Payout policy in family firms

(1)

Payout policy in family firms

– A study on payout levels and dividend smoothing in Sweden

Master’s Thesis 30 credits

Department of Business Studies Uppsala University

Spring Semester of 2019

Date of Submission: 2019-05-29

Patrick Bolin Carl Widerberg

Supervisor: Adri De Ridder

(2)

Acknowledgements

We want to thank our supervisor Adri De Ridder for valuable advice, feedback and guidance throughout the thesis process. In addition, we would like to express our gratitude towards the many people at Uppsala University who have been part of our past five years and helped us in laying the foundation for this thesis.

________________________________ ________________________________

Patrick Bolin Carl Widerberg

(3)

Abstract

This study investigates payouts in Swedish family firms by focusing on both the level and speed of adjustment of dividends. In addition, the use of dual-class shares in family firms is examined to further identify potential drivers of payout differences between family-controlled companies and non-family firms. Agency theory and previous studies suggest that high and stable payouts are used by controlling families to mitigate minority shareholders’ concerns of being expropriated. We find that family firms in Sweden do not differ from non-family firms in their payouts. The results could be seen as an indication of expropriation if minority shareholders should be compensated for higher agency costs, but it could also be that family control does not worsen agency conflicts between majority and minority shareholders. Rather, other ownership structures such as the use of dual-class shares to gain control in excess of ownership seem to be associated with higher levels of payouts. Neither do family firms smooth their dividends more than non-family firms. Instead, they adapt towards their target dividend at a higher pace.

Keywords: Family firms, payouts, dividends, dual-class shares, expropriation, dividend smoothing

(4)

1. Introduction

Controlling shareholders have the power to affect firms’ financial decisions to satisfy personal preferences. Therefore, the dividend policy in companies with a controlling shareholder can differ from the dividend policy in widely held firms. Family owned firms is the most common form of concentrated ownership in continental Europe (Faccio and Lang, 2002). Despite them being a cornerstone in many economies, the knowledge about family firms as an economic phenomenon is still relatively shallow (Credit Suisse Research Institute, 2018). According to prior research, controlling families can use dividends as a governance tool to mitigate minority shareholders’ fear of being expropriated (Pindado et al., 2012). An opposing view is that family ownership reduce agency costs and decrease the need for payouts as a governance tool (Gugler, 2003). By examining payouts in Swedish family-controlled companies, this paper studies whether family firms in Sweden differ from non-family firms in payouts.

1.1 Background

The dividend irrelevance theorem presented by Miller and Modigliani (1961) suggests that payout policy should not matter to investors. However, empirical findings reveal that investors tend to appreciate raised dividends and dislike dividend cuts (Richardson Pettit, 1972). Due to the separation of ownership and control, agency problems arise as there might be a conflict of interest between management and shareholders (Jensen and Meckling, 1976). Jensen (1986) argue that dividends can be used as a disciplining tool to hinder management from investing in projects with negative net present value. Managements’ awareness of investor preferences is evident since many managers are reluctant to cut dividends due to the negative share price effect (Brav et al., 2005; Lintner, 1956). Shiller (1981) confirms these tendencies as dividends are shown to be more stable than earnings.

Besides the potential conflict of interest between shareholders and management, agency problems can also arise if companies are controlled by a majority shareholder. Potential costs emerge since minority shareholders face the risk of being expropriated by the controlling shareholder (Morck et al., 2005; Shleifer and Vishny, 1997). By influencing financial decisions to satisfy personal preferences, a controlling shareholder may steer the firm to engage in projects that are not value maximizing instead of distributing excess cash to shareholders. For example, the founding family of the Swiss chemical and construction company Sika received criticism from other shareholders after the annual general meeting in 2017 (Reuters, 2017).

(6)

The family utilized their voting power to reject the suggested dividend as they preferred that Sika would retain a larger portion of the earnings. Instead, they proposed a lower dividend which they were able to approve thanks to their majority ownership.

Contrary to theories of expropriation, an alternative view suggests that family control is not less effective from an agency perspective. Instead of distributing funds to hinder inefficient investments by management (Jensen, 1986), the large stake held by controlling families could serve as a monitoring mechanism itself. Controlling families might be more concerned with the long-term health of the company which could prohibit inefficient investments and expropriation (Miller et al., 2008). Anderson and Reeb (2003a) argue that family ownership actually seem to reduce agency conflicts instead of exacerbating them. Furthermore, Chen et al. (2010) state that reputational concerns of the family name could hinder controlling families from expropriating minority shareholders.

1.2 Problematization

Considering that prior research associates family control with both benefits and costs, the effect on agency problems is not clear. Family firms may have incentives to diverge from non-family firms in payouts if their ownership structure affects agency costs. Results on differences between the two are mixed. For example, Pindado et al. (2012) show that family-controlled firms in the Eurozone pay higher dividends, which is interpreted as an effort from the controlling owners to alleviate minority shareholders’ concerns of expropriation. On the other hand, Gugler (2003) study dividends in Austrian family firms and find that family firms distribute less funds to shareholders. He argues that family control serves as a substitute to dividends in corporate governance. Attig et al. (2016) and Mulyani et al. (2016) find similar results for East Asian firms. However, they argue that lower dividends in family firms can be interpreted as expropriation of minority shareholders.

In addition to disparity in payout levels, several studies investigate differences in how quickly firms adapt dividends to changes in earnings (Pindado et al., 2012; Gugler, 2003).

Lintner (1956) presents the concept of dividend smoothing and the speed of adjustment towards a target ratio. The smoothing arises because firms aim to maintain stable payout levels and avoid cutting dividends. High and stable dividends can be a way for firms to mitigate agency problems (Leary and Michaely, 2011). More dividend smoothing can therefore be seen as a tool

(7)

to manage minority shareholders’ concerns of expropriation. This view is supported by Pindado et al. (2012), who show that family firms are more likely to smooth dividends. The authors argue that their results are evidence of controlling families using dividends as a governance mechanism. On the contrary, Gugler (2003) find that family firms in Austria smooth their dividends less than other companies.

The severity of agency problems could also partly depend on the institutional setting in which a firm is based (Isakov and Weisskopf, 2015). In Sweden, family-controlled companies is a common form of ownership structure among publicly traded firms (Faccio and Lang, 2002).

According to La Porta et al. (1998), the Swedish institutional setting provides intermediate protection for minority shareholders. The law enforcement is considered to be strong, but regulations allow for several restrictions of minority shareholders’ rights. One such example is the use of shares with a dispersion between voting and cash flow rights. Faccio and Lang (2002) examine the use of such dual-class shares in Western European countries and find that Sweden is the country where dual-class shares are most frequently used. The ownership structure and institutional setting make Sweden suitable for examining payouts in family-controlled companies.

1.3 Purpose

Despite the theoretical ambiguity and practical relevance of payouts, previous literature on payouts in family firms is scarce (Isakov and Weisskopf, 2015). The purpose of this study is to contribute to the understanding of payouts’ role in agency problems between controlling families and minority shareholders. By studying both payout levels and the speed of adjustment, this study contributes with a complementary analysis in Swedish family firms. The understanding of firms’ payout decisions is also important from a monetary perspective since payouts provide a potential source of return for investors. The paper builds on to existing knowledge by examining the following question:

How does family control affect payouts in dividend-paying Swedish firms?

(8)

1.4 Disposition

The remainder of this paper is organized as follows: The next section portrays the theoretical framework of the study, reviews previous findings on payouts in family firms and presents the hypothesis development. The methodological procedures used to test the hypotheses are presented in section three in combination with summary statistics of the sample. In section four, descriptive statistics are presented, the regression results are analyzed and the robustness of these findings are tested with alternative specifications. In section five, the main findings are discussed and conclusions highlighted. Finally, potential implications and suggestions for future research are presented.

(9)

2. Theory

2.1 Payouts

Decisions on how much capital companies should distribute to their shareholders and how much to retain is a widely discussed topic. Historically, payouts have been viewed as a residual available after investments in value generating projects (Miller and Modigliani, 1961).

However, over time both the methods and means for distributing funds to shareholders have broadened. There are different alternatives available for companies, and market reactions indicate that investors tend to appreciate some payout decisions while reacting negatively to others.

2.1.1 Dividends

The most common form of cash distribution to shareholders is dividends, meaning that shareholders entitled to payouts receive a share of the company’s equity (Floyd et al., 2015). In general, dividend paying firms are bigger, more profitable and make less investments in relation to their earnings compared to companies that do not pay dividends (Fama and French, 2001).

Managers are generally reluctant to cut dividends and stable payouts can earn companies a positive reputation (Brav et al., 2005; Lintner, 1956). One example is the American multinational corporation Procter & Gamble that has increased its dividends 63 years in a row, earning them a place on the list of Dividend Champions¹. In addition, there are also examples of firms paying dividends which exceeds the earnings for the year. In 2019, the board of global retailer Hennes & Mauritz proposed an unchanged dividend at 9.75 SEK per share despite earnings per share of 7.64 SEK (H&M, 2019). The fact that companies are seemingly willing to borrow money or reduce cash reserves to finance dividend payouts suggests that insiders and outsiders view dividends as more than a residual (Brav et al., 2005). Fama and French (2001) report a drop in the number of dividend paying industrial companies in the United States from 1978 to 1999. DeAngelo et al. (2004) claim that dividends on an aggregate level increased over the period even though the fraction of firms paying dividends decreased. Furthermore, Floyd et al. (2015) find that the declining trend reverses after 2002. A similar pattern is found by von Eije and Megginson (2008) for European firms, and Baker and De Ridder (2018) report that the fraction of Swedish dividend paying companies on average is high compared to other regions.

1 Dividend Champions are companies which have increased dividends for at least 40 consecutive years (Miglietta et al., 2018).

(10)

2.1.2 Share Repurchases

By repurchasing its own shares, a company increases each shareholder’s portion of the company. Thus, instead of providing each shareholder with cash, share repurchases provide owners with funds in the form of increased capital value. Compared to dividends, repurchasing shares is a relatively new approach of distributing money to shareholders. Repurchasing shares has grown in popularity since the mid-1980s and today account for a substantial part of the distributions to shareholders (Fama and French, 2001; Floyd et al., 2015). In Sweden, popularity has surged after Swedish authorities removed restrictions on share repurchases in 2000 (Baker and De Ridder, 2018). Brav et al. (2005) claim that the overall view among company insiders is that the primary advantage of repurchases is the flexibility. They are perceived to be less sticky than dividends, which provides management with more leeway in year-to-year changes. Furthermore, reducing the number of shares instead of paying dividends is more tax efficient (Ibid).

2.1.3 Compulsory Share Redemptions

A third alternative when distributing cash to shareholders is a stock split with mandatory redemption. Companies split existing stocks and force shareholders to receive proceeds by adding a compulsory share redemption program at a predetermined price (Baker and De Ridder, 2018). Thus, the shares are listed on the stock exchange for a short period of time before they are redeemed. Baker and De Ridder state that it is a relatively new innovation in Sweden and that the approach can be beneficial for institutional investors as the proceeds are taxed as capital gains.

2.2 Agency Theory

As described by Jensen and Meckling (1976), the separation of ownership and control could lead to potential agency problems. The owners of a company (principals) hire managers (agents) to run the daily operations of the firm. If the priorities of the principal and agent are not aligned, a conflict of interest arises. Since managers are the ones in control, they can engage in self-centered behavior at the expense of shareholders. The occurrence of such problems is most salient in widely held firms in which control is concentrated in the hands of managers (Cronqvist and Nilsson, 2003). Berle and Means’ (1932) description of corporations in the United States as most commonly being widely held was long believed to be true for the rest of

(11)

the world as well (La Porta et al., 1999). However, more recent studies show that the presence of a controlling shareholder is the most common form of ownership in many parts of the world, especially outside common law countries (Ibid). Controlling shareholders mitigate the agency conflict between managers and shareholders by being able to monitor and influence management (Shleifer and Vishny, 1997).

At the same time, concentrated ownership gives rise to another type of agency conflict as controlling shareholders have the power to extract private gains (Shleifer and Vishny, 1997).

Minority shareholders’ fear of being expropriated by controlling owners is therefore the main agency problem in concentrated ownership settings. If there is a divergence in interests between controlling and minority shareholders, controlling owners might redistribute wealth at the expense of minority shareholders. The risk of expropriation for minority shareholders and other stakeholders is argued by Shleifer and Vishny to be especially prevalent in companies with deviation from the one-share-one-vote system. According to them, incentives for controlling owners to treat themselves preferentially increases if they have voting rights in excess of cash flow rights. In such situations, the controlling owner can fully enjoy private benefits while only partially carrying the burden of potential stock depreciation (Cronqvist and Nilsson, 2003). As payouts could be a governance mechanism for companies to alleviate agency problems, the payout level might be affected by a company’s ownership structure.

2.3 Dividend Theories

Several theories aim to explain the role of dividends and present different reasons to why payout decisions might matter. In their dividend irrelevance theorem, Miller and Modigliani (1961) state that in a perfect world, dividend payouts are irrelevant and that firm value is driven by a company’s investment policy. Thus, business risk and firms’ ability to generate earnings should be the factors relevant for investors. However, when easing the assumptions of the dividend irrelevance theorem, dividend decisions could matter to investors. Empirical findings show that investors tend to appreciate raised dividends and dislike dividend cuts (Richardson Pettit, 1972). Companies seem to be aware of these preferences as dividends are shown to be more stable than earnings (Shiller, 1981). The resistance from management to cut dividends is primarily due to the negative share price effect (Brav et al., 2005; Lintner, 1956).

(12)

Many explanations of dividends are rooted in agency theory and information asymmetry. For example, Jensen (1986) argues that dividends can be used as a disciplining tool to hinder management from investing in projects with negative net present value. Since management’s primary goal is growing the business, they have a tendency to overinvest. Due to agency problems, firms can distribute excess cash to alleviate shareholders’ concerns of inefficient usage of the company’s funds. Brav et al. (2005) show that firms with stable earnings growth are inclined to increase dividends, which is interpreted as a pressure to return capital to shareholders.

Another popular explanation to the role of payouts is the concept of signaling. The logic is that increased dividends or initiated share repurchases can be interpreted as positive signals from company insiders about expected future earnings (Bhattacharya, 1979). Since management are reluctant to cut dividends, they only initiate or increase payouts if they are convinced that the company is profitable enough to maintain payouts in the long run. Share repurchase programs can also be used in situations where management perceives the firm’s stock to be undervalued (Vermaelen, 1981). Brav et al. (2005) show that a strong majority of insiders interviewed and surveyed view firms’ payout decisions as reflecting the future prospects of their firm.

The clientele hypothesis states that companies set their payout policy in response to demand from different types of investors. Decisions on payouts are therefore made to attract certain clientele. Allen et al. (2000) argue that tax differences between institutional and individual investors can influence payout decisions in firms that want to favor institutions. Institutional investors’ ability to monitor the firm effectively may also be a driver behind firms attempting to attract them. Grinstein and Michaely (2005) show that institutions are attracted to firms that pay dividends and prefer firms that engage in share repurchases. Becker et al. (2011) further describe differences among individual investors as they state that retail investors appreciate local stocks and older investors prefer dividend-paying companies.

2.4 The Swedish Institutional Setting

The World Economic Forum ranks Sweden as a top ten nation regarding both the institutional- and financial setting by taking into account factors such as corporate governance, security, transparency and ethics (Schwab, 2018). Compared to other countries, Sweden’s financial system is more stable, which mitigates excessive risk-taking and opportunistic behavior.

(13)

A well-developed financial sector reduces information asymmetries and improves the allocation of capital. In addition, Holmén and Knopf (2004) argue that Sweden is characterized by strong extralegal institutions which provide protection for shareholders. These include for example media, tax authorities and social norms. Furthermore, La Porta et al. (1998) state that the law enforcement in Sweden is strong. The Swedish corporate governance model consists of a combination of legal rules, self-regulation and tacit rules. Important components are the Swedish Companies Act, the Swedish Corporate Governance Code and different rules connected to the stock exchanges (Swedish Corporate Governance Board, 2018).

Despite the top-tier ranked financial system, La Porta et al. (1998) state that Sweden only provides intermediate legal protection for investors. This is primarily related to the use of anti- shareholder devices (Ibid). La Porta et al. (1999) find that countries with strong protection for minority shareholders have more widely held firms compared to countries with weaker protection, in which family firms are more common. The Swedish market has a relatively large portion of concentrated ownership. Faccio and Lang (2002) investigate the ultimate ownership of European firms and report that almost half of the companies in Sweden are family-controlled.

The prevalence of dual-class shares, pyramiding and cross-ownership enables higher concentrations of votes compared to cash flow rights. Faccio and Lang (2002) show that Sweden has the highest fraction of firms with different voting rights. Sweden is also the European country in which the fewest cash flow rights are needed to control a company. On average, only 9.83 percent of the cash flow rights are needed to have 20 percent of the voting rights (Ibid). Högfeldt (2005) suggests that a reason for Sweden’s high level of concentrated ownership can be connected to firms’ historic dependence on external equity financing and that political decisions have determined how large that dependence has been. Mechanisms such as dual-class shares and pyramiding have received continuous political support and led to a decrease in firms’ willingness to use external equity financing (Ibid). Instead, controlling owners want to retain their control without having to compensate new external shareholders for the related agency costs.

The extensive use of dual-class shares is one of the reasons La Porta et al. (1998) view Sweden as comparatively weak in investor protection. The use of such anti-shareholder devices tends to expose minority shareholders to the risk of being expropriated. Ownership groups with large block holdings in affiliated companies are common in Sweden. A few families own substantial voting rights in other companies through investment companies managed as closed-end

(14)

funds (Doukas et al., 2002). The control is often obtained through dual-class shares and pyramidal structures. These family-controlled spheres emerged from Ivar Kreuger's fallen empire when commercial banks obtained control in firms on the brink of bankruptcy. As regulations came to prohibit banks from this type of ownership, holding companies were created to keep control of the firms (Ibid). A majority of these powerful shareholders have maintained their controlling positions in the investment companies.

2.5 Literature Review and Hypothesis Development

The potential conflict of interest between minority and controlling shareholders is argued by some to be most relevant in companies controlled by families (Cronqvist and Nilsson, 2003;

Villalonga and Amit, 2006). The rationale being that controlling families often are directly involved in management which increases the risk of expropriation. Faccio and Lang (2002) show that a member of the controlling family is part of management in over 70 percent of Swedish family firms. Dividends can therefore be used to reduce controlling families’ capacity to extract private benefits at the expense of minority shareholders (Pindado et al., 2012).

However, there are also contrasting views suggesting that family control is positive in terms of agency conflicts (Anderson and Reeb, 2003b; Gugler, 2003). The controlling family’s long- term commitment might reduce the risk of expropriation and therefore the need for payouts as a governance tool.

Empirical findings on dividend payouts in family firms provide mixed results.

Faccio et al. (2001) argue that individual investors are concerned with the possible expropriation when firms have a controlling owner. In response, companies distribute more capital as dividends in order to alleviate the concerns. Pindado et al. (2012) show that family firms in the Eurozone pay higher dividends compared to non-family firms. Their findings suggest that family firms use higher and more stable dividend ratios to alleviate agency conflicts. Maury and Pajuste (2002) investigate dividends in Finnish family firms, one of the countries in the study of Pindado et al. (2012). However, Maury and Pajuste do not find any significant difference in payouts between the groups. Isakov and Weisskopf (2015) show that Swiss family firms on average distribute a higher portion of their earnings as dividends compared to non-family firms. The authors suggest that the higher dividends paid reduce available cash and therefore limit the risk of inefficient use by the controlling family. In line with their reasoning, Setia-Atmaja et al. (2009) find a positive relationship between family

(15)

control and dividend yield in Australia, suggesting that dividends are used to prevent expropriation. Yoshikawa and Rasheed (2010) argue that family owners may use high payouts to build a reputation of fair treatment of minority investors. La Porta et al. (2000) state that treating minority investors fairly is even more important in countries with weak legal protection of minority shareholders.

In contrast, Attig et al. (2016) find that family firms pay lower dividends in East Asia. The authors argue that family control is connected to higher agency costs, and the lower dividends are therefore interpreted as expropriation of minority shareholders. Lower dividend ratios are also reported by Gugler (2003) in Austrian family firms. However, he argues that low dividend ratios in family firms might be connected to a substitution mechanism. Instead of worsening the agency conflict, family control improves the corporate governance and reduces the need for dividends as a governance tool. Pindado et al. (2012) addresses the view of family control as a corporate governance mechanism. They state that it is not clear whether ownership structure and dividends are complements or substitutes in reducing agency costs in family firms. Previous studies also suggest that controlling families may be more concerned with the long-term survival of the firm, which translates into decisions with a strong focus on long-term growth and avoidance of short-term threats to financial stability (Lins et al., 2013; Miller et al., 2008).

Anderson and Reeb (2003b) argue that their findings indicate that family ownership is not detrimental to minority shareholders. Further, the authors argue that when outsiders are able to engage in effective monitoring of the controlling family, the agency problems are less apparent.

In another study, Anderson and Reeb (2003a) even state that family ownership reduces conflicts of interest between the controlling family and outside investors rather than worsening the agency problems.

The view of family ownership in relation to corporate governance, as well as the relationship between family control and payout ratios varies. By adopting the view of family ownership leading to larger agency conflicts, and that payouts can be used as a tool to mitigate them, the expectation is that family firms have higher payout ratios than non-family firms. The intermediate shareholder protection in Sweden together with the strong prevalence of family- controlled companies support the expected relationship. The first hypothesis is therefore formulated as:

Hypothesis 1: Family firms have higher payouts than non-family firms.

(16)

Minority shareholders’ fear of being expropriated can be fueled by the ownership structure used by controlling families. Adoption of anti-shareholder devices that violate the traditional one- share-one-vote system can increase agency problems (Renneboog and Szilagyi, 2015). As the controlling shareholder hold voting rights in excess of cash flow rights, the risk of expropriation increases (Shleifer and Vishny, 1997). The wedge between ownership and control means that the owner only internalizes a portion of the negative valuation consequences following poor financial decisions (Cronqvist and Nilsson, 2003). At the same time, the controlling owner can fully enjoy the private gains which increases incentives for personal wealth extraction (Ibid).

Thus, the need for dividends as a governance mechanism increases although controlling owners might have reasons for laxing payouts. Lemon and Lins (2003) argue that the ownership structure, and more specifically family ownership together with the separation of voting and cash flow rights, is important for understanding incentives of controlling owners to expropriate minority shareholders.

Jordan et al. (2014) find that American companies using dual-class shares on average pay higher dividends than firms without dispersion between voting and cash flow rights. The authors argue that companies with a wedge use higher dividends as a pre-commitment signal to mitigate concerns of expropriation. Faccio et al. (2001) investigate dividend ratios of firms with a controlling shareholder. They find that deviations between ownership and control, in companies that are tightly affiliated within a business group, increase dividend ratios. The authors theorize that owners might do so to decrease concerns of expropriation. Related to this explanation, Hu and Kumar (2004) suggest that entrenched managers might accept paying higher dividends to avoid being sanctioned by outsiders. Furthermore, Jog et al. (2010) find that companies in Canada with restricted voting shares pay higher dividends to compensate for the risk of expropriation.

Pindado et al. (2012) hypothesize that family firms, in which the controlling shareholder has voting rights in excess of cash flow rights, pay higher dividends to mitigate larger agency conflicts. However, their findings show that companies using such control enhancing mechanisms pay lower dividends. They argue that controlling owners in these companies take advantage of their wedge by expropriating minority shareholders. Gugler and Yurtoglu (2003) find support for the expropriation theory in the German setting. They call for better governance systems and more transparency as deviations from the one-share-one-vote structure tend to lower dividends paid. Renneboog and Szilagyi (2015) present similar findings in their study of

(17)

Dutch companies. The stakeholder-oriented governance regime used in the Netherlands often restrict minority shareholders’ rights, for example through dual-class shares. The authors show that the use of such devices by controlling shareholders tend to lower the propensity to pay dividends.

As previously mentioned, Faccio and Lang (2002) show that Sweden is the country in Western Europe where dual-class shares are most frequently used. On average, less than 10 percent of the capital is enough to control 20 percent of the voting rights. Cronqvist and Nilsson (2003) study the use of a wedge between control and ownership on a sample of Swedish firms. They argue that the agency costs are highest in family firms using dual-class shares. The common usage of dual-class shares alongside the relatively large fraction of family firms make Sweden an appropriate setting for investigating whether separation of ownership and control in family firms further affect payout levels.

Agency theory suggests that divergence between ownership and control should increase the conflict between controlling and minority shareholders. By combining family firms and dual- class shares, the agency problems might be even more prevalent than in the first hypothesis.

Raising payouts might be one way of alleviating minority shareholders’ concerns. Even though empirical findings provide mixed results, being consistent in the view of payouts’ role in mitigating agency conflicts leads to the second hypothesis being formulated as:

Hypothesis 2: Family firms in which the controlling family owns voting rights in excess of cash flow rights have higher payouts than non-family firms.

Lintner (1956) elaborate on managers’ reluctance to cut dividends in his study on dividend policy. One of the primary reasons discussed by Lintner is that managers are aware of shareholders’ preferences of stable dividend levels. Therefore, companies partially adjust their payouts towards a target dividend level. As markets put a premium on companies with stable payouts, managers avoid making changes in dividends that might have to be reversed in subsequent periods. Lintner’s line of reasoning is supported by the findings of Brav et al. (2005) almost 50 years later. Several studies focus on determinants of dividend smoothing behavior, and some of these include agency conflicts as one of the components. According to Michaely and Roberts (2012), privately owned corporations are much less likely to smooth dividends and show a higher propensity to both cut and omit their payouts. They focus on the agency conflict

(18)

between managers and owners and argue that the direct involvement of family members in privately held corporations mitigates agency conflicts. Therefore, the authors claim that dividend smoothing is less important in these firms as the ownership structure manages free cash flow problems. Gugler (2003) find that family firms in Austria tend to smooth their dividends less than non-family firms and argue that the direct involvement of controlling families in management reduces the need for dividend smoothing.

However, as previously mentioned the most salient agency problem in family firms is the one between controlling and minority shareholders. In family-controlled companies, minority shareholders face the risk of having their wealth expropriated by controlling shareholders.

Therefore, family firms could be more prone to pay high and stable dividends to please minority shareholders and mitigate their concerns (Leary and Michaely, 2011). Leary and Michaely argue that the severity of agency conflicts affect dividend smoothing and that firms with a dispersed ownership structure smooth less due to lower agency costs. Pindado et al. (2012) also present empirical evidence for family firms’ higher propensity to smooth dividends. They argue that controlling families use stable dividends to mitigate the agency conflict between the controlling family and minority shareholders. Due to previous results on dividend smoothing, the predicted outcome is that family firms pay more stable dividends and have a slower speed of adjustment towards their target ratio. Leary and Michaely also find that companies with higher dividend ratios tend to smooth more. A logical continuation of the first hypothesis therefore leads to the third hypothesis being formulated as:

Hypothesis 3: Family firms have more stable payout ratios than non-family firms.

(19)

3. Method

3.1 Sample Selection

The study is based on annual data ranging from 2009 to 2017. The sample period is chosen in an attempt to capture contemporary payout decisions under normal business conditions while avoiding the epicenter of the financial crisis. Inclusion in the sample requires that companies pay dividends and are traded on the Stockholm Stock Exchange (SSE) as a Mid- or Large Cap company. The model used to determine dividend smoothing requires that companies pay dividends in at least six consecutive years in the sample period. Focusing on larger firms is therefore appropriate as these, in general, are more consistent in their payouts. Cross-listed firms in which the primary listing is not on the SSE are excluded from the sample. In line with previous studies on payout decisions, financial firms are also excluded from the sample (Attig et al., 2016; Isakov and Weisskopf, 2015; Pindado et al., 2012). Financial companies are tightly regulated which affects both dividend decisions and payout variables.

Data on companies’ ownership structure is hand-collected from their respective annual report.

An advantage of obtaining information on Swedish firms is that their annual reports are known to be more informative and transparent compared to firms in other countries (La Porta et al., 1999). Classifying a company as being family-controlled or not also requires follow-up on the owners presented in the annual reports. Manually obtaining detailed ownership data on the ultimate owners for each year in this manner is necessary for accurately classifying companies as family or non-family firms. Financial data related to the performance and payouts are extracted from Thomson Reuters Eikon and all statistical estimations are performed in Stata (version 14). An important consideration is the treatment of the investment companies managed as closed-end funds on the SSE. To enhance the comparability with previous studies, investment companies are managed as any other company in the sample (Attig et al., 2016;

Faccio and Lang, 2002; Pindado et al., 2012). A few closed-end funds founded by families hold significant parts in several of the largest companies in Sweden, and thus have the ability to influence payout decisions. Although investment companies might operate differently than firms considered to be traditional family firms (Schmuck and Hamberg, 2019), they are still relevant for this study. These firms are included to enhance the sample’s representativeness of the Swedish market.

(20)

3.2 Classification of Family Firms

One crucial consideration when investigating family firms is establishing the classification scheme determining which companies are considered to be family-controlled. Previous studies adopt different ways of separating family firms from non-family firms and there is no general definition available. In this study, four criteria needs to be fulfilled for a company to be classified as family-controlled. These are set up to reflect the classical view of family firms as being long-term oriented and entrenched in the company instead of merely wealthy individuals investing in listed companies. More specifically, the criteria used for firm classification are: (1) The ultimate owners of a company should be a family or an individual. In accordance with prior studies, (2) 20 percent of the voting rights is used as threshold for control (Faccio and Lang, 2002; Isakov and Weisskopf, 2015; La Porta et al., 1999). Although it could be seen as a modest level of control, there is a widely accepted view that low attendance in annual general meetings allow owners of stakes around 20 percent to affect company policies such as dividend decisions (Isakov and Weisskopf, 2015). In cases where several shareholders have stakes above 20 percent, the owner with the largest holding is considered to be in control. To further ensure the owners ability to influence decisions, (3) companies must have a family member on the board of directors or in the role as acting CEO to be considered family-controlled. And finally, (4) companies must fulfill all of the three aforementioned criteria during the whole period to be classified as a family firm.

If a company is controlled through a pyramid structure, the capital rights are calculated as the product of the different ownerships (Faccio and Lang, 2002). Making these calculations requires available ownership data for all companies in the pyramid. If the ultimate owner is not a listed company and detailed ownership data is unavailable, their stake is instead assumed to be 100 percent of the figure presented in the annual report of the listed firm. The level of control is determined by the weakest link in the chain of control. In cases where the ultimate owner cannot be determined, it is classified as a family firm. As described by Faccio and Lang (2002), it is unlikely that an unlisted firm is controlled by a widely held corporation, the state or a widely held financial institution. This leaves family-controlled companies as the most likely type of ownership structure for these unlisted entities (Claessens et al., 2000; Faccio and Lang, 2002). This stance is also adopted by Pindado et al. (2012) who conduct robustness checks to investigate whether the definition of family firms affects their empirical findings. The outcome shows that their results are not driven by unlisted companies assumed to be family-controlled.

(21)

An example of the classification process is described below as the ownership structure of Beijer Ref is outlined. The control and ownership stakes are denoted C and O respectively.

Figure 1. Ownership structure in Beijer Ref.

As depicted in Figure 1, the ultimate owners of Beijer Ref according to the framework used in this study is the American company Carrier. As Carrier exceeds the 20 percent threshold and there are no family ties between the other three large shareholders, Carrier is the largest owner of Beijer Ref. To determine whether Beijer Ref should be classified as a family firm or not, the ultimate owners of Carrier needs to be decided. Carrier is listed on the NYSE and do not have any controlling owners since no shareholder own more than 20 percent of the votes. Therefore, Beijer Ref is classified as a non-family firm in this study. For Beijer Ref to be classified as a family firm, Carrier should have been controlled by a family who also should have had a member on the board of directors in Beijer Ref during the whole sample period.

3.3 Summary Statistics

Panel A in Table 1 presents some ownership characteristics of the sample. As shown, a total of 1019 firm-year observations constitute the basis for the primary analysis. The sample is divided into 539 non-family firm observations and 480 observations of family firms. Panel A also depicts the average percentage of ownership (cash flow rights) and control (voting rights) for the largest owners in both family and non-family firms. As reported in the table, the average percentage of voting rights for the largest owner of a family firm is 47.7 percent while the equivalent for non-family firms is 23.1 percent. The average percentage of cash flow rights for the largest owner in family firms is 31 percent while the largest owners of non-family firms on average hold 17.8 percent of the cash flow rights. The average voting and cash flow rights show the common use of dual-class shares in the investigated sample as there is a clear wedge

Beijer Ref

Carrier

C: 32.5% O: 40%

State Street Corporation

C: 11.3% O: 11.3%

Vanguard Group

C: 8.2% O: 8.2%

BlackRock

C: 6.8% O: 6.8%

Joen Magnusson

C: 14.5% O: 2.6%

Per Bertland

C: 11% O: 2.1%

Peter Jessen Jurgenssen

C: 7.8% O: 1.3%

(22)

between ownership and control. Furthermore, the figures reported in Panel A also show that owners of family firms on average hold larger stakes, both in terms of voting and cash flow rights. As depicted in Panel B, industrials is the most common industry classification and make up almost a third of the firms in the sample. Some differences in ownership is observed among the industry classifications. For example, investment companies are usually family-controlled while non-family firms dominate the communication service and information technology classifications.

Table 1. Summary statistics

This table presents summary statistics of the sample. Panel A illustrates the number of firm observations for each year by ownership classifications while Panel B depicts a breakdown of the sample by industry categorization according to the GICS taxonomy. Obs denotes the number of observations for each period. Family and non-family specifies the number of observations classified as family and non-family firms respectively. The columns for votes and cash flow in Panel A displays the average percentage of voting and cash flow rights of the largest owner in family and non-family firms for each period. The fraction column in Panel B summarizes the percentage of firms in the sample that belongs to each GICS category.

PANEL A: Summary Statistics

Average Voting (%) Average Cash flow (%) Period Obs Family Non-family Family Non-family Family Non-family

2010 107 55 52 49.1 23.6 31.8 17.5

2011 118 57 61 49.0 23.6 32.0 17.7

2012 117 56 61 49.5 23.5 31.8 17.7

2013 116 55 61 49.5 24.1 32.2 18.4

2014 122 57 65 48.1 23.3 30.9 18.1

2015 139 65 74 46.6 22.7 29.8 18.0

2016 153 69 84 45.6 22.2 30.3 17.8

2017 147 66 81 44.5 21.5 29.5 17.2

TOTAL 1019 480 539 47.7 23.1 31.0 17.8

PANEL B: Industry Categorization

Industry No. firms Fraction (%) Family Non-family

COMMUNICATION SERVICES 6 3.7 0 6

CONSUMER DISCRETIONARY 30 18.5 12 18

CONSUMER STAPLES 8 4.9 4 4

ENERGY 2 1.2 1 1

HEALTH CARE 13 8.0 7 6

INDUSTRIALS 49 30.2 21 28

INFORMATION TECHNOLOGY 15 9.3 3 12

INVESTMENTS 12 7.4 10 2

MATERIALS 7 4.3 2 5

REAL ESTATE 20 12.3 11 9

TOTAL 162 100.0 71 91

(23)

3.4 Models

3.4.1 Payout Levels

Multiple regressions are used to investigate the first and second hypothesis of the study. In accordance with previous studies, clustered robust standard errors based on firms are used to calculate the t-statistics to account for heteroscedasticity in the sample (Attig et al., 2016;

Isakov and Weisskopf, 2015). In model (1), payouts are explained by a dummy and a set of control variables as follows:

𝑃𝐴𝑌𝑂𝑈𝑇_𝑖𝑡 = 𝛼₀+ 𝛽₁𝐹𝐴𝑀𝐼𝐿𝑌_𝑖𝑡+ 𝛽₂𝑇𝐴_𝑖𝑡+ 𝛽₃𝑇𝑄_𝑖𝑡+ 𝛽₄𝐷/𝐸_𝑖𝑡+ 𝛽₅𝑁𝐼_𝑖𝑡+ 𝜀_𝑖𝑡, (1)

where the dependent variable is the total payout ratio (PAYOUT). The dummy variable FAMILY equals 1 for family firms and 0 otherwise, TA is the natural logarithm of total assets, Tobin’s Q (TQ) is the market value divided by the book value of assets, D/E is the debt-to-equity ratio and NI is net income scaled by total assets.

The dependent variable includes dividends, share repurchases and compulsory share redemptions to capture all payouts to shareholders. In accordance with Pindado et al. (2012), the dependent variable is scaled by total assets. One possible drawback of scaling payouts by assets is that differences among firms and industries in capital intensity could affect the outcome. Some studies use earnings to scale payouts, but assets are chosen since earnings more easily can be manipulated and affected by accounting procedures. Furthermore, firms can maintain payouts despite negative earnings which causes difficulties in how to treat their payout ratios. A common approach is to treat negative payout ratios as a ratio of 100 percent since these companies distribute more than their earnings. Comparing these observations to the years in which earnings were not negative, or to other firms with positive earnings, could be misleading.

The dummy for family firms is included to capture the effect of family ownership on payout levels. Total assets is included to control for the effect of company size on payout ratios as previous studies indicate that larger firms have higher payout ratios (Brockman and Unlu, 2009;

Shao et al., 2010). Tobin’s Q is included to check for companies’ growth opportunities (Attig et al., 2016; Isakov and Weisskopf, 2015; Pindado et al., 2012). Higher Tobin’s Q indicate better growth opportunities. The payout ratio is therefore predicted to be negatively related to Tobin’s Q as more growth opportunities could be connected to more investments, thus leaving

(24)

less money available for payouts. A higher level of debt is predicted to decrease payout as debt can serve as a substitution to dividends in mitigating agency problems (Setia-Atmaja et al., 2009). Net income is used as a control variable to account for the effect of earnings on the payout ratio (Pindado et al., 2012). A higher level of earnings is predicted to increase the level of payouts as more money is available for distribution to shareholders.

Some studies use a Tobit regression model instead of an ordinary least squares (OLS) to handle skewness in the dependent variable (Attig et al., 2016; Isakov and Weisskopf, 2015). If non- paying firms are included in the sample, the distribution curve of the payout variable is skewed towards the left hand side. Tobit models are more suitable to handle censoring of large amounts of zero values. However, the choice of using OLS over Tobit is motivated by the exclusion of non-paying firms in the sample of this study. Thus, payout ratios of zero do not affect the distribution curve of the dependent variable.

The second hypothesis is tested through two multiple regression models similar to the one used in hypothesis 1. Instead of using a dummy for family firms, model (2) includes a dummy that differentiates family firms with a wedge between voting and cash flow rights from non-family firms. Thus, family firms in which the controlling owner do not have voting rights in excess of cash flow rights are excluded in this model. More specifically, model (2) is constructed as follows:

𝑃𝐴𝑌𝑂𝑈𝑇_𝑖𝑡 = 𝛼₀+ 𝛽₁𝑊𝐸𝐷𝐺𝐸_𝑖𝑡 + 𝛽₂𝑇𝐴_𝑖𝑡 + 𝛽₃𝑇𝑄_𝑖𝑡 + 𝛽₄𝐷/𝐸_𝑖𝑡+ 𝛽₅𝑁𝐼_𝑖𝑡 + 𝜀_𝑖𝑡, (2)

where the dependent variable PAYOUT is explained by the dummy WEDGE and the same set of control variables. The dummy WEDGE equals 1 for family firms in which the owner has voting rights in excess of cash flow rights and 0 for non-family firms. The predicted relation between the payout ratio and WEDGE is positive. That is, payouts are expected to be higher for family firms with a wedge between ownership and control to mitigate larger agency problems.

The dispersion between ownership and control in relation to agency costs is further investigated in model (3). If control enhancing mechanisms such as dual-class shares influence agency conflicts, it is possible that there are differences in payouts within family firms as a group. The dummy FAMILYWEDGE only includes family firms and differentiates between family firms

(25)

with a wedge and family firms without a wedge. Model (3) enables an extended analysis of payouts in family firms and is designed as:

𝑃𝐴𝑌𝑂𝑈𝑇_𝑖𝑡 = 𝛼₀+ 𝛽₁𝐹𝐴𝑀𝐼𝐿𝑌𝑊𝐸𝐷𝐺𝐸_𝑖𝑡+ 𝛽₂𝑇𝐴_𝑖𝑡 + 𝛽₃𝑇𝑄_𝑖𝑡 + 𝛽₄𝐷/𝐸_𝑖𝑡+ 𝛽₅𝑁𝐼_𝑖𝑡 + 𝜀_𝑖𝑡, (3)

where the dependent variable PAYOUT is explained by the dummy FAMILYWEDGE and the same set of control variables as in model (1) and (2). The dummy FAMILYWEDGE equals 1 for family firms in which the family has voting rights in excess of cash flow rights and 0 for family firms without a wedge. In comparison to model (2), non-family firms are excluded to better understand potential drivers of payouts in family firms. The expected positive coefficient for the dummy variable would indicate that family firms with a wedge have higher payouts to mitigate minority shareholders’ concerns of being expropriated.

3.4.2 Speed of Adjustment

Lintner’s (1956) model of partial adjustment provides the basis for the methodology used to investigate the third hypothesis. As described by Lintner, managers are reluctant to cut dividends and prefer to keep payouts at stable levels. Companies therefore use a target payout ratio (TPR) related to the profits earned. The pace at which firms adapt their dividends to their current profits is determined by their speed of adjustment (SOA). Since companies tend to adjust their dividends proportionally less than profits fluctuate, the pace at which companies move toward the target is in general 0 < SOA < 1. A lower value on the SOA coefficient indicates more dividend smoothing as the company keeps dividends stable despite possible fluctuations in earnings. In contrast, a higher value of SOA means that the firm is quick to adjust dividend payments to changes in earnings. According to Lintner, a company’s change in dividends from one year to the next is determined by:

𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑_𝑖𝑡 = 𝛼_𝑖+ 𝑆𝑂𝐴(𝑇𝑃𝑅_𝑖 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠_𝑖𝑡 − 𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑_{𝑖(𝑡−1)}) + 𝜀_𝑖𝑡, (4)

where the dependent variable is the change in actual dividends from year t-1 to year t. The intercept (α_𝑖) will in general be positive to denote the reluctance to cut dividends. The SOA- coefficient is estimated by the difference between the target dividend payout and the dividends paid in the previous period. The target dividend in year t is computed as the TPR multiplied with the current earnings. The error term (𝜀_𝑖𝑡) represents the divergence between the observed

(26)

and expected change in dividends. One issue with Lintner’s (1956) model is determining the TPR. Pindado et al. (2012) state that the target is not directly observable by researchers.

Previous studies present different approaches in how to treat the target measure. Fama and Babiak (1968) modify Lintner’s model by suppressing the constant term and estimating the SOA and TPR simultaneously as follows:

𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑_𝑖𝑡 = 𝛼_𝑖 + 𝛽₁(𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑_{𝑖(𝑡−1)}) + 𝛽₂(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠_𝑖𝑡) + 𝜀_𝑖𝑡, (5)

where the SOA is calculated as -𝛽̂₁ and the TPR is given by (- 𝛽̂₂/𝛽̂₁). However, estimating both the SOA and the TPR simultaneously creates biased results due to inflated standard errors caused by small SOA coefficients (Leary and Michaely, 2011). Instead, Leary and Michaely present a two-step approach in which median payout ratios are computed and used as a measure for the target. Their results show that investigating dividend smoothing with the two-step approach significantly increases the precision of the SOA estimate. Following their approach, median payout ratios for all firms are computed and used as a proxy for each company’s TPR. The firm specific effect on payouts () is then estimated by the difference between the target and the actual payouts for firm i in year t. This is done by running OLS regressions for each company as follows:

𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑_𝑖𝑡 = 𝛼_𝑖+  𝐷𝑒𝑣_𝑖𝑡+ 𝜀_𝑖𝑡, (6)

in which the deviation (𝐷𝑒𝑣_𝑖𝑡) is computed as:

𝐷𝑒𝑣_𝑖𝑡 = 𝑇𝑃𝑅_𝑖 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠_𝑖𝑡− 𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑_𝑖𝑡. (7)

In line with previous studies, the estimated  in model (6) is interpreted as a measure for the SOA (Baker and De Ridder, 2018; Leary and Michaely, 2011). Thus, model (6) is used to investigate the third hypothesis whether family firms smooth their payouts more than non- family firms. To further strengthen the reliability of the SOA measure, inclusion in the analysis for the third hypothesis requires that companies are dividend payers in at least six consecutive years during the investigated period.

(27)

4. Results and Analysis

4.1 Descriptive Statistics

Table 2 presents main descriptive statistics for the sample. The means, medians and quartiles for key variables are reported in Panel A. All variables are winsorized at 1 percent to reduce the effect of outliers. Still, the mean for Tobin’s Q (TQ) is well above the median, suggesting that the sample is positively skewed and the maximum value indicate a long tail of the distribution curve for the variable.

Table 2. Descriptive statistics

Panel A presents descriptive statistics for the variables included in the regression model and Panel B reports correlation coefficients. PAYOUT is total payouts scaled by total assets. TA is the natural logarithm of total assets and TQ (Tobin’s Q) is the company market value divided by the book value of total assets. D/E is the ratio of debt-to-equity while NI is the ratio of net income to total assets. FAMILY is a dummy variable which equals 1 if the firm is classified as a family firm and 0 otherwise.

All of the variables are presented in more detail in Appendices A-B. *, ** and *** denotes statistical significance on 10%, 5%

and 1% level.

PANEL A: Descriptive Statistics

Obs Mean Q1 Median Q3 Std. Dev. Min Max

PAYOUT 1 019 0.050 0.018 0.032 0.052 0.057 0.005 0.314

TA 1 019 22.673 21.388 22.614 24.100 1.736 18.453 26.761

TQ 1 019 1.546 0.627 0.989 1.700 2.572 0.321 6.179

D/E 1 019 0.613 0.138 0.460 0.916 0.590 0.000 2.418

NI 1 019 0.080 0.040 0.066 0.103 0.082 -0.613 0.950

PANEL B: Correlation Matrix

PAYOUT FAMILY TA TQ D/E NI

PAYOUT 1.000

FAMILY 0.097*** 1.000

TA -0.291*** -0.092*** 1.000

TQ 0.550*** -0.071*** -0.421*** 1.000

D/E -0.355*** -0.112*** 0.324*** -0.349*** 1.000

NI 0.484*** 0.146*** -0.211*** 0.446*** -0.320*** 1.000

Panel B of Table 2 shows the correlation coefficients for the variables. PAYOUT and Size (TA) are negatively correlated which means that larger firms have lower payouts. However, it might be explained by the fact that PAYOUT is scaled by total assets. Thus, the payouts of firms with larger asset bases are divided by bigger denominators. There is a positive correlation between PAYOUT and net income (NI) and a negative correlation between PAYOUT and debt (D/E), which are both expected. The correlation coefficient between PAYOUT and FAMILY is positive but not strong. Among the correlation coefficients between the independent variables none exceed 0.5. This is reassuring since multicollinearity can be an issue when independent variables are highly correlated. Multicollinearity do not bias the model as a whole but it can

(28)

bias coefficient estimates leading to invalid interpretations of individual predictors. VIF tests are performed to check for the potential issue, these show that there are no signs of multicollinearity between the variables².

4.2 Univariate Analysis

Table 3 presents results from the univariate analysis of the sample obtained by testing differences in means through t-tests³. Panel A shows the results when the sample is grouped as family and non-family firms. The difference in means for the dependent variable is significant at the 1 percent level, suggesting that family firms have higher payout ratios than non-family firms. The result provides initial support for the first hypothesis that family firms distribute more capital to shareholders, which is in line with several previous studies (Isakov and Weisskopf, 2015; Pindado et al., 2012; Setia-Atmaja et al., 2009). The univariate tests performed on the control variables yields significant results as well. Family firms tend to have lower debt ratios while also being more profitable in terms of net income divided by total assets.

Worth noting is also that the market seems to put a premium on family firms relative to non- family firms as the former exhibit significantly higher Tobin’s Q. Overall the results in Panel A illustrate that family firms differ from non-family firms across all dimensions considered.

However, univariate tests do not control for the effect of differences in firm level characteristics. Although the payout ratio seems to be higher in family firms, further analysis is required.

Panel B of Table 3 shows additional univariate results when dispersions between voting and cash flow rights in family firms are considered. When testing the second hypothesis, family firms with a wedge is compared to two groups. The first group is made up of non-family firms, and the second group consists of the family firms with no difference between voting rights and cash flow rights. The results of the univariate tests in column (6) show that family firms with a wedge exhibit higher payout ratios compared to non-family firms. This result differs from the univariate findings of Pindado et al. (2012), who show that dividends are lower in family firms with a wedge. As reported in column (7), no difference in mean payout ratio can be

2 The Variance Inflation Factor (VIF) tests are presented in Table 8 in Appendix C.

3 In addition to parametric t-tests, non-parametric Wilcoxon rank-sum (Mann-Whitney) tests are performed. This enables a relaxation of the assumption of normality of the underlying distribution. The results are presented in Table 9 in Appendix C. The results are consistent with the parametric analysis, with the exception that the difference in payout is non-significant.

Payout policy in family firms