Supervisor: Niuosha Samani
Master Degree Project No. 2016:22 Graduate School
Master Degree Project in Accounting
Earnings Quality in Family Firms
A quantitative study in Sweden
Mårten Andersen
i
Acknowledgements
I would like to thank all the people who have contributed to the accomplishment of this thesis. The process of creating this thesis has been both interesting and rewarding. I would like to thank my supervisor Niuosha Samani for the guidance she provided during the process. I would also like to thank the members of the seminar group, who all provided valuable insights and interesting discussions during the seminar meetings.
Thank you!
Gothenburg, May 20th 2016
Mårten Andersen
ii
Abstract
This study investigates the relation between family firms and earnings quality.
There are two competing theories on the effect that family firms have on earnings quality, the entrenchment effect with decreased earnings quality and the alignment effect with increased earnings quality. By using a sample of firms listed at the Stockholm Stock Exchange between the years 2005 and 2014. The study empirically examines the effect that the ownership structure has on the discretionary accruals as a proxy for earnings quality. The findings suggest that family firms are associated with lower levels of discretionary accruals. This would support that family firms in Sweden provide earnings of higher quality than their non-family counterparts, which is in line with the alignment theory.
Keywords: earnings quality, family firms, earnings management, dual class shares, accruals, entrenchment, alignment.
iii
Table of Contents
1. Introduction ... 1
2. Institutional Setting ... 4
2.1 Dual-Class Shares ... 4
2.2 Swedish Institutional Setting ... 4
3. Theoretical framework ... 6
3.1 Agency Theory ... 6
3.2 Earnings Quality ... 6
3.3 Earnings Management ... 7
3.4 Entrenchment Effect ... 8
3.5 Alignment Effect ... 8
3.6 Prior Studies ... 9
4. Method ... 13
4.1 Research Design ... 13
4.2 Sample and Data Collection ... 13
4.3 Variables ... 14
4.4 Models ... 15
5. Empirical Results ... 17
5.1 Descriptive Statistics... 17
5.2 Pearson Correlation ... 18
5.3 Results and Discussion ... 20
5.4 Sensitivity Analysis and Additional Tests ... 21
6. Summary and Conclusion ... 23
References: ... 24
Appendix ... 27
1
1. Introduction
This chapter will start with a description of the background of family ownership. The background is followed by the problem definition on the family ownership structure by presenting the theories of entrenchment and alignment. This leads to the purpose of the thesis, the hypothesis and presenting the results of the study. The chapter ends with a description of the study’s outline.
The family owned firm is a unique corporate governance type that can reduce the traditional agency issue identified by Jensen and Meckling (1976) by aligning the interest of owners and management (Demsetz & Lehn, 1985; Shleifer & Vishny, 1997). This is achieved by the historical presence of the family, their long term perspective and their poorly diversified portfolios (Anderson & Reeb, 2003).
While reducing the agency issues between owners and managers, the family owned ownership structure could introduce another agency issue, a conflict of interests between minority shareholders and controlling shareholders (Villalonga & Amit, 2006; Renders & Gaeremynck, 2012; Bardhan et al., 2015).
In Europe more than 60 % of all firms are considered family firms (Ec.europa.eu, 2016). Most of these firms are SME1, however some of the largest businesses in Europe are family owned for example the German Volkswagen, French Peugeot, Italian Exor and Swedish H&M (Familybusinessindex.com, 2016). This is not only the case in Europe but also in the US. In the American Standard & Poor fortunes 500 list (S&P 500) one third was defined as family owned (Wang, 2006). Previous research has used several definitions when defining what constitutes a family firm. Some commonly used definitions when classifying firms as a family firm or non-family firm in public firms is that the family controls at least between 10 – 25 % of the votes, that the family holds senior management positions, that several board members comes from the family or that the founder of the firm serves as the CEO.
(Ali et al., 2007; Achleitner et al., 2014; Chen et al., 2008; Prencipe et al., 2008; Wang, 2006). The guidelines for defining a family firm in this study, is that the majority shareholder is a family group active in one firm and control at least 15 % of the votes and / or that the founder of the firm also serves as CEO in the firm.
Studies have shown that the family manages control of the firm through the use of enhancing control mechanism e.g. pyramid ownership2, dual-class shares3 and cross holding ownership structures4 (Villalonga & Amit, 2010; Claessens et al., 2000). The sample of this study shows that in Sweden 52
% of the firms used dual-class shares and of the sample 35 % are classified as family firms with accordance to the guidelines of this study, this is also consistent with La Porta et al. (1999) study that investigated the ownership structures around the world and in Sweden and the study by Khosravi (2015) where the CEO compensation was investigated in firms listed at the Stockholm Stock Exchange. The use of dual-class shares diverges from the “one share one vote principle”. This allows the family owners to keep control of the firm without a proportional equity stake of the firm (Ehrhardt
& Nowak, 2003). A recent example is the Chinese company AliBaba where the IPO was moved from the Hong Kong Stock Exchange (HKSE) to the New York Stock Exchange (NYSE) since HKSE did not permit the use of dual-class shares (Chan, 2014; Jung et al., 2015).
In firms with a concentrated ownership there is a possibility that the information asymmetry between insiders and outsiders could create incentives for controlling shareholders to expropriate wealth from
1 Small and Medium Entities
2 Pyramids are control mechanism where the family control shares in the firm through entities such as trusts, funds, foundations, limited partnerships, holdings, or other form of corporation where the family is the sole owner (Villalonga & Amit, 2006).
3 Shares where there is a difference between the voting rights and cash-flow rights (ABL 2005:551).
4 Cross-holdings are a control mechanism where the firm owns shares in a corporation that belongs to the family's sphere of control (Villalonga & Amit, 2006).
2
the firm at the expense of minority shareholders (Akerlof, 1970; Wang, 2006). This is in line with the entrenchment effect where firms can manage earnings opportunistically (Morck et al., 1988). In this theory the family firm is seen to have inferior internal control mechanism due to the presence of family members in the management and the board. This allows for the family members to access information without it first being made public. With this kind of insider information which stems from their management positions in the firm and their positions as controlling shareholders, it would reduce the demand for high earnings quality information from shareholders. These issues are summarized in the entrenchment effect (Bardhan et al., 2015; Kang, 2014; Fan & Wong, 2002).
However, the entrenchment effect is not unknown to investors, the market is aware of the potential issues of ownership concentration. Therefore, the shares of a family firm would be traded with a discount, this discount can for example be seen in the Tobin’s Q5, when the family ownership increases past 31 % according to Anderson and Reeb (2003) and 40 - 50 % in an older study by McConnell and Servaes (1990) which finds that the firm value decreases (McConnell & Servaes, 1990; Anderson & Reeb, 2003). In order to reduce this discount, family firms have incentives to report earnings of high quality and in good faith. These incentives are gathered in the alignment effect.
Another consequence is that these kinds of firms have stronger monitoring from owners which further reduces the manager’s opportunities to manage earnings. For family firms the owner’s interest in the firm is long-term which reduces the incentives for short term manipulation. Further, scandals will have a negative effect on the family's reputation (Wang, 2006; Anderson & Reeb, 2003). This negative effect is achieved through medias “naming and shaming” when the firm behaves in a manner that is perceived to deviate from the desirable norm (Jansson, 2013).
During the last couple of decades, the focus on family firms as a performer has increased; there have been several studies that have identified the family firm as a superior performer compared to non- family firms (Andres, 2008; Miller & le-Breton Miller, 2006; Lee, 2006; Zellweger et al., 2007;
Villalonga & Amit, 2006). However, there are conflicting views (Wang, 2006; Kang, 2014; Bardhan et al., 2015) on how the earnings quality is affected due to the conflicting nature of the alignment and entrenchment effects.
This makes the Swedish setting ideal to investigate. In Sweden there is a large presence of family owned firms, which allows for either an entrenching or aligning behavior. Further Sweden has a long history of dual-class shares in listed firms, this wedge between equity rights and voting rights could induce opportunistic behavior (La Porta et al., 1999; Carlsson, 2007). The private ownership of shares is among the highest in the world. Finally, the Swedish institutional setting is interesting with long history of minority owner protection and a strong code law.
Finally, to conclude the family owned firm is an important ownership structure that is present and controlling a vast number of firms in both the Europe and the US. This is firms that often deviate from the “one share one vote” principle. This ownership structure has faced an increased level of interest during the last decades. The primary focus of these studies has been on the American market or on the emerging markets, the European setting has been largely overlooked and Sweden notably. This makes the opposing theories of alignment and entrenchment interesting, and also what the effects will be on the reported earnings. The question therefore arises, will the earnings be of a higher or lower quality in family firms compared to non-family firms in Sweden. As of such there is a gap in the literature that this study will help mitigate.
The purpose of this study is to investigate how the family firm ownership structure in Sweden affects the earnings quality compared to the earnings quality of non-family firms and investigate the effects of
5 Tobin’s Q is the market value of the company scaled by total assets. If the value is above 1, the market values the assets at more than their recorded value, this could be due to unrecorded assets such as intangible assets stemming from R&D. However, if the value is below 1 the market values the assets at less than book value, this could be due to dual-class shares, poor accounting quality etc. (Morck et al., 1988).
3
the alignment and entrenchment effects in the Swedish setting. Is there a difference between the earnings quality of family owned firms compared to non-family owned firms.
The results from this study find that family firms provide lower levels of discretionary accruals compared to their non-family counterparts. This would suggest that the reported earnings are of higher quality than in non-family firms. This is in support of the alignment theory. This is consistent with the studies of Wang (2006), Ali et al. (2007) and Alves (2012). Previous studies that supported the entrenchment theory were mainly conducted in emerging markets (Fan & Wong, 2002; Claessens et al., 2002). These results could be useful for investors when making decisions, to use ownership concentration as one decision factor. However, generalizing the results to other countries with similarities to the Swedish setting should be conducted with care due to the sensitivity of the results.
Further, the results could be helpful when constructing future research.
This study will contribute to the extant literature of family ownership and earnings quality by focusing on the unique Swedish institutional setting, a setting with a high ownership concentration and a high presence of family owned firms. Further, the combination of earnings quality and family firms are a relatively new subject. Much of the previous research has been conducted on the American market and focused on S&P 500 firms (Wang, 2006; Ali et al., 2007; Bardhan et al., 2015). In contrast there is a lack of research conducted in the European setting, where there is less ownership dispersion compared to the American market. Most research outside America on ownership concentration has been conducted on emerging markets, primary East Asia (Claessens et al., 2002; Fan & Wong, 2002) a setting with weak institutional regulation and inferior code law compared to Sweden.
This paper consists of six chapters. Chapter two describes the institutional setting in Sweden. Chapter three presents the theoretical framework. Chapter four explains the methodology. In chapter five the empirical findings are presented and analyzed. Finally, in the sixth chapter a summary and the conclusion are presented and the chapter ends with suggestion for future research.
4
2. Institutional Setting
In this chapter the concept of dual-class shares will be explained first and followed by an explanation of the Swedish institutional setting.
2.1 Dual-Class Shares
The dual-class share system means a divergence from the “one share one vote” principle. This creates a wedge between voting rights and cash-flow rights. In Sweden dual-class shares are regulated in the Swedish Companies Act6 (ABL 2005:551). ABL 4:2 describes that the use of shares with different voting power is permitted. Further it describes how and where it should be disclosed; it should be disclosed in their article of association. In the article of association, it should be described what letter represents each share and what voting power each share has. The voting difference is regulated in ABL 4:5 where the maximum divergence is ten to one. In Sweden most firms with dual-class shares use A and B class shares. Where the B class is publicly traded and the A class has the voting deviation.
There is however exceptions where both the A and B class shares are publicly traded (e.g. Volvo, Handelsbanken).
The dual-class share system is frequently used by family-controlled firms. This allows the family to maintain control without the sufficient equity share. This wedge between cash-flow rights and voting rights could induce agency issues, which will be explained in chapter 3.1.
2.2 Swedish Institutional Setting
When compared to the corporate governance structure of the Anglo-Saxon and also continental counterparts the Swedish corporate governance structure deviates in multiple ways. The private ownership of shares in Sweden are among the highest in the world, there are a long history of dual- class shares, the distribution of roles and responsibilities of the annual general meeting (AGM), boards of directors and executive management are clear cut in the Swedish Companies Act (Carlsson, 2007).
Figure 1:
Source: Hellman, 2011
The above figure 1 shows the connections and relationships between the AGM, the board of directors, the CEO and the auditor.
6 Aktiebolagslagen
5
The highest decision making body in a limited Swedish company are the AGM where the shareholders have an opportunity to exercise their influence. Any shareholder is entitled to ask questions at the meeting. This is also where the annual accounts are presented. The AGM has a sovereign role compared to both the board of directors and the CEO. This is also the event where the board of directors and CEO are discharged of responsibility of the past year (Jonnergård & Kärreman, 2004;
Hellman, 2011).
The boards of directors are elected at the AGM and are responsible for the company’s organization and also the running of the company’s business. The board of directors has a vast authority but cannot elect new board members, auditors or adopt the annual accounts. The members of the board of directors can be categorized in three categories. First there are the owner representatives, second the labor representatives and finally the independent directors. Only one of the members is allowed from the management, this is often the CEO. Further the owner representatives are often considered to be of a higher rank since they are considered to represent the owner’s interest (Jonnergård & Kärreman, 2004; Hellman, 2011).
The CEO is a subordinate to the board of directors and responsible for the daily management. For matters that fall outside the scope of day to day operations the CEO must present this to the board of directors for instructions. The CEO is a full time position while even the chairman of the board of directors often is only part-time. This gives the CEO a stronger position. The difference in strength could possibly induce agency issues, especially between a founder CEO which holds extensively knowledge of the firm and often holds a large share post of high vote dual-class shares (Söderström et al., 2003; Hellman, 2011).
The auditors are appointed at the AGM and serve the owner’s interest when examining the income statement, balance sheet and the company's profit or loss presented by the board of directors. The auditor's tasks are also to review the board of directors and CEO management of the company (Söderström et al., 2003; Hellman, 2011).
6
3. Theoretical framework
In this chapter the agency theory will be presented first. Followed by the earnings quality concept and how it is affected by first entrenchment and later alignment effects. The chapter is concluded with a presentation of prior studies and followed by the formulation of the hypothesis.
3.1 Agency Theory
When there is a separation between ownership and control the possibility of conflicting interests arise, this is examined and explained in the agency theory. The following quote by Jensen & Meckling (1976) describes the core of the agency theory:
“We define an agency relationship as a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent. If both parties to the relationship are utility maximizers there is good reason to believe that the agent will not always act in the best interest of the principal.”
(Jensen & Meckling 1976, p. 308)
The classic agency issue described by Jensen & Meckling (1976) concerns the issues between management and shareholders. The information asymmetry between management and shareholders allows the management to participate in moral hazard behavior and transfer resources from the shareholders to themselves. This is due to the conflicting interests between management and shareholders, where the manager will always try to maximize their own wealth. This issue is mitigated by contracts that can prove to be costly; these contracts align the interests between the parties. It can be performance based pay, share based pay or other contracting forms (Eisenhardt, 1989; Jensen &
Meckling, 1976; Demarzo & Berk, 2011).
The literature (Bebchuk et al., 2000; Morck et al., 1988) identifies a second agency problem between large controlling (family) shareholders and minority shareholders (Shleifer & Vishny, 1997). The minority shareholders are concerned that the family could engage in behavior of moral hazard nature.
This moral hazard behavior could be the extraction of private benefits. The minority shareholders bear an unproportioned part of the cost when the controlling (families) shareholders extract private benefits.
In this case the moral hazard occurs when the agent does not bear the full effect of the actions which in turn influences the course of action (Johnson et al., 2000).
The incentives for the extraction of private benefits are reduced when the controlling (family) shareholder increase their cash-flow rights. When the wedge is reduced the cost for the extracted private benefits are to a larger extent transferred to the controlling (family) shareholder. Further large cash flow rights increase the incentives of the large shareholder to monitor the management because the time and effort spent on monitoring benefits large shareholders the most. The monitoring activity is costly which limits the benefits for minority shareholders, since the monitoring cost could exceed the agency cost (Tirole, 2006).
Incentives to engage in moral hazard activities increase when there is a divergence from the “one share one vote” principle. The trade-off from private benefits extraction described above is larger and less cash-flow rights decrease the monitoring incentives. However, this is just incentives. Families could be superior monitors or possess other competencies which are superior to other shareholders leading to a better performance despite the wedge (Anderson & Reeb, 2003).
3.2 Earnings Quality
There is no widely accepted definition of earning quality. Rather the definition depends on the context.
Francis et al. (2005) identifies seven attributes of earnings. Accrual quality is to place a relation
7
between the current period working capital accruals and the operating cash flow from the period prior, present and of the future periods. By comparing the management's estimates of cash flow and to what extent the accruals differ from the cash flows, changes in revenue and PPE due to estimation errors is a way to measure accruals quality. The persistence of earnings is important since if an earning is returning it can to a larger part be considered sustainable, this is also important for investors as it suggest that the earnings process is stable and low-risk. Predictability is a desirable attribute since earnings of high quality can be predicted; this also reduces the volatility of earnings forecasts.
Smoothness; where managers use their private information about the business and future events in an effort to smooth out earnings fluctuations to increase the usefulness of the reported earnings. However, there is some controversy as smoothed earnings could be considered a way to mask the firm's real performance. The presence of smooth earnings could be an indicator of earnings management. Value relevance is the ability of earnings to clarify a variation of returns. Thus the earnings are of high quality and value relevant if they are able to explain the firm’s market price. Timeliness and Conservatism come from the perspective that the purpose of accounting earnings is to measure economic income which is defined as changes in market value of equity (Francis et al., 2005).
Dechow, Ge and Schrand (2010) define earnings quality as follows:
“Higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker.”
(Dechow, Ge and Schrand, 2010 p 344) According to Dechow et al. (2010) there are also three important aspects to take into consideration.
First earnings quality is dependent on the decision-relevance of the information. Secondly the quality of the reported earnings number depends on if it reflects the performance of the firm. Finally, earnings quality depends on the quality of the underlying financial performance and by the capacity of the accounting system to measure performance.
3.3 Earnings Management
An important aspect of financial reports is for the managers to transfer information on the firm's future performance. In order to do this the manager must use judgment and knowledge of the business to give as a precise picture as possible. There is a possibility that the manager instead uses this chance opportunistically. When managers manage earnings and act opportunistically it is called earnings management (Healy & Wahlen, 1999). One widely used definition of earnings management (EM) is:
“Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on reported accounting numbers.”
(Healy & Wahlen 1999, p 368)
There are two paradigms in EM, first there is the real EM. Here operational decisions are made with the intention of boosting short-term earnings. It could be achieved through decreasing necessary investments, abstaining from profitable projects or delaying impairments. While a destructive behavior it can be difficult to detect (Cohen et al., 2008). Secondly there is accrual EM which can according to Scott (2012) be divided in four subsections. The big bath accounting during reorganizations or a bad year, when a negative result cannot be avoided the management is willing to take an extra-large loss, either by excessive provisions, impairments or shifting earnings into the next period. The income maximization to maximize the earnings in order to maximize profit either for debt covenants or bonus plans. The income minimization; where in order to not attract attention in regulated markets the income is minimized. The income smoothing where the management in an effort to reduce the
8
volatility in the reported earnings choose to smooth the earnings, by reducing spikes and dips. By sustaining smooth level of earnings the firm appears as a stable and good firm (Scott, 2012).
There could be several incentives for management to manipulate the earnings. As both analyst and investors rely on the financial reports for information it could create incentives for management to mislead the earnings in order to boost the short-term stock valuation by target beating. A second incentive for management could be to maximize the bonus plans used to align the interest between managers and shareholders. A third incentive could be that the firm’s loans have debt covenants where certain financial ratios must be upheld. Otherwise the loan could be cancelled or otherwise altered resulting in an increased cost of capital (Cheng & Warfield, 2005; Healy & Wahlen, 1999).
In order to detect accrual EM several models have been constructed in order to identify the discretionary accruals. This study will use two versions of the Modified Jones in order to measure the discretionary accruals; these will be presented and explained in 4.4. The discretionary accruals will further be used as a proxy to measure the earnings quality.
3.4 Entrenchment Effect
The entrenchment effect is that concentrated ownership creates incentives for the controlling shareholders to expropriate wealth from the other shareholders (Bardhan et al., 2015). In firms where the controlling shareholder is a family it means that family shareholders could consume firm resources at the expense of non-family shareholders (Wang, 2006). This behavior is described by Johnsson et al.
(2000) as tunneling when the firm's assets are transferred out of the company to the controlling shareholders. This expropriation could take the form of pet projects, salary benefits or transfer-pricings benefitting the controlling shareholders. These forms of entrenchment could be legal but still destroy significant firm value (Zerni et al., 2010).
Further the ownership structure of a firm significantly affects the reported earnings quality (Katz, 2009). Studies have shown that family firms have fewer incentives for reporting earnings of high quality. Francis et al. (2005) find that US firms with dual-class shares compared to firms with single class shares have lesser corporate governance and report earnings with an inferior informative aspect.
This could support the entrenchment theory as the main possessor of dual-class shares are family shareholders (Francis et al., 2005). Bardhan et al. (2015) also examined the US setting with a sample of S&P 500 firms. They found that family firms with dual-class shares exhibited an increased likelihood of material weaknesses in the firm's internal control. Also Chen et al. (2008) found that family firms disclose less voluntary information compared to non-family firms. In a study conducted in East Asia, Claessens et al. (2002) examine the wedge between cash-flow rights and voting and finds that when the difference increases the valuation of the firm decreases. This is due to the entrenching behavior of the majority shareholder.
The demand of earnings quality is influenced by the market. The entrenchment effect assumes that the earnings quality is of low quality in family firms. The reported inferior earnings quality raises the cost of capital and could offer family firms worse contracting terms, the contracting terms could be more closely tied to the quality of the earnings. The inferior corporate governance in family firms could increase the demand of higher quality earnings as an opportunity to offset the discount on the equity (Wang, 2006).
3.5 Alignment Effect
In the alignment effect the interest of family shareholders and minority shareholders are better aligned.
The family shareholders are controlling shareholders and have a long-term perspective combined with a long term presence. Due to this family firms are less likely to expropriate wealth. There is also stronger monitoring from the owners as they are often present on the board (Wang, 2006; Cascino et al., 2010). The management is more closely monitored as the wealth of the family is closely tied to
9
firm value and any scandals would also damage the family's reputation, this and their undiversified portfolio (Anderson et al., 2003). Fan and Wong (2002) agrees with that the reputation of the family is important and also that there is a positive effect when the majority owner increases their cash-flow rights beyond the minimum amount required for control. There are incentives for the family to communicate more effectively and report higher quality earnings quality to shareholders and debtors, thus reducing cost of debt (Anderson et al., 2003). Thus the family’s long-term perspective, their interconnectedness with the firm and will to prevent reputational loss, prevents families from engaging in earnings management. The alignment effect anticipates that the family ownership is tantamount to stronger corporate governance. This stronger corporate governance reduces the possibility of earnings management by managers.
Studies have shown that the earnings quality of family firms compared to non-family firms are of high quality. Wang (2006) found that founding family firms are associated with higher earnings quality than non-family firms. Cascino et al. (2010) investigated Italian listed firms and found that the information reported by family firms was of higher quality compared to non-family firms (Cascino et al., 2010). In Portugal Alves (2012) found that ownership concentration reduced the occurrence of earnings management. Bardhan et al. (2015) argues against the alignment effect and is concerned with the low level of internal control mechanism in firms with concentrated ownership. In a further study by Ali et al. (2007), they examine family firms also in the US. By using similar variables as the Wang (2006) study and also investigating the disclosures of their corporate governance practices. By measuring earnings quality by the level of discretionary accruals they find that family firms provide better quality earnings. Another finding is that firms with founder CEO are mainly responsible for providing better disclosures, however family firms otherwise provide fewer details of their corporate governance practices. The Swedish study by Cronqvist and Nilsson (2003) supports the alignment effect. However, the largest discount of firm value is found in family firms with the presence of controlling minority shareholders; this is due to their long term involvement and control mechanisms.
3.6 Prior Studies
In the study by Fan and Wong (2002), they investigate the relation between earnings informativeness and concentrated ownership in seven East Asian countries. Their sample consists of 977 firms with a total of 3,572 firm-years between the years 1991 and 1995. Their hypothesis is that the ownership concentration and wedge between cash-flow rights and voting rights would weaken the earnings quality. They provide empirical support for two explanations of the low earnings quality; first that the minority investors anticipate that the controlling owners will entrench themselves which weakens the credibility of the reported earnings. Their second explanation is that the owner tries to minimize the leak of firm specific information to competitors and the public.
In another East Asian study by Claessens et al. (2002) unravels the entrenching effect of large shareholders and family groups, through the use of dual-class shares and pyramid ownership. Through examining 1,301 publicly listed firms in eight East Asian counties during the year 1996. They find that the valuation of a firm decreases as the controlling shareholder increases their voting rights of the firm.
When they separate the effects by ownership type, the results are mainly driven by family ownership.
The researchers suggest that institutional functions such as the quality of banking systems, legal protection of minority ownership and the level of disclosures required in the financial reports could affect the ownership driven discount.
A more recent study by Prencipe and Bar-Yosef (2011) that examines the effect the board independence has on earnings management in family firms. By using a sample of 249 observations of firms listed at the Milan Stock Exchange during the years 2003 and 2004. Traditionally the presence of independent board members would be associated with a reduced level of earnings management in a firm. The study investigates abnormal working capital accruals which are used as a proxy for earnings management. The findings suggest that in family firms the independent board members are less independent due to the long term time aspect of the family and the earnings management decreasing
10
effect of independent board members is weaker than in non-family firms. These effects strengthen when the CEO is a part of the family.
In the US study by Bardhan et al. (2015) the relationship between family-firm characteristics and the quality of the internal control over the financial reporting compared to non-family firms. The study uses a sample of 446 firms between the years 2002 and 2006. By examining the material weaknesses of the financial reporting under the SOX regulation they find that family firms with dual-class shares present more material weakness compared to non-family firms. This is partly due to the absence of internal control mechanisms. However, in family firms that do not employ the use of dual-class shares there is no difference between family and non-family firms. The results suggest that the incentives of entrenchment outweigh the concern for the firm’s long-term wellbeing.
In contrast to the studies that support the entrenchment theory there is the study by Wang (2006) that investigates if founding family firms has an effect on the quality of financial reporting in the US. The study uses a sample of 542 firms during the years 1994 and 2002. By using discretionary accruals as a proxy for earnings quality and including variables for the nature of the CEO, if the CEO is the founder, a descendant or externally hired, and the impact this has on the financial reporting. The results show that a founder CEO has a positive effect on the earnings quality and that founder owned family firms are positively associated with decreased levels of earnings management and increased levels of earnings quality.
Further in the Ali et al. (2007) study that investigates the disclosures by family firms in the US. Their sample consists of 500 firms during the years 1998 and 2002 in the S&P 500 list. By examining firms discretionary accruals in the reported earnings, the disclosures of the corporate governance practices and the voluntary disclosures of negative news. Their findings suggest that the reported earnings of family firms are of higher quality. Family firms are more likely to warn of poor earnings when compared to non-family firms. However, family firms make less disclosures of their internal control governance than non-family firms. Firms with founder CEO are also more associated with better disclosures, this is also an effect seen in firms with dual-class shares.
In Europe the study by Alves (2012) investigates the ownership structure and earnings management in Portugal. The sample consists of 34 listed firms between the years 2002 and 2007. By using discretionary accruals as a proxy for earnings management and examining the effect the ownership structure has on managed earnings. The results suggest that ownership concentration and managerial ownership decreases the levels of earnings management and increases the earnings quality. Further the study finds that earnings management increases when the political cost and leverage is high.
The study by Cascino et al. (2010) explores the quality of accounting information in family firms. By using 778 listed firms during the years 1998 and 2004 in Italy. The study compares the informativeness of the financial statements in family firms and non-family firms as a proxy for earnings quality. The results show that family generally report earnings of higher quality compared to non-family firms. The findings further show that the financial reporting in family firms is more transparent and less vulnerable to managerial manipulation.
In the Swedish study by Cronqvist and Nilsson (2003), they investigate the agency costs of the controlling minority shareholders. Their sample consists of 309 listed firms during the years 1991 and 1997. The minority controlling shareholders uses dual-class shares and other control mechanisms.
Their findings suggest that family controlling minority shareholders are associated with the largest discount on firm-value compared to the owners of corporations and financial institutions. This discount is associated with that the return on assets are lower for family firms and the fact that family owners long-term perspective decreases the prospect of the firm being taken over. However, they suggest there is a possibility that this discount on firm value were already present when the firm initially was listed and resulted in a lower share price.
11
Finally in the Swedish study by Khosravi (2015) the agency problems between majority and minority owners concerning CEO compensation and corporate governance are investigated. The study uses a sample of 1164 firm years between the yeas 2005 and 2009. By examining how family firms with dual-class shares affect the CEO compensation. In family firms the CEO are compensated less with performance based compensation than in non-family firms. The findings suggest that performance based pay for a family CEO would be effective in reducing the conflicting interests in family firms.
Prior research has found conflicting results when examining the accruals as a proxy for earnings quality of family firms. The theories of entrenchment and alignment yield predictions of opposite directions for the earnings quality in family firms. This study’s hypothesis therefore states:
H1: Family firms provide higher (lower) quality earnings
Table 1 Summary of Prior Studies
Study Description of the study and main results
Fan & Wong, 2002 977 firms in seven East Asian countries between 1991 and 1995. The study investigates the wedge between cash-flow and voting rights and the effect this wedge has on earnings quality. Minority owners anticipate the entrenchment and make adjustments.
Claessens, et al., 2002 1301 firms in eight East Asian countries in 1996. By investigating the effects that the controlling shareholder has on the valuation. They find that when the wedge between cash-flow rights and voting-rights increase the firm value decreases consistent with the entrenchment theory. This is mainly driven by family ownership.
Bar-Yosef and Prencipe, 2011 249 observations in Italy, during the years 2003 and 2004.
They investigate the effect that independent board members have on the earnings management in family firms. They find that the effect is weakened in family firms and the presence of a family CEO increases the effects.
Bardhan et al. 2015 446 firms in the US between the years 2002 and 2006. The study finds that family firms have more material weakness in their financial reports. They also find that the presence of dual-class shares further drives the presence of material weaknesses.
Wang, 2006 542 firms in the US between 1994 and 2002. The study investigates earnings quality of founding family firms.
Family firms report a higher earnings quality by having lower levels of discretionary accruals and more informative earnings.
Ali et al., 2007 500 firms in the US between 1998 and 2002. The study find that family firms provide better quality earnings but provide less disclosures on their internal corporate
12
governance practices. Family firms have also more analysts following.
Alves, 2012 34 firms in Portugal between the years 2002 and 2007. The study finds that managerial ownership and concentrated ownership are negatively associated with earnings management, which would indicate that the earnings quality is of higher quality since the discretionary accruals are smaller.
Cascino et al., 2010 778 firm-year observations between the years 1998 and 2004 in Italy. Find that the financial reporting in family firms is of higher quality and less inclined to be manipulated by the management.
Cronqvist & Nilsson, 2003 309 listed firms in Sweden between the years 1991 and 1997. They find that family controlling minority shareholders are associated with the largest discount on firm value. This is due to their long term involvement and control mechanisms.
Khosravi, 2015 1164 firm-year observations in Sweden between the years 2005 and 2009. The study investigates the agency problems in family controlled firms and the issue with the voting wedge. The study finds that performance based compensation is an effective mechanism to mitigate agency cost in family firms.
13
4. Method
In this chapter the research design will be presented first. The sampling process will be described subsequently. This will be followed by a description of the models used.
4.1 Research Design
The purpose of this thesis is to examine the earnings quality of family firms in Sweden. This is achieved by comparing all publicly listed firms in Sweden, except financial firms and banks, during the years 2005 and 2014. A quantitative research design is deemed most appropriate for testing the hypothesis developed in the theoretical framework. The purpose is fulfilled through the outcome of the statistical test and the subsequent analysis. The hypothesis is presented below:
H1: Family firms provide higher (lower) quality earnings
4.2 Sample and Data Collection
The initial sample for this thesis includes all firms listed at the Stockholm Stock Exchange (SSE) during the years 2005 - 2014. During these years there are a total of 2618 firm year observations. The data on vote percentage, equity percentage and if the CEO are the founder was for the years 2010 – 2014 hand-collected through their annual reports and the corporate governance section. The classification of firms as family or non-family was manually conducted. There is an issue when manually collecting data, the risk that the collected data diverges from the original or subjectivity from the collector when making judgments on how the data is interpreted and entered. To mitigate these issues the data was collected over a concentrated period of time in a consistent fashion. All values were double checked to avoid that the data diverged from the original. For the ownership data for the years 2005 – 2009 the data was provided by Khosravi (2015). The empirical data required for the statistical models was collected through DataStream for the years 2004 - 2014, the year 2004 were included due to that several of the variables in models is calculated through the change of variables between years. This could induce issues since IFRS was made mandatory in Sweden during 2005, this could further result in different numbers between the years due to a difference in accounting regulation. However, to mitigate this potential problem the test was conducted twice, first with the year 2005 excluded and subsequent test included. The test where 2005 was excluded (can be found in Appendix 1) showed consistent results as the test where it was included. This resulted in the decision of including the year 2005.
The exclusion of firm observations was conducted as follows. First financial firms and banks were identified through their ICB code (8300 - 8500) and excluded. Foreign firms and firm years with missing data on the ownership structure were removed. Finally, firm years that had insufficient data required for the models or other variables included in the regression were also excluded. When the missing values had been excluded there were 16 remaining observations in the ICB 0001 and 0 observations in the ICB 7000. This leads to a final sample of an unbalanced panel dataset of 1773 firm-year observations.
14
Table 2 Sample composition
Panel A: Sample selection criteria Total
Initial firm-year observation for years 2005-2014 2618
1. Less banks and insurance companies (ICB codes 8300 - 8500) (40)
2. Less missing ownership data (296)
3. Less missing other data from DataStream (509)
Final sample 1773
Panel B: Industry composition
Number of Percent of
ICB code observation observations
0001 Oil & Gas 16 0.9
1000 Basic Materials 99 5.6
2000 Industrials 628 35.4
3000 Consumer Goods 179 10.1
4000 Health Care 226 12.8
5000 Consumer Services 213 12.0
6000 Telecommunications 22 1.2
8000 Financials 90 5.1
9000 Technology 300 16.9
Total 1773
Panel C: Distribution of observations by fiscal year
Year 2005 2006 2007 2008 2009
Number of observations 164 181 187 185 184
Year 2010 2011 2012 2013 2014 Total
Number of observations 176 180 176 168 172 1773
4.3 Variables
The variables required for the modified Jones model by Dechow et al. (1995) and the adjustments to the model made by Kothari (2005) was collected from DataStream and their DataStream code and description can be found in Appendix 2.
Consistent with prior studies (Wang, 2006; Ali et al., 2007; Alves, 2012; Cascino et al., 2010) the following control variables are included; if the founder of the firm serves as CEO (Ceofoundert), the presence of dual-class shares (dualt). The performance (roat) and the risk for bankruptcy or of violating debt covenants (losst & levt), the firm size measured by the natural logarithm of total assets (sizet) and growth measured by the change in net sales between t and t-1 (growtht) and if there is a large non- family shareholder exceeding 15 % of the votes (concownt).
15
Table 3 Variables
Variable Definition
Model1t The discretionary accruals from the modified Jones (Dechow et al., 1995) model at t;
Model2t The discretionary accruals from the modified Jones modified by Kothari (2005) model at t;
Ceofoundert Dummy variable for if the CEO also is the founder at t. It is uncertain how the presence of a founder CEO will affect the earnings quality (Ali et al., 2007),
Dualt Dummy variable for if there are more than one type of shares with different voting rights at t. Firms with dual-class shares are expected to provide lower quality earnings (Francis et al., 2005),
Roat Used as a proxy for performance. Firms with low performance is expected to manipulate earnings to a larger extent than firms with higher performance (Dechow et al., 2010)
Capitallargestt The amount of the cash flow rights that is owned by the largest shareholder in percentage at t;
Votinglargestt The percentage of voting power of the owner with most voting power at t;
Famownt Dummy variable for if the largest voting holder exceeds 15 % and are a family owner or if the CEO of the firm also is the founder at t, it is uncertain what effect the family ownership will have on the earnings quality (Bardhan et al., 2015; Alves, 2012; Wang, 2006;
Claessens et al., 2002);
Concownt Dummy variable for if the largest voting holder exceeds 15 % and is not a family owner at t, Firms with a large block holder is expected to supervise the management to a larger extent than firms with dispersed ownership, this will result in higher earnings quality (Wang, 2006; Alves, 2012);
Votedift The excess voting rights calculated as the difference between capital largest and voting largest at t;
Losst Dummy variable for if the firm reported a loss at t. Firms that report a loss is expected to provide lower quality earnings (Wang, 2006);
Levt Leverage at t, measured as total liabilities scaled by total assets, Leverage is expected to affect earnings quality negatively (Wang, 2006);
Sizet The natural logarithm of total assets at t;
Growth The change in sales between t and t-1.
4.4 Models
In order for the financial reporting to be informative it is essential that the earnings are of high quality.
The firms use accruals as a tool to measure and report the earnings. But the accruals are to a large extent under the influence of management due to their large reliance of judgment and assumptions
16
(Healy & Wahlen, 1999). Therefore the quality of the earnings is directly linked to the level of discretionary accruals. This study will therefore use first the modified Jones model adjusted by Dechow et al. (1995) and also adjust it according to the suggestions of Kothari et al. (2005) in order to control for the effects of firm performance. This is achieved by controlling for the lagged ROA due to the possibility of correlation between performance and the residuals. Equation 1 will present the modified Jones by Dechow et al. (1995) and equation 2 will present the model with the adjustments suggested by Kothari et al. (2005).
𝑇𝐴𝑖𝑡
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1 = 𝛽0+ 𝛽1 1
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1+ 𝛽2∆𝑅𝐸𝑉𝑖𝑡− ∆𝑅𝐸𝐶𝑖𝑡
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1 + 𝛽3 𝑃𝑃𝐸𝑖𝑡
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1+ 𝜀𝑖𝑡 𝑇𝐴𝑖𝑡 (1)
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1 = 𝛽0+ 𝛽1 1
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1+ 𝛽2∆𝑅𝐸𝑉𝑖𝑡− ∆𝑅𝐸𝐶𝑖𝑡
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1 + 𝛽3 𝑃𝑃𝐸𝑖𝑡
𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1+ 𝛽4𝑅𝑂𝐴𝑖𝑡−1+ 𝜀𝑖𝑡 (2) In the models TAit is the total accruals for firm i in the year t. It is calculated by withdrawing the change in current liabilities minus depreciation and amortization from non-cash current assets (TAit = [∆CurrentAssetsit - ∆Cashit] – [∆Currentliabit - ∆Short Term Debtit] – Depreciationit). Assetsit-1 is total assets for firm i in the year t. ∆REVit is the change in sales for firm i in the year t-1 and the year t, ∆RECit is the change in receivables in the for the firm i in the year t-1 and the year t. PPEit is the gross property, plant and equipment for firm i in the year t. Finally ROAit-1 is the lagged return on assets.
This study will use the residuals from equation 1 and equation 2 to use as two estimations for discretionary accruals in the regression model. Further an increase in the residuals indicates lower earnings quality.
The regression models are presented below, in Model 1 the dependent variable is the absolute value of the discretionary accruals given from the modified Jones model (Dechow et al., 1995), in Model 2 the dependent variable is the absolute value of discretionary accruals that stems from the adjustments suggested for performance by Kothari et al (2005), in both regressions this study will use OLS standard errors.
𝐷𝐴 1𝑡 =0+1𝑓𝑎𝑚𝑜𝑤𝑛𝑡+2𝑠𝑖𝑧𝑒𝑡+3𝑟𝑜𝑎𝑡+4𝑑𝑢𝑎𝑙𝑡+5𝑙𝑜𝑠𝑠𝑡+6𝑙𝑒𝑣𝑡 +7𝑐𝑒𝑜𝑓𝑜𝑢𝑛𝑑𝑒𝑟𝑡+8𝑐𝑜𝑛𝑐𝑜𝑤𝑛𝑡+ 𝑦′𝑖𝑐𝑏+𝑦′𝑦𝑒𝑎𝑟 +ε𝑡
(3) 𝐷𝐴 2𝑡 =0+1𝑓𝑎𝑚𝑜𝑤𝑛𝑡+2𝑠𝑖𝑧𝑒𝑡+3𝑟𝑜𝑎𝑡+4𝑑𝑢𝑎𝑙𝑡+5𝑙𝑜𝑠𝑠𝑡+6𝑙𝑒𝑣𝑡
+7𝑐𝑒𝑜𝑓𝑜𝑢𝑛𝑑𝑒𝑟𝑡+8𝑐𝑜𝑛𝑐𝑜𝑤𝑛𝑡+𝑦′𝑖𝑐𝑏+𝑦′𝑦𝑒𝑎𝑟 +ε𝑡
(4)
