) examined the role of dividend pay-outs, levels of debt, and board structure to mitigate the conflicts of interest between controlling and non-controlling shareholders. They found that family firms pay out higher levels of dividends and have higher levels of debt than non-family firms. This is successful in mitigating type 2 agency problems, because high dividend pay outs and interest and debt repayments reduce the firm’s free cash flows, which otherwise would be vulnerable for cash flow tunneling. Recall from the debt covenant hypothesis that high levels of debt lead to earnings management in order to prevent violation of debt covenants. In case of a family firm, high levels of debt might actually be used to mitigate type 2 agency problems and thereby reducing earnings management. Therefore, the net effect of debt on earnings management in family firms depends on which demand is stronger. Either to manage earnings in order to prevent debt covenant violation, or to reduce earnings management by reducing rent extraction opportunities.

Another important tool to mitigate type 2 agency problems are legal protection rights for minority shareholders (Ali et al., 2007). These protection rights are in place to minimize rent extraction by controlling shareholders. Legal systems protect investors by providing them the rights to discipline insiders (to replace managers), and by enforcing contracts designed to limit insiders’ private control benefits. Leuz et al. (2003) find that earnings management is increasing in private benefits of control and decreasing in minority protection rights. A study by La Porta et al. (2000) showed that the United States are one of the few countries where minority shareholders gain strong legal protection against controlling shareholders in the decision making process. On the other hand, continental European countries tend to have less legal protection rights for minority shareholders. This could mean that controlling shareholders in continental Europe are more likely to extract rents from non-controlling shareholders than in the United States. Based on Ali et al. (2007), this might indicate that controlling shareholders in continental Europe are also more likely to hide rent extraction by managing earnings.

(20)

20

To sum up, due to the specific characteristics of a family firm there is less misalignment between the family and shareholders. This results in a decrease of type 1 agency problems, known as the alignment effect. Because of the better alignment between family and shareholders, there is no need to base management compensation upon reported earnings and therefore earnings management to maximize bonuses is less likely (Chen, 2006). On the other hand, there are increased type 2 agency problems, which is known as the entrenchment effect. According to the entrenchment effect, the family will extract rents from the company and hide this rent extraction by managing earnings (Ali et al., 2007). The level of rent extraction and earnings management depends on the strength of the minority shareholders protection rights. In addition, debt and dividend pay-outs are internally used to mitigate the agency problems between the controlling and non-controlling shareholders (Setia-Atmaja, 2009). The overall effect of family firms on agency costs depends on the relative decrease in type 1 agency problems compared to the increase in type 2 agency problems. This is consistent with Gilson and Gordon (2003), who state that non-controlling shareholders prefer the presence of the increase in costs of rent extraction.

2.4 Family firms and earnings management

The relationship between financial reporting quality and controlling shareholders, and family firms in particular, has been examined before in the USA and several Asian countries. Warfield et al. (1995) studied if managerial ownership influences the accounting choices and informativeness of earnings. They found that managerial ownership is positively related to the informativeness of earnings and negatively related to the magnitude of accounting accrual adjustments. Recent studies looked at the effect of founding family firms in particular on financial reporting quality. All U.S. studies are consistent and show that family firms report higher quality earnings, as measured by greater earnings informativeness, lower abnormal accruals and less earnings restatements (Tong, 2008; Ali et al., 2007; Wang, 2006). In addition, Jiraporn and Dadalt (2009) find that family firms in the U.S. use less accrual-based earnings management. Prencipe et al. (2008) is the only study about public family firms and earnings management in Europe. Consistent with Jiraporn and Dadalt (2009), they also find that family firms use less accrual-based earnings management than non-family firms. However, this study has some limitations, as it focusses on just Italian firms and uses just one specific accrual, namely R&D cost capitalization. Therefore, the results of this study are not generizable to other

(21)

21

European firms and other ways of accrual-based earnings management. These studies seem to be inconsistent with the rent extraction hypothesis. According to the rent extraction hypothesis, family firms should manage their earnings in order to protect their private benefits.

However, there are also studies consistent with the rent extraction hypothesis. Fan and Wong (2002) replicated the Warfield et al. (1995) study for seven Asian countries. Contrary to the prior studies for U.S. firms, they found that corporate ownership leads to lower informativeness of earnings in Asia. They argue that the lower informativeness is due to self-interested reporting purposes by the owners, causing the reported earnings to lose credibility to outside investors. The concentrated ownership prevents leakage of information about management’s rent-seeking activities. This is consistent with the increased type 2 agency problems in family firms (Ali et al. 2007), in which the families manage earnings to mislead the investors in order to protect their rent extraction opportunities. The differences between both studies are explained by the legal and constitutional differences.

Leuz et al. (2003) show that institutional differences, like investor protection rights, influence levels of earnings management across countries over the world. Countries with weak investor protection rights face higher levels of earnings management compared to countries with strong investor protection rights. Based on prior literature it can be expected that family firms use more earnings management than non-family firms in countries with weak minority investor protection rights (Fan and Wong, 2002). Weak investor protection rights lead to high private benefits of control, providing the controlling shareholders the opportunity to extract rents from the firm. In order to protect their rent extraction opportunities they will manage earnings. On the other hand, in countries with strong investor protection rights, like the USA, family firms use less earnings management than non-family firms and actually produce higher quality financial statements (Jiraporn and Dadalt, 2009; Wang, 2006). This occurs when the benefits of providing high quality reports, reflected in a lower cost of capital, are higher than the private benefits of control (Gopalan and Jayaraman, 2012). According to Gopalan and Jayaraman (2012), this trade-off is influenced by the strength of a country’s legal protection rights. Thus, the amount of earnings management by family firms depends on how effectively rent extraction opportunities are reduced, which is based on the strength of a country’s legal protection rights. Despite some variations between European nations, overall continental European countries have weaker investor protection rights than the USA (La Porta et al., 1998). So based on the USA studies, it can be expected that family firms in continental Europe have higher private benefits of control and therefore earnings management is more likely by family

(22)

22

firms than by non-family firms. However, continental European countries have stronger investor protection rights than Asian countries (La Porta et al., 1998). So based on prior studies in Asia, it can be expected that family firms have less private benefits of control and therefore earnings management by family firms is less likely than earnings management by non-family firms. Taken together, it is unclear whether family firms in continental Europe use more or less accrual-based earnings management than non-family firms. Therefore, the first hypothesis is stated in a non-directional way:

Hypothesis 1: firms with a founding-family ownership structure have different levels of

accrual-based earnings management than non-family firms.

In addition to accrual based earnings management, I will examine real activities-based earnings management as well. Prior studies show that managers not only use accounting choices to manipulate earnings, but also use real economic activities (Roychowdhury, 2006). In addition, Zang (2012) finds that the level of accrual-based earnings management depends on the level of real activities-based earnings management realized. Therefore, focusing on just accrual-based earnings management would provide an incomplete image of the total level of earnings management within family firms. Since real activities-based earnings management is a relatively new stream of literature, it has not been examined before in relation to family firms. It could be expected that real activities-based earnings management is used to boost current earnings and cover up rent extraction. Under this expectation, the amount of real activities-based earnings management by family firms also depends on the strength of legal protection rights. However, there is no literature to rely upon this expectation. On the other hand, contrary to accrual-based earnings management, real activities-based earnings management actually destroys firm value (Roychowdhury, 2006). Because of the concentrated equity holdings by families in their firms, there is a close link between firm value and family wealth. Since real activities-based earnings management destroys firm value, and thereby family wealth, families might be reluctant to real activities-based earnings management within their firms. Therefore, real activities-based earnings management could be less likely in family firms than in non-family firms. Recall that real activities-based earnings management is more likely in firms with high levels of debt, stock of inventory and receivables, and growth opportunities. Furthermore, firms trying to avoid negative annual forecast errors are more likely to conduct real-based earnings management. Finally, real-activities based earnings management is less likely in firms with sophisticated investors (Roychowdhury, 2006). As previously mentioned, families could be seen as more sophisticated investors (Anderson and Reeb, 2003). Therefore real

(23)

activities-23

based earnings management would be less likely. In addition, family firms could rely more on debt than non-family firms, to mitigate the type 2 agency problems and thereby mitigating earnings management (Setia-Atmaja et al., 2009). However, according to Roychowdhury (2006), real activities-based earnings management is more likely in firms with high levels of debt. The other factors of real activities-based earnings management, level of inventory, receivables and growth opportunities, are unlikely to differ between family firms and non-family firms. Taken together, the net effect on real activities-based earnings management by family firms compared to non-family firms is unclear. Therefore, the second hypothesis is stated in a non-directional way as well:

Hypothesis 2: firms with a founding-family ownership structure have different levels of real

activities-based earnings management than non-family firms.

3. Research Methodology

In this section the research methodology will be discussed. First, a thorough description of the sample collection and deleted observations will be provided. Second, the empirical models used and all the control variables in it will be explained.

3.1 Sample description

This study examines the differences in earnings management between family firms and non-family firms. The family firm sample is based upon the top 100 best performing family firms in Europe (Campden FB, 2012). As in other studies (Jiraporn & Dadalt, 2009; Ali et al., 2007; Wang, 2006), I use a preconceived list of family firms, because it would be too time consuming to identify family firms manually based on financial statements. Family firms are defined based on the same definition as used by the European Commission (EC, 2009, p. 10)3 (Campden FB, 2012). The focus of this study will be on publicly listed firms. Therefore, all private family firms are removed from the sample. After removing the private family firms, 44 unique publicly listed family firms were left. Second, firms from the United Kingdom are

3_{Consistent with the definition of a family firm by the European Commission (EC, 2009, p. 10), a firm is included}

in the list if: “At least one representative of the family is formally involved in the governance of the firm; and listed companies meet the definition of family enterprise if the persons who established or acquired the firm from their families or descendants possess at least 25% or more of the decision-making rights mandated by their share capital” (Campden FB, 2012).

(24)

24

excluded from the sample, because firms in the U.K. have similar legal and institutional characteristics with the United States (La Porta et al., 1998). Therefore, studies based on firms in the U.K. are likely to have similar results compared to studies based on firms in the U.S. and Australia (Leuz et al., 2003). Excluding these firms provides a unique sample of continental European firms, which might provide different results. Third, as in Becker et al. (1998) firms from the financial and utilities industries are excluded. Financial institutions with Standard Industrial Classifications (SIC’s) between 6000 and 6999 are excluded, because computing discretionary accruals for these firms is problematic. Utility firms with SIC’s between 4900 and 4999 are excluded, because regulation is likely to influence the incentives to manage earnings differently compared to companies in unregulated industries. Finally, family firms with unavailable data to compute discretionary accruals, production costs and discretionary expenses are deleted. This resulted in a sample of 26 unique family firms.

The non-family firm sample is based on the Euro STOXX index. This index is used as non-family firm sample because it already excludes firms from the U.K. and financial industries. As in the family firm sample, firms in the utility industry and firms with insufficient data are deleted. Finally, some of the family firms are listed on the Euro STOXX index as well, so there were a number of duplicates in the sample. After deleting the family firms from the Euro STOXX index there were 130 unique non-family firms left. Taken together, there are 156 individual firms. In table 2, I tabulated the sample in firm-year observations, and made a distinction between family firms and non-family firms. In total, there are 1,045 firm-year observations, of which 172 are family firms and the other 873 are non-family firms. Despite the small differences between the years, approximately 20% of the sample consists out of family firms.

TABLE 2

Sample distribution in firm year observations

2006 2007 2008 2009 2010 2011 2012 Total

Family Firms 24 25 25 26 26 25 21 172

Non-Family Firms 120 125 129 126 127 129 117 873

Total Firms 144 150 154 152 153 154 138 1,045

(25)

25 3.2 Empirical models

In order to examine earnings management I will look at accrual-based earnings management and real activities-based earnings management. The empirical model used to examine accrual-based earnings management will be discussed in the first subsection, and the models used to examine real-activities based earnings management will be discussed in the second subsection.

3.2.1 Accrual-based earnings management

As stated before, managers try to influence reported earnings with different accounting policies and choices. This is known as accrual-based earnings management. There are many ways to detect the effects of accounting choices on earnings. McNichols and Wilson (1998) examined this on the estimation of an individual account, bad debt expenses. Sweeney (1994) examined the changes in specific accounting method choices. A recent study by Prencipe et al. (2008) examined accrual based earnings management based on R&D cost capitalization. However, consistent with Becker et al. (1998), I will focus on the net effect of all accounting choices that impact reported income, instead of specific individual accounting choices. Therefore, I will examine the behavior of total discretionary accruals. To measure the total level of discretionary accruals I use the modified Jones model as in Dechow et al. (1995). The only modification compared to the original Jones model is that the change in revenues is adjusted for the change in receivables. The modified model implicitly assumes that all changes in credit sales result from earnings management. This is based on the assumption that it is easier to manage earnings by exercising discretion over the recognition of revenue on credit sales, than it is over the recognition on cash sales (Dechow et al., 1995). I use the modified model because this modification generates the fewest type 2 errors. Further, I use the cross-sectional variation. This means that the modified Jones model is ran for each year and industry. According to Subramanyam (1996), the cross-sectional variations of the Jones models are better specified than the time series models.

In order to estimate discretionary accruals, normal accruals have to be computed first. The model estimates normal accruals as the change in revenue minus the change in receivables, and the level of property, plant, and equipment (PPE). These variables control for changes in accruals that are due to changes in the firm’s normal operating activities. The portion of total accruals unexplained by the firm’s economic condition are discretionary accruals. These

(26)

26

discretionary accruals are assumed to be due to accrual-based earnings management. This is done based on the following model:

𝑇𝑂𝑇𝐴𝐿 𝐴𝐶𝐶𝑅𝑈𝐴𝐿𝑆𝑖,𝑡

𝐴𝑆𝑆𝐸𝑇𝑆𝑡−1 = α

1

𝐴𝑆𝑆𝐸𝑇𝑆𝑡−1+ β1

(∆REVi,t− ∆RECi,t) 𝐴𝑆𝑆𝐸𝑇𝑆𝑡−1 + β2

𝑃𝑃𝐸𝑖,𝑡

𝐴𝑆𝑆𝐸𝑇𝑆𝑡−1+ εi,t

(1)

Where:

𝑇𝑂𝑇𝐴𝐿 𝐴𝐶𝐶𝑅𝑈𝐴𝐿𝑆𝑖,𝑡 = income before extraordinary items minus operating cash flows for firm i in

industry j for year t;

∆𝑅𝐸𝑉𝑖,𝑡 = change in revenues for firm i in industry j for year t;

∆𝑅𝐸𝐶𝑖,𝑡 = change in receivables for firm i in industry j for year t;

𝑃𝑃𝐸𝑖,𝑡 = property, plant, and equipment for firm i in industry j for year t;

𝜀𝑖,𝑡 = error term for firm i in industry j for year t.

All variables are deflated by lagged total assets (𝐴𝑆𝑆𝐸𝑇𝑆𝑡−1) to control for scale

differences. Financial data to compute total accruals is attained from COMPUSTAT global database. Industry membership is assessed using two-digit SIC codes. An ordinary least squares (OLS) regression is used to obtain the industry specific estimates of the coefficients. Finally, discretionary accruals (DA_industry) are defined as the error term in the above equation (1).

The second step to examine the differences in earnings management between family firms and non-family firms is a multivariate analysis. This analysis is used to control for potential differences between family firms and non-family firms. In this multivariate analysis, discretionary accruals per year and industry are regressed on a dummy variable indicating family firm or non-family firm and several control variables to control for the differences between family and non-family firms.

The multivariate analysis is performed using the following regression model:

𝐷𝐴_{𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦} = 𝜕₀+ 𝜕₁𝐹𝐴𝑀𝐼𝐿𝑌 𝐹𝐼𝑅𝑀 + 𝜕₂ 𝑆𝐼𝑍𝐸 + 𝜕₃𝑅𝑂𝐴 + 𝜕₄𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝜕₅𝐺𝑅𝑂𝑊𝑇𝐻 +

𝜕₆𝐿𝑂𝑆𝑆 + 𝑂𝐶𝐹 + 𝜀_𝑡 (2)

Where:

DA_industry = estimated discretionary accruals per industry and year; FAMILY FIRM = dummy variable equal to one if firm is family firm; SIZE = natural logarithm of total assets;

(27)

27

LEVERAGE = leverage: total liabilities divided by total assets; GROWTH = growth rate in sales;

LOSS = dummy variable equal to one if firm reported negative net income (=loss) OCF = operating cash flows scaled by lagged assets

With respect to the control variables, I follow prior literature based on Becker et al. (1998) and Wang (2006). The first control variable is SIZE, which controls for the potential effects of size on the choice of discretionary accruals. The second control variable, return on assets (ROA), controls for the potential effect of profitability on the choice of discretionary accruals. The third control variable is LEVERAGE. High leveraged firms may be close to violation of debt covenants, and the likelihood of debt covenant violation tends to be associated with discretionary accrual choice (Watts and Zimmerman, 1990). The fourth control variable is GROWTH, which controls for the effect of the growth potential of the firm on the choice of discretionary accruals. The fifth control variable is LOSS, a dummy variable indicating a negative net income. This controls for the effect of a loss on the choice of discretionary accruals. Finally, I control for the potential effect of operating cash flows from operations on the choice of accruals (OCF). Note that a positive coefficient indicates that discretionary accruals are higher, and this means that earnings are managed upwards. On the other hand, a negative coefficient indicates that discretionary accruals are lower.

3.2.2 Real activities-based earnings management

Real activities-based earnings management will be examined based on the three manipulation methods. As previously explained in the literature review, real activities-based earnings management might be conducted through sales manipulation, discretionary expenses and overproduction (Roychowdhury, 2006). If a firm would manage it earnings upwards through one of these three manipulation methods, this would lead to abnormally low CFO, abnormally low discretionary expenses and abnormally high production costs. I refer to table 1 in section 2.3.2 to see which specific effect the manipulation methods will have on abnormal CFO, abnormal discretionary expenses and abnormal production costs.

The first step to examine levels of real activities-based earnings management is to compute the normal levels of CFO, production costs and discretionary expenses. This is based on three specific formulas. Normal cash flows from operations is expressed as a linear function of sales and change in sales in the current period:

(28)

28 𝐶𝐹𝑂𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1

= α

0

+ α

1 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1

+ β

1 𝑆𝑎𝑙𝑒𝑠𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1

+ β

2 ∆Salest 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1

+ ε

t

(3)

Normal production costs consist out of normal costs of goods sold and normal change in inventory. This is expressed as a linear function of sales in the current period, change in sales in the current period and change in sales in the previous year:

𝑃𝑅𝑂𝐷𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1 = α0+ α1 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1+ β1 𝑆𝑎𝑙𝑒𝑠𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1+ β2 ∆Salest 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1+ β3 ∆Salest−1 𝐴𝑠𝑠𝑒𝑡𝑠𝑡−1+ εt (4)

Normal discretionary expenses is expressed as a function of lagged sales:

𝐷𝐼𝑆𝐸𝑋𝑃_𝑡 𝐴𝑠𝑠𝑒𝑡𝑠_𝑡−1

= α

0

+ α

1 1 𝐴𝑠𝑠𝑒𝑡𝑠_𝑡−1

+ β

𝑆𝑎𝑙𝑒𝑠_𝑡−1 𝐴𝑠𝑠𝑒𝑡𝑠_{𝑡− 1}

+ ε

t (5) Where:

CFO = Cash flows from operations;

PROD = Production costs: costs of goods sold + change in inventory; DISEXP = Discretionary expenses: selling, general & administrative expenses; Δ Sales = Change in sales in year t;

Δ Sales (t-1) = Change in sales in year t-1.

Again, all variables are deflated by lagged total assets (𝐴𝑠𝑠𝑒𝑡𝑠_𝑡−1) to control for scale differences. The regressions are ran for each year end industry, in which industry membership is assessed using two-digit SIC codes. An ordinary least squares (OLS) regression is used to obtain the industry specific estimates of the coefficients. Finally, the abnormal amounts of CFO (ABN_CFO), production costs (ABN_Prod) and expenses (ABN_DisExp) are defined as the error terms (𝜀_𝑡) in the above equations. Those abnormal amounts are separately used as proxies for real activities-based earnings management (RM_proxy) in the following multivariate analysis (6). In addition to the individual real activities-based earnings management proxies, I use two comprehensive measures to capture the total level of real activities-based earnings management. First, based on Chi et al. (2011), I use a summary measure by combining the three individual measures (RM_index). RM_index is computed by the sum of the standardized values of the individual measures, i.e. RM_index = ABN_Prod – ABN_CFO – ABN_DisExp.

Earnings management in family firms

Earnings Management in Family Firms

Master Thesis

Earnings Management in Family Firms

Table of Contents

= α

+ α

+ β

+ β

+ ε

(3)

= α

+ α

+ β

+ ε