The impact of family ownership on dividend payout policy

(1)

The impact of family ownership

on dividend payout policy

– An examination on the Swedish context

Master’s Thesis 15 credits

Department of Business Studies

Uppsala University

Spring Semester of 2020

Date of Submission: 2020-06-03

Fanny Lundvall

Marcus Wibom

(2)

Abstract

This study investigates whether family ownership impacts firms’ dividend payout policies by examining firms publicly listed on the Stockholm Stock Exchange (OMX Stockholm) during the years 2013–2018 (1,363 firm-year observations). The investigation is made by performing multiple regression analyses including the dependent variable DIVIDEND PAYOUT. The findings reveal that family firms distribute higher dividend payouts than non-family firms, suggesting that dividends are used as a corporate governance mechanism to mitigate agency problems. Family firms without a second blockholder present have the highest dividends. A family second blockholder appears to collude with the controlling family resulting in lower dividends. A separation between ownership and control results in higher dividends as it implies a worse corporate governance structure. In sum, the results imply that family ownership positively impacts firms’ dividend payout policies in Sweden.

(3)

1. Introduction

In contrast to Berle and Means’ (1932) idea of widely held firms, subsequent studies have shown that firms in most countries are controlled by large blockholders1 (e.g. La Porta, Lopez-de-Silanes & Shleifer, 1999; Claessens, Djankov & Lang, 2000; Faccio & Lang, 2002). By holding a substantial amount of a firm’s cash-flow rights and the attached voting rights, these blockholders have a considerable impact on the firm’s decisions, for example concerning dividend payout policy (Shleifer & Vishny, 1997; Isakov & Weisskopf, 2015). High ownership concentration, i.e. that one or a few blockholders hold the majority of a firm’s shares, is found to be a response to a country’s level of investor protection (La Porta, Lopez-de-Silanes, Shleifer & Vishny, 1998; La Porta et al., 1999). In a setting of strong investor protection, controlling blockholders have less fear of being expropriated if they would lose control and consequently tend to hold fewer voting rights. In a setting of weak investor protection, controlling blockholders consider it too costly to relinquish the private benefits of control and hence do anything to keep control. In this setting, controlling blockholders tend to hold more voting rights(La Porta et al., 1999).

The presence of blockholders implies that agency problems between controlling and non-controlling shareholders are more common than the ‘classical’ agency problems between managers and shareholders (Faccio, Lang & Young, 2001; Cronqvist & Nilsson, 2003). A controlling blockholder may have incentives to monitor management and to align the interests of all shareholders (alignment effect) (Claessens, Djankov, Fan & Lang, 2002). However, it may also have incentives to entrench itself by extracting private benefits at the cost of non-controlling shareholders (entrenchment effect) (Ibid). When investigating which of these opposing effects that prevail, the identity of the controlling blockholder has been emphasized (Isakov & Weisskopf, 2014). There are also indications that the presence and identity of the second largest blockholder may have an impact on which effect that prevails (Ramli, 2010; Pindado, Requejo & de la Torre, 2012; Renneboog & Szilagyi, 2020). Either, the second largest blockholder may encourage the controlling blockholder to align the shareholders’ interests, or, it may engage in the expropriating activities (Pindado et al., 2012; Isakov & Weisskopf, 2015).

1_{The term ‘blockholder’ is commonly used in prior studies to denote a shareholder that holds a substantial amount}

(5)

4 Particular attention has been raised to family ownership2 related to the aforementioned agency problems (Runesson, Samani & Marton, 2018). Families constitute the most common blockholder category in many countries, for example in Sweden (La Porta et al., 1999; Claessens et al., 2000; Faccio & Lang, 2002). While reviewing prior studies, the impact of family ownership and the severity of agency problems in these firms appear somewhat ambiguous. The findings of Cronqvist and Nilsson (2003) and Croci, Doukas, and Gonenc (2011) imply that decisions in family firms are influenced by control motives. These motives appear related to the possibility of extracting private benefits by utilization of power (Ibid). Family firms are more likely than non-family firms to use control enhancing mechanisms, as dual-class shares (Cronqvist & Nilsson, 20033). These instruments imply a separation of ownership (cash-flow rights) and control (voting rights) allowing shareholders to exercise more control over the firm with less invested capital. The enhanced control increases the ability to monitor management. It also increases the risk of agency problems as the controlling blockholder may get enticed to waste corporate resources and expropriate non-controlling shareholders (Cronqvist & Nilsson, 2003; Pindado et al., 2012; Isakov & Weisskopf, 2014). Furthermore, family firms are unwilling to issue equity as it dilutes ownership, consequently preferring debt financing to stay in control (Croci et al., 2011).

Beyond the control motives, decisions in family firms are influenced by trust motives (Pindado et al., 2012; Isakov & Weisskopf, 2014, 2015). Controlling families seem more inclined than non-families to generate trust among non-controlling shareholders and build a reputation of treating them well (Isakov & Weisskopf, 2014, 2015). This appear related to controlling families’ close ties to their business. For example, they usually have a significant proportion of their wealth invested in the firm (Isakov & Weisskopf, 2015) and consider it part of their identity (Isakov & Weisskopf, 2014). These close ties suggest a relatively low risk of expropriation of non-controlling shareholders. Accordingly, Pindado et al. (2012) and Isakov and Weisskopf (2015) show that controlling families use dividends as a signal of commitment not to expropriate non-controlling shareholders. Dividend payouts lessen the amounts of cash

2_{A family firm in this study is a firm ultimately owned by a shareholder holding at least 20% of the firm’s total}

voting rights and is a family or an individual. Section 3.3.2 provides a more detailed description of the family firm definition.

(6)

5 on hand, hence the controlling family is considered voluntarily emitting control in exchange for building trust (Pindado et al., 2012; Isakov & Weisskopf, 2015).

The above-mentioned motives of controlling families suggest that agency problems differ in family firms and non-family firms. Consequently, it is expected that important corporate governance decisions differ between these firms. An example of such decision is dividend payout policy (Bhattacharya, 1979). Dividends constitute a common means of disbursing capital to shareholders and could signal expectations of a firm’s future profits to outsiders (Bhattacharya, 1979; Dittmar, 2008). Additionally, dividends are acknowledged as an essential control function to discipline managers from misusing firm resources for self-fulfilling purposes (Easterbrook, 1984; Jensen, 1986; Dittmar, 2008; Pindado et al., 2012). Similarly, dividends can be used as a trust-generating mechanism to ease non-controlling shareholders’ concerns of expropriation by controlling blockholders (Pindado et al., 2012; Isakov & Weisskopf, 2015).

Swedish firms are characterized by highly concentrated ownership structures and are often controlled by families (La Porta et al., 1998; Faccio & Lang, 2002). The country has a weak level of investor protection (La Porta et al., 1998; Leuz, Nanda, & Wysocki, 2003), strong extralegal institutions (Holmén & Knopf, 2004) and allows the use of control enhancing mechanisms (La Porta et al., 1998; Cronqvist & Nilsson, 2003). Furthermore, controlling blockholders have a central role in corporate governance in Sweden (Swedish Corporate Governance Board, 2020). Considering the above-mentioned characteristics, the Swedish setting differs from the U.K. and the U.S.4 which most prior research on dividend payout policy has focused on (Pindado et al., 2012). The characteristics of the Swedish setting motivates an examination of family ownership and dividend payout policy on this context.

The high presence of families (Claessens et al., 2000; Faccio & Lang, 2002) and their conflicting objectives create interest to examine if the control motive or the trust motive dominate related to families’ dividend behavior. If the control motive (Cronqvist & Nilsson, 2003; Croci et al., 2011) prevails, it is expected that family firms distribute low dividends (Gugler, 2003). If the trust motive prevails, it is expected that family firms distribute high dividends (Pindado et al., 2012; Isakov & Weisskopf, 2015). By employing a Swedish context, the study extends prior studies on family ownership structure and dividend payout policy in

4_{E.g., the U.K. and the U.S. are common-law countries with strong investor protection (La Porta et al., 1998)}

(7)

6 Europe (Pindado et al., 2012). Studying a sample within a single legal jurisdiction enables an analysis of firms covered by the same legal restrictions but that voluntarily have chosen to adopt different ownership and governance structures (Cronqvist & Nilsson, 2003). The study’s contribution is to further increase knowledge regarding the impact of family ownership in Sweden, particularly considering its low protective institutional environment. From a practical perspective, the study assists investors in their expectations on dividends in family firms.

The purpose of this study is to examine the impact of family ownership on dividend payout policy in Swedish listed firms during the years 2013–2018. The aim is to investigate whether family firms differ in their dividend payout policies compared to non-family firms. Additionally, the influence of the second largest blockholder and a separation of ownership (cash-flow rights) and control (voting rights) is considered. The research question of the study is formulated as follows:

Does family ownership impact firms’ dividend payout policies?

(8)

7

2. Theoretical framework

2.1 Agency theory

The agency relationship is described by Jensen and Meckling (1976) as a contract in which a party (the principal) appoints another party (the agent) to act on his or her behalf. The agent is provided with some decision-making authority and shall perform according to the principal’s instructions. Assuming both parties are utility maximizers and act in their own best interest, there is a high risk that the agent’s actions will deviate from the interest of the principal and as to what has been assigned. Conflicts of interests between the principal and the agent may therefore cause agency problems (Ibid). Agency problems in firms are related to the separation of ownership and control (Jensen & Meckling, 1976; Shleifer & Vishny, 1997).

The classical agency problem between managers and shareholders is often referred to as the Type I agency problem (Paiva, Lourenço & Branco, 2016). Another agency problem is between controlling and non-controlling shareholders, often referred to as the Type II agency problem (Paiva et al., 2016). As firms in many countries are characterized by highly concentrated ownership structures (La Porta et al., 1998, 1999) the Type II agency problem is more prominent (Faccio et al., 2001; Cronqvist & Nilsson, 2003). It is unclear whether the presence of a controlling blockholder is beneficial or not from non-controlling blockholders’ perspective. Two counteracting effects are noted; the alignment effect and the entrenchment effect (Claessens et al., 2002). Controlling blockholders may provide enhanced monitoring of management (which reduces the free-rider problem between shareholders5) aligning shareholders’ interests. As a result of the increased monitoring, the Type I agency problem is diminished or even eliminated. Also, it may have incentives to engage in value-maximizing activities benefiting all shareholders (Cronqvist & Nilsson, 2003; Isakov & Weisskopf, 2014).

Besides aligning shareholders’ interests, controlling blockholder may get entrenched and act in their self-interest (Claessens et al., 2002). The entrenchment effect appears most present in firms where blockholders hold voting rights above their cash-flow rights using various control enhancing mechanisms (La Porta et al., 1998; Claessens et al., 2000; Faccio & Lang, 2002). Common examples of such mechanisms are cross-shareholdings, dual-class shares and

5_{The free-rider problematic is about how to successfully organize dispersed shareholders to collectively monitor}

(9)

8 pyramid structures. The power received from this deviation of rights enables controlling blockholders to exploit the situation and extract private benefits at the expense of non-controlling shareholders (Cronqvist & Nilsson, 2003; Isakov & Weisskopf, 2014). Cronqvist and Nilsson (2003) show that agency costs6 increase linearly with the potential to extract private benefits. According to the authors, restrictions in the legal system and the financial incentives of the controlling blockholder are the only limitations hindering controlling blockholders from expropriating non-controlling shareholders (Ibid).

Whether the presence of a controlling blockholder can mitigate Type I and Type II agency problems also seems to be dependent on the blockholder’s level of control (in terms of voting rights). Isakov and Weisskopf (2014) find that family presence is only beneficial (i.e. lowers agency costs) within a certain range of control (20–80%). In firms where the level of control exceeds 80%, investors perceive the family too powerful leaving them to mistrust which has a negative effect on firm value (Ibid).

Similar as the impact of a controlling blockholder on agency problems is unclear, so is the impact of the second largest blockholder. The second largest blockholder may encourage controlling blockholders to align shareholders’ interests, hence allay agency problems (Pindado et al., 2012; Isakov & Weisskopf, 2015). However, the second largest blockholder may also collude with the controlling blockholder to expropriate non-controlling shareholders, hence aggravate agency problems (Ibid).

2.2 Dividends

The use of dividends can be traced back to 1685 but is believed to have existed even long before that (Morck & Yeung, 2005). The original idea of dividends was to divide the firm’s profit among its shareholders over time, which would also attract more investors hence increase firm value (Ibid). This idea of dividends has remained over time and appears to be complemented by other functions. For example, dividends could function as a corporate governance mechanism (Easterbrook, 1984; Jensen, 1986; Dittmar, 2008; Setia-Atmaja, Tanewski & Skully, 2009; De Cesari, 2012). Holding high levels of cash could be costly and expose shareholders to involuntary risk (Jensen, 1986; Dittmar, 2008). According to Jensen’s (1986) free cash-flow theory, excess cash may induce managers to invest in unprofitable investments. This so-called ‘overinvestment problem’ is a cost for the shareholders. Disbursing

(10)

9 excess cash through dividends therefore mitigates the risk of overinvestment, forcing managers to focus on lucrative investments only (Ibid). Consequently, it lowers the risk of shareholder expropriation thus mitigates agency problems (Dittmar, 2008; De Cesari, 2012).

Dividends also appear to have a potential signaling function. Compared to share repurchases, dividends are a costlier signal to alleviate expropriation concerns as it implies a stronger future commitment to disburse capital (Pindado et al., 2012). Bhattacharya (1979) finds that initiation or a raise of dividends constitutes a signal to the market that the firm is prosperous and has high expectations on future earnings. A reason for this is management’s reluctance to cut dividends hence only initiate or increase dividends when they are certain about future earnings (Ibid).

2.3 The Swedish institutional setting

The Swedish institutional setting is characterized by relatively weak investor protection (La Porta et al., 1998; Leuz et al., 2003). La Porta et al. (1998) compare legal rules covering investor (shareholders and creditors) protection, the legal origins of those rules and the quality of enforcement in 49 countries worldwide. Shareholder protection is measured in terms of antidirector rights, for example conformity with the one-share-one-vote concept and allowance of proxy voting by email. Creditor protection is measured in terms of creditor rights, for example restrictions on reorganization and the ranking of creditors in case of distributions of proceeds. In general, La Porta et al. (1998) find that common-law countries (e.g. the U.K. and the U.S.) have stronger investor protection than civil-law countries (including French, German and Scandinavian-civil law countries).

(11)

10 as an insider economy with a less-developed stock market, concentrated ownership, weak investor rights and strong legal enforcement (Ibid).

The weak investor protection in Sweden is also to a large extent explained by the legal allowance and use of control enhancing mechanisms (La Porta et al., 1998; Cronqvist & Nilsson, 2003). The Swedish corporate governance model is mainly based on the Swedish Companies Act (Aktiebolagslagen) containing general legal rules regarding limited share companies (Runesson et al., 2018). Cronqvist and Nilsson (2003) argue that these rules are aimed at protecting blockholders by the allowance of different control enhancing mechanisms. Faccio and Lang (2002) find that Swedish firms rank top three on the use of cross-shareholdings, dual-class shares and pyramid structures. Furthermore, they find that Sweden has the lowest level of cash-flow rights needed to control (at a threshold of 20% of total voting rights) a firm among the countries in their sample (Ibid).

Although the high use of control enhancing mechanisms has a detrimental impact on corporate governance in Sweden, other factors related to the institutional environment seem to improve the protection of investors. Holmén and Knopf (2004) investigate mergers related to the corporate governance structure in Sweden. They find evidence that the poor corporate governance in the country (e.g. referring to the low level of investor protection found by La Porta et al. (1999)) can be offset by its extralegal institutions. Examples of extralegal institutions are organized product markets and organized labor (Holmén & Knopf, 2004).

2.4 Literature review and hypotheses

2.4.1 The impact of family ownership on dividend payout policy

(12)

11 have different objectives and incentives that affect their actions in the firm (Isakov & Weisskopf, 2014). According to prior studies, the impact of family ownership appears to be unclear, for instance related to dividend payout policy (Gugler, 2003; Pindado et al., 2012; Isakov & Weisskopf, 2015).

Several studies indicate that decisions made in family firms are influenced by control motives (Cronqvist & Nilsson, 2003; Croci et al., 2011; Isakov & Weisskopf, 2015). Whether these control motives negatively impact non-controlling shareholders or not may depend on the reasons why the blockholder wants to possess control. The common view seems to be that the control motives are related to the ability to extract private benefits at the non-controlling shareholders’ expense (Croci et al., 2011).

Croci et al. (2011) examine the effect of family ownership on financial policy decisions in Western Continental European firms. They show that families prefer debt financing over equity financing. The unwillingness to issue equity is related to the implicated dilution of ownership by which control is relinquished to other shareholders (Ibid). Cronqvist and Nilsson (2003) find that Swedish family firms are more likely to use control enhancing mechanisms than other blockholder categories. They also find a low use of these instruments in firms with dispersed ownership. The interpretation of their results is that there is a control motive present in firms with high ownership concentration, particularly in family firms (Ibid). Furthermore, Gugler (2003) investigates the relationship between the ownership and control structure and dividends in Austrian firms. The author concludes that family firms tend to have significantly lower target payout ratios7 and are more prone to cut dividends, compared to other firms (Ibid).

Other studies indicate that decisions in family firms are influenced by trust motives (Pindado et al., 2012; Isakov & Weisskopf, 2014, 2015). Controlling families are considered more caring about their business than non-families. As a consequence, controlling families are more prone to generate trust among non-controlling shareholders and build a reputation of treating them well (Isakov & Weisskopf, 2014, 2015). Families usually have much of their wealth invested in the firm by which they are closely tied to the firm’s performance (Isakov & Weisskopf, 2015). Families also tend to perceive the firm as part of their identity, in contrast to non-families. Hence, families should have greater incentives to monitor management and not engage in expropriating activities compared to non-families (Isakov & Weisskopf, 2014). On

(13)

12 this basis, controlling families are expected to distribute higher dividends as a means of mitigating concerns of expropriation, as confirmed by the studies by Pindado et al. (2012) and Isakov and Weisskopf (2015).

Pindado et al. (2012) study the impact of family ownership on dividend policy examining nine European countries (Sweden excluded). They focus on whether family firms use dividends as a corporate governance mechanism to mitigate agency problems between controlling and non-controlling shareholders. Paying dividends instead of holding excess cash could generate trust among non-controlling shareholders showing that the controlling shareholder will not engage in wealth expropriating activities (Ibid). In general, their findings indicate that disbursing cash through dividends has the potential to align the interests of controlling and non-controlling shareholders. Family firms distribute higher and more stable dividends compared to non-family firms, interpreted as dividends are used as a mechanism to overcome agency problems. Dividends, compared to other corporate governance mechanisms, may constitute a more credible signal for family shareholders, especially due to their special connection to the business (Ibid). Isakov and Weisskopf (2015) examine the impact of family ownership on payout policy in Switzerland. They find that founding family firms have higher total payouts and distribute higher dividends than non-family firms. These findings are partly explained by the interest of families to remain long-term control of the business and reputational considerations (Ibid).

In sum, the incentives and motivations of family firms related to dividend policy are somewhat ambiguous, associating family ownership with both benefits and costs. However, considering the weak investor protection in Sweden (e.g. La Porta et al., 1998) and findings in prior research (Pindado et al., 2012; Isakov & Weisskopf, 2015), it is suggested that Swedish family firms are inclined to distribute high dividends to allay agency problems. In this way, dividend payouts constitute a means of ensuring non-controlling shareholders that they will not be expropriated. On this basis, the study’s first hypothesis is formulated as follows:

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

(14)

non-13 controlling shareholders (Pindado et al., 2012; Isakov & Weisskopf, 2015). However, depending on its identity the blockholder’s presence could also result in higher dividends, positively affecting non-controlling shareholders (Pindado et al., 2012).

Faccio et al. (2001) investigate ownership structure related to dividend payouts on the West European and East Asian contexts. They show that the presence of several large blockholders has a positive influence on dividends in Europe, but the opposite effect in Asia. The negative influence in Asia appears to be explained by blockholders colluding with each other in expropriating activities (Ibid). Following the findings of Faccio et al. (2001) in Europe, Ramli (2010) shows that the presence of a second blockholder positively impacts firms’ dividend policies in Malaysia, encouraging the controlling blockholder to pay higher dividends.

The findings of Pindado et al. (2012) indicate that the influenceof the second blockholder in family firms depends on the second blockholder’s identity. Dividends are lower in the presence of a family second blockholder, compared to a non-family second blockholder. A family second blockholder appears to collude with the controlling family. Family second blockholders appear to prefer low dividends and use cash for extraction of private benefits at the non-controlling shareholders’ expense. The positive impact of non-family second blockholders is explained by these shareholders being an effective monitoring and controlling factor, functioning as a disciplining force that encourages families to pay higher dividends (Ibid). Besides, Pindado et al. (2012) find that family firms with no second blockholder8 disburse higher dividends than non-family firms.

In line with the findings of Pindado et al. (2012), Isakov & Weisskopf (2015) show that the second blockholder in family firms could function as a corporate governance mechanism to prevent expropriation of non-controlling shareholders. Total payouts are highest when there is a significant difference in the level of ownership between the controlling family and the second blockholder or in family firms without a second blockholder. Their results suggest that the level of ownership stake of the second largest blockholder has an impact on its monitoring function of the controlling blockholder. The power and motivation of the second blockholder increases when its stake is of similar size as the stake of the controlling family (Ibid).

8_{In family firms without a second largest blockholder present no shareholder except the controlling family}

(15)

14 Prior studies indicate that the presence of a second blockholder influences dividend payout policy. Whether the influence is positive or not seems to depend on the context and the second blockholder’s incentives. Considering the findings of Pindado et al. (2012) and Isakov and Weisskopf (2015) in family firms, dividend payouts indicate to be highest in family firms without a second blockholder present. Based on these findings, the study’s second hypothesis is formulated as follows:

Hypothesis 2: Family firms without a second blockholder have higher dividend payouts than family firms with a second blockholder.

Also, the identity of the second blockholder appears to have an impact on dividend payout policy. Pindado et al. (2012) find that family second blockholders negatively influence dividend payout policies. Thus, it is expected that family firms without a family second blockholder disburse higher dividends. The study’s third hypothesis is formulated as follows:

Hypothesis 3: Family firms without a family second blockholder have higher dividend payouts than family firms with a family second blockholder.

2.4.3 The impact of the separation of ownership and control on dividend payout

policy

Commonly, controlling blockholders have more voting rights than cash-flow rights by using control enhancing mechanisms, for example dual-class shares (La Porta et al., 1999; Faccio & Lang, 2002; Cronqvist & Nilsson, 2003). This separation of ownership and control implies that the blockholder can make decisions at a lower risk in proportion to the invested capital. Besides, it increases the possibilities and incentives to extract self-serving benefits at the expense of non-controlling shareholders (Cronqvist & Nilsson, 2003). Accordingly, as these instruments strengthen the influence of the blockholder, the use impacts important corporate governance decisions, for example dividend payout policy (Faccio et al., 2001; Pindado et al., 2012).

(16)

15 these findings, Pindado et al. (2012) find that family firms with no separation of voting rights and cash-flow rights disburse higher dividends than family firms with this separation. Contrary to the just mentioned studies, De Cesari (2012) finds that dividends constitute a higher proportion of total payouts in Italian firms with separation of voting rights and cash-flow rights. A weaker corporate governance structure is related to a higher risk of expropriation of non-controlling shareholders, hence dividends function as a means of mitigating agency problems (Ibid). This is supported by the findings of Jordan, Liu and Wu (2014) on American firms. They find that firms using dual-class shares distribute higher dividends, as to alleviate agency problems, than firms that are not using dual-class shares (Ibid).

The results in prior literature on the separation of ownership and control appear to be mixed. However, referring to the study’s first hypothesis, family firms are expected to have higher dividends than non-family firms as a means of generating trust among non-controlling shareholders. Based on this, dividends are expected to be even more needed as a ‘trust-generating mechanism’ to mitigate agency problems in firms where control and ownership separate, as the risk of expropriation is higher. Hence, it is expected that family firms disburse higher dividend payouts when there is a wedge between voting rights and cash-flow rights. The study’s fourth hypothesis is formulated as follows:

Hypothesis 4: Family firms with a wedge between voting rights and cash-flow rights have higher dividend payouts than family firms without a wedge between voting rights and cash-flow rights.

(17)

16

3. Methodology

3.1 Research method

To examine the impact of family ownership on dividend payout policy in Sweden, a quantitative research strategy with a deductive approach is applied. The study’s four hypotheses are formulated based on prior studies and the findings are analyzed in relation to existing theory, which characterizes a deductive method (Bryman & Bell, 2015). The hypotheses are empirically tested by four multiple regression analyses. The remaining of this chapter provides detailed information about the data sample and analyses performed.

3.2 Data description

Table 1 summarizes the sample selection procedure. The study is based on annual data ranging over a six-year period from 2013 to 2018. The sample include publicly listed firms on the Stockholm Stock Exchange (Nasdaq OMX Stockholm) constituting Small, Mid and Large Cap9 firms. The examination period is chosen to provide an as representative view of firms’ dividend payout policies as possible. The length of the period (six years) is similar to what Isakov and Weisskopf (2015) use in their study (eight years) examining family firms and payout policies in Switzerland. This further strengthens the chosen length of the examination period.

Before any observations are excluded, the sample consists of 1,792 firm-year observations (378 unique firms). Inclusion in the sample does not require a firm to have been listed during the entire examination period. Following prior studies on dividend payout policy (Setia-Atmaja et al., 2009; Pindado et al., 2012; Isakov & Weisskopf, 2015) foreign, financial and investment firms are excluded from the sample since the dividend payout policies of these firms may be covered by certain regulations. ‘Other eliminations’ (presented in Table 1) include observations for which financial data is missing or the ultimate shareholder10 by various reasons cannot be accurately identified. The financial data is trimmed when considered necessary and constitute 12 observations with extreme values for the control variable debt-to-equity ratio (D/E). These eliminated values are included in ‘Other eliminations’.

9_{Firms listed on Small, Mid and Large Cap have a market capitalization below EUR 150 million, between EUR}

150 million to EUR 1 billion and over EUR 1 billion, respectively (Nasdaq, 2020b).

10_{An ultimate shareholder is the shareholder that ultimately controls a firm either directly or indirectly through a}

(18)

17 The final sample consists of 1,363 firm-year observations (293 unique firms). As the sample size is similar to what Isakov and Weisskopf (2015) use when examining family firms in Switzerland11, the sample is considered sufficient for an examination of firms within a single legal jurisdiction (Sweden).

Table 1. Sample selection

_#Firm-years _#Firms

Firms listed on Nasdaq OMX Stockholm 2013–2018 1,792 378

Financial and investment firms 131 25

Foreign firms 144 30

Other eliminations 154 30

Final sample 1,363 293

Notes. The initial sample consists of publicly listed firms on Nasdaq OMX Stockholm during the years 2013–2018. ‘Other eliminations’ include observations for which financial data is missing in Thomson Reuters Eikon (Datastream), the ultimate shareholder cannot be accurately identified and extreme values on the control variable debt-to-equity ratio (D/E). If a firm constitutes of several observations, where some (but not all) are drop-outs, the firm is included in the number of #Firms for the final sample and not in the number of #Firms for the eliminating category.

The information required to perform the study is financial data for the dependent variable (DIVIDEND PAYOUT) and the control variables, as well as detailed information on firms’ ownership structures for the independent variables. Financial data is extracted from Thomson Reuters Eikon, using Datastream, for the last day of each financial year over the studied period. Information on firms’ ownership structures is used to distinguish family firms from non-family firms and identify the presence and identity of the second largest blockholder12. Besides, it is used to establish if there is a wedge between ownership (cash-flow rights) and control (voting rights). Information on firms’ ownership structures is collected from the database Holdings for the last day of each financial year over the studied period. In comparison, Pindado et al. (2012)13 apply single year ownership data from each firm on several years’ financial data. This is done under the assumption that firms’ ownership structures remain relatively stable over time (referring to reasoning by La Porta et al. (1999)). However, collecting ownership data for each respective year will provide a more accurate view of firms’ ownership structures, motivating the choice in this study. The data extracted from Thomson Reuters Eikon

11_{Isakov and Weisskopf’s sample (2015) includes 1,228 firm-year observations (187 unique firms).} 12_{These processes are described in detail in Section 3.2.2.}

13_{Pindado et al. (2012) base their study on information from the ownership database developed by Faccio and}

(19)

18 (Datastream) and Holdings is merged and compiled in Microsoft Excel. The regression analyses are performed in SPSS.

In case the data from Holdings do not provide any or not adequate information of the ultimate shareholder of a firm, the ultimate shareholder is tracked manually based on firms’ annual reports and, if necessary, various internet sources (e.g. firm websites). The annual reports are mainly accessed through Retriever Business. Identifying the ultimate shareholder is a vital procedure to classify the observations correctly, i.e. as family or non-family firms, and highly affects the accuracy of the investigation. If the ultimate shareholder is an unlisted firm for which the shareholders cannot be identified, the observation is classified as a family firm, following Faccio and Lang (2002). According to their reasoning, the likelihood is low that an unlisted firm is controlled by a financial institution, a state or a widely held corporation, leaving a family as the most likely controller (Ibid). Figure 1 illustrates an example of the process of identifying the ultimate shareholder of a firm for a specific year.

Figure 1. Ownership structure in Securitas AB 2018

Notes. The figure describes the process of identifying the ultimate shareholder of the firm. Cash-flow rights (ownership) are denoted by ‘‘C’’ and voting rights (control) are denoted by ‘‘V.’’ The voting rights of the controlling blockholder of Securitas AB and the ultimate shareholder are marked in bold. The box of the controlling blockholder (ultimate shareholder) of Securitas AB is marked in bold (with dotted lines).

(20)

19

3.3 Family definitions

3.3.1 Family definitions in prior studies

To investigate the impact of family ownership on dividend payout policy, family firms are distinguished from non-family firms. Prior studies investigating family ownership use different definitions to classify a family firm, indicating an absence of a widely agreed family firm definition. La Porta et al. (1999) define a family firm as a firm with an individual as controlling blockholder. Faccio and Lang’s (2002) definition is a firm with an ultimate shareholder which is a family (or individual) or a firm that is unlisted on any stock exchange. Setia-Atmaja et al. (2009) define a family firm as a firm in which the founding family or family member or a private individual owns at least 20% of total cash-flow rights and is involved in the top management. Croci et al.’s (2011) definition is a firm in which a family or an individual is the largest shareholder in terms of voting rights (and the rights exceed the determined threshold of control). Isakov and Weisskopf (2014, 2015) define a family firm as a firm controlled by one or several individuals or families that have either founded the firm or have been shareholders over a long time and substantially shaped the firm.

Prior studies have also used different thresholds (a percentage of total voting rights) to define a controlling blockholder. This includes 10% (La Porta et al., 1999; Faccio & Lang, 2002; Croci et al., 2011), 20% (La Porta et al., 1999; Faccio & Lang, 2002; Setia-Atmaja et al., 2009; Isakov & Weisskopf, 2014, 2015) and 25% (Cronqvist & Nilsson, 2003).

3.3.2 Family definitions in this study

The family firm definition used in this study is based on those used in prior studies (Faccio & Lang, 2002; Isakov & Weisskopf, 2014, 2015). A threshold of control is set at 20% of total voting rights14 following Isakov and Weisskopf (2014, 2015) examining ownership in Swiss firms. Choosing the same level when examining Swedish firms is motivated by the similarities between the Swedish and Swiss institutional settings; weak level of investor protection (La Porta et al., 1998), highly concentrated ownership structures, high presence of family firms and high use of dual-class shares (Faccio & Lang, 2002).

14_{In excess of tests using a control threshold of 20%, an additional test is performed using a threshold of 25%}

(21)

20 In this study, a family firm (FAM) is defined as a firm ultimately owned by a shareholder that holds at least 20% of the firm’s total voting rights and is a family15_{or an individual. A firm}

that does not meet the family firm definition could be controlled by various ownership identities such as institutions, states or many small shareholders (widely held). These firms are referred to as family firms. Panel A in Table 2 provides the distribution of family firms and non-family firms according to this definition.

To check the robustness of the results, a second family firm definition (FAM2) is used with a different degree of strictness. In addition to a control threshold (20%) this definition includes active ownership considerations, also following family firm definitions used in prior studies (Setia-Atmaja et al., 2009). According to the second family firm definition, a family firm (FAM2) is a firm ultimately owned by a shareholder that holds at least 20% of the firm’s total voting rights, and is a family or an individual, and is himself/herself active or has a family member16 that is active in the firm; either as CEO, chairman or member of the board of directors. Panel B in Table 2 provides the distribution of family firms and non-family firms according to this definition.

Similar procedures are applied to determine the presence of the second largest blockholder and to classify it as a family or not. The second largest blockholder (2NDBH) is defined as the second largest shareholder that holds voting rights above a threshold of 10% of total voting rights. Holding 10% of total voting rights is considered enough to constitute a substantial amount of shares, characterizing a blockholder (La Porta et al., 1999; Faccio & Lang, 2002; Croci et al., 2011). A family second blockholder (2NDFAM) is defined as the second largest shareholder that holds at least 10% of the firm’s total voting rights and is a family or an individual. Panel C in Table 2 provides the distribution of second largest blockholders and which of these are families according to these definitions.

15_{A family in this study consists of several individuals sharing the same last name, e.g. Lena Jonsson, Stefan}

Jonsson and Peter Jonsson.

16_{A family member in this study is an individual who shares the same last name as the/those individual/s in}

(22)

21 To check the robustness of the results, a stricter definition of family second blockholder is also applied. In this test, a family second blockholder (2NDFAM2) is defined as the second largest shareholder that holds at least 10% of the firm’s total voting rights, and is a family or an individual, and is himself/herself active or has a family member that is active in the firm; either as CEO, chairman or member of the board of directors. A second largest blockholder (2NDBH) is defined as the second largest shareholder that holds voting rights above a threshold of 10%

Table 2. Family firms vs. non-family firms & second largest blockholders

Panel A: Family firms vs. non-family firms #Firm-years

Final sample 1,363

Of which are family firms (FAM) 713 52%

Of which are non-family firms 650 48%

Panel B: Family firms vs. non-family firms (robustness) #Firm-years

Final sample 1,363

Of which are family firms (FAM2) 529 39%

Of which are non-family firms 834 61%

Panel C: Second largest blockholders #Firm-years

Final sample 1,363

Of which do not have a second largest blockholder 741 54% Of which have a second largest blockholder (2NDBH) 622 46%

Of which are families (2NDFAM) 513

Of which are not families 109

Panel D: Second largest blockholders (robustness) #Firm-years

Final sample 1,363

Of which do not have a second largest blockholder 741 54% Of which have a second largest blockholder (2NDBH) 622 46%

Of which are families (2NDFAM2) 426

Of which are not families 196

(23)

22 of total voting rights. Panel D in Table 2 provides the distribution of second largest blockholders and which of these are families according to these definitions.

Figure 2 illustrates an example of classifying a firm according to the main family firm definition and the second family firm definition. The example builds on the prior example in Section 3.2.

Figure 2. Ownership structure and board participation/CEO position in Securitas AB 2018

Notes. The figure describes the process of classifying a firm according to the main family firm definition and the second family firm definition. Cash-flow rights (ownership) are denoted by ‘‘C’’ and voting rights (control) are denoted by ‘‘V.’’ The voting rights of the controlling blockholder of Securitas AB and the ultimate shareholder are marked in bold. The box of the controlling blockholder (ultimate shareholder) of Securitas AB is marked in bold (with dotted lines). The active shareholder or his/her family member is marked in bold.

In 2018 the controlling blockholder of Securitas AB was Investment AB Latour (holding ca. 30% voting rights and ca. 11% cash-flow rights). In turn, Investment AB Latour was controlled by Gustaf Douglas (holding ca. 79% voting rights and ca. 76% cash-flow rights), making Gustaf Douglas the ultimate shareholder of Securitas AB by holdings through multiple chains. Securitas AB is considered a family firm according to the main definition as Gustaf Douglas is the ultimate shareholder of Securitas AB holding more than 20% of the total voting rights and is an individual. Securitas AB is also considered a family firm according to the second family firm definition. The requirements of the main definition are fulfilled and Carl Douglas, an individual sharing the same last name as Gustaf Douglas hence considered a family member, participates on the board of directors in Securitas AB.

3.4 Variables

(24)

23 constructed as a firm’s total cash dividends divided by total assets, following prior studies (Pindado et al., 2012). Other studies have used dividends divided by earnings (Ramli, 2010). Due to the risk of earnings management dividing dividends by total assets could provide a more correct view of whether firms use dividends as a corporate governance mechanism (Pindado et al., 2012). The independent variables are family firm (FAM), second largest blockholder (2NDBH), family second blockholder (2NDFAM), and wedge between voting rights and cash-flow rights (WEDGE). All independent variables are dummy variables. Regressions are performed including each of these variables.

Additional analyses based on the second family firm definition are performed to check if the findings are robust to an alternative family firm definition. Following independent variables are included in these additional analyses; family firm (FAM2), second largest blockholder (2NDBH), family second blockholder (2NDFAM2), and wedge between voting rights and cash-flow rights (WEDGE).

A selection of control variables is used to minimize the risks of drawing the wrong conclusions regarding the relationship between the dependent variable and the independent variables (Pallant, 2016). The control variables include debt-to-equity ratio (D/E), net income (NI), firm size (SIZE), share repurchases (SR) and Tobin’s q (TQ). The variables chosen are factors either known from prior studies to have an impact on a firm’s dividend payout policy or expected to have an impact according to related studies. The variable debt-to-equity ratio (D/E) is used to control for debt which, in the same way as dividends, could constitute a corporate governance mechanism to alleviate agency problems. In this sense, firms may use debt as a substitute for dividends (Jensen, 1986; Setia-Atmaja et al., 2009). Therefore, a negative relation is expected between dividend payout and debt-to-equity ratio. The variable is trimmed for extreme values to improve its inclusion in the analyses (12 observations excluded). The variable net income (after preferred dividends) (NI) is used to control for profitability as more profitable firms are expected to have higher dividends than firms relatively not as profitable (Fama & French, 2001). Consequently, a positive relation is expected between dividend payout and net income. The variable is divided by total assets to make the measure more comparable and representative.

(25)

24 Pindado et al., 2012; Attig, Boubakri, Ghoul & Guedhami, 2016). The log-transformation is done to adjust for skewness in the data and make it more normally distributed (Pallant, 2016). Even though larger firms tend to make higher dividend payouts in absolute numbers, it is expected that they pay out a smaller percentage of total assets than smaller firms. Since dividend payout is a relative measure a negative relation to firm size is expected. The variable share repurchases (SR) is used as repurchases may be used as substitutes for dividends to disburse cash to shareholders (Pindado et al., 2012). Hence, a negative relation between dividend payout and share repurchases is expected. The variable is created as a dummy that equals 1 if the firm has done a share repurchase, and otherwise 0. By creating this variable as a dummy there is a risk of not capturing firms that eventually both have done a rights issue and a share repurchase. However, this risk is not expected to have any considerable impact on the results. The variable Tobin’s q (TQ), calculated as the market value of equity divided by the book value of total assets, is used as a proxy and control for investment opportunities (Yoshikawa & Rasheed, 2010; Pindado et al., 2012; Isakov & Weisskopf, 2015). Investment opportunities is used to control for dividends as having a signaling function (Pindado et al. 2012). A firm may disburse high dividends if having a lot of profitable investment opportunities as a means of sending information of the firm’s future prospects to the market (Bhattacharya, 1979). Based on this, a positive relation between dividend payout and Tobin’s q is expected.17

Table 3. Definition of variables

Dependent

variable Description Source

DIVIDEND PAYOUT

Total cash dividends divided by total assets. Thomson Reuters Eikon (Datastream).

Independent

variables Description Source

FAM A dummy variable taking the value 1 if the firm is a family firm according to the family firm definition, otherwise 0. Section 3.3.2 provides a further description.

Holdings and manually gathered from e.g. annual reports.

FAM2 (robustness)

A dummy variable taking the value 1 if the firm is a family firm according to the second family firm definition, otherwise 0. Section 3.3.2 provides a further description.

17_{A negative relation between dividend payout and Tobin’s q is also possible as both are two alternatives of using}

(26)

25 Table 3. continued

Independent

2NDBH A dummy variable taking the value 1 if there is a second blockholder present according to the second blockholder definition, otherwise 0. Section 3.3.2 provides a further description.

2NDFAM A dummy variable taking the value 1 if there is a second blockholder present which is a family according to the family second blockholder definition, otherwise 0. Section 3.3.2 provides a further description.

2NDFAM2 (robustness)

A dummy variable taking the value 1 if there is a second blockholder present which is a family according to the second family second blockholder definition, otherwise 0. Section 3.3.2 provides a further description.

WEDGE A dummy variable taking the value 1 if the firm has a wedge between voting rights and cash-flow rights, otherwise 0.

Holdings.

Control

D/E Debt-to-equity ratio. Calculated as total debt divided by common equity. Trimmed (12 observations).

Thomson Reuters Eikon (Datastream).

NI Net income (after preferred dividends) divided by total assets.

SIZE Firm size. The common logarithm of total assets. Thomson Reuters Eikon (Datastream).

SR A dummy variable taking the value 1 if the firm has done a share repurchase, and otherwise 0. Calculated as: SR = (SHOit-1 - SHOit) where SHO is the number of common

outstanding shares in the corresponding period of time.

TQ Tobin’s q. Calculated as the market value of equity divided by the book value of total assets.

(27)

26 capture any variation in the outcome that is related to a change over time or a specific industry, i.e. variation that is not attributable to the independent variables. A firm’s industry is determined based on the Nasdaq OMX Stockholm classification, including the industry categories ‘Basic Materials’, ‘Consumer goods’, ‘Consumer services’, ‘Financials’, ‘Health care’, ‘Industrials’, ‘Oil and Gas’, ‘Technology’, ‘Telecom’ and ‘Utilities’. As previously mentioned, financial firms are excluded hence this category is excluded. An additional category named ‘Other’ is created including firms for which an industry category is missing or inappropriate. Examples of such firms are real estate firms. The categories ‘Oil and Gas’ and ‘Utilities‘ are eliminated as these consist of only a few observations. These observations are included in ‘Other’. Table 4 discloses the industry distribution of the sample.

Table 4. Industry distribution

Industry #Firm-years #Firms

Basic Materials 64 5% 12 4% Consumer goods 136 10% 28 10% Consumer services 155 11% 35 12% Health care 214 18% 51 17% Industrials 400 29% 80 27% Technology 190 13% 44 15% Telecommunications 28 2% 5 2% Other 176 12% 38 13% Total 1,363 100% 293 100%

Notes. The number of firm-year observations and firms for each industry. Source: Nasdaq, 2020. The category ‘Other’ includes firms for which an industry category is missing, is inappropriate or are classified to the categories ‘Oil and Gas’ or ‘Utilities’ (categories excluded as these consist of few observations).

3.5 Multiple regression models

(28)

27 dummy), the same set of control variables18 (Controls) and controls for industry and year fixed effects.

Model one (1) is used to test the study’s first hypothesis that family firms have higher dividend payouts than non-family firms. Thus, a positive relation is expected between DIVIDEND PAYOUT and FAM. The model includes a dummy that equals the value 1 if the firm is a family firm, and 0 if the firm is a non-family firm.

𝐷𝐼𝑉𝐼𝐷𝐸𝑁𝐷 𝑃𝐴𝑌𝑂𝑈𝑇 = 0 + 1FAM + 2Controls + 3Industry + 4Year + it (1)

Model two (2) is used to test the study’s second hypothesis that family firms without a second blockholder have higher dividend payouts than family firms with a second blockholder. Thus, a negative relation is expected between DIVIDEND PAYOUT and 2NDBH. In this analysis, non-family firms are excluded as the focus is to assess the impact on dividend payout policy from a certain family ownership structure. The model includes a dummy that equals the value 1 if there is a second blockholder present, and 0 if no second blockholder is present.

𝐷𝐼𝑉𝐼𝐷𝐸𝑁𝐷 𝑃𝐴𝑌𝑂𝑈𝑇 = 0 + 12NDBH + 2Controls + 3Industry + 4Year + i (2)

Model three (3) is used to test the study’s third hypothesis that family firms without a family second blockholder have higher dividend payouts than family firms with a family second blockholder. Thus, a negative relation is expected between DIVIDEND PAYOUT and 2NDFAM. As in model 2, non-family firms are excluded as the focus is to assess the impact on dividend payout policy from a certain family ownership structure. Model 3 includes a dummy that equals the value 1 if there is a family second blockholder present, and 0 if not.

𝐷𝐼𝑉𝐼𝐷𝐸𝑁𝐷 𝑃𝐴𝑌𝑂𝑈𝑇 = 0 + 12NDFAM+ 2Controls + 3Industry + 4Year + it (3)

Model four (4) is used to test the study’s fourth hypothesis that family firms with a wedge between voting rights and cash-flow rights have higher dividend payouts than family firms without a wedge. Thus, a positive relation is expected between DIVIDEND PAYOUT and 2NDFAM. As in models 2 and 3, non-family firms are excluded as the focus is to investigate the impact on dividend payout policy from a certain family ownership structure. Model 4

(29)

28 includes a dummy that equals the value 1 if there is a wedge between voting rights and cash-flow rights, and 0 otherwise.

𝐷𝐼𝑉𝐼𝐷𝐸𝑁𝐷 𝑃𝐴𝑌𝑂𝑈𝑇 = 0 + 1WEDGE + 2Controls + 3Industry + 4Year + it (4)

3.6 Regression model assumptions

3.6.1 Endogeneity

The classical OLS regression assumptions need to be in order for the regression models to provide the most accurate estimates for what is intended (Studenmund, 2011). One of these is that the independent variables are uncorrelated with the error term, in broad terms referred to as endogeneity. The assumption can be violated due to an issue called ‘omitted variable bias’. This issue occurs if a variable that is a determinant of the dependent variable is excluded from the regression, making the coefficient estimates biased and inconsistent (Studenmund, 2011). Likewise, the assumption can be violated by adding an interaction effect as the product of two variables to the regression model. To make the regression models in this study as accurately specified as possible, the variables included are chosen based on what has been used in prior studies making similar investigations (e.g. De Cesari, 2012; Pindado et al., 2012; Isakov & Weisskopf, 2015). Also, no interaction terms are added to the models.

3.6.2 Homoscedasticity

Another classical OLS assumption is constant variance, or homoscedasticity, of the residuals. This means that when the value of an independent variable (x) increases, the variation in the dependent variable (y) is constant. The assumption can be checked by assessing the residuals scatterplot, in which the standardized residuals are plotted against the standardized predicted values. The assumption is suggestively violated if the pattern of the residuals shown deviates from a centralized rectangle, and the distribution shows a systematic pattern (Pallant, 2016). To assess the condition of homoscedasticity a residuals scatterplot is done. This plot shows that the assumption is violated, which is not surprising considering the inclusion of non-dividend paying firms in the sample. This implication of this is that many observations will have a zero value for the dependent variable DIVIDEND PAYOUT.

3.6.3 Multicollinearity

(30)

29 Multicollinearity exists when the independent variables are highly correlated, i.e. has a correlation value of ±0.9. The presence of multicollinearity negatively impacts the explanatory power of the regression model (Pallant, 2016). Testing for multicollinearity in this study is done by conducting a Pearson’s correlation test and examine variance inflation factors (VIFs) through VIF-tests. A VIF-value above 10 indicates a problem with multicollinearity (Pallant, 2016).

3.6.4 Normality

(31)

30

4. Results and analyses

4.1 Descriptive statistics

Table 5 reports the Pearson’s correlation coefficients for the test variables. From the table it can be interpreted how the different variables correlate with each other. DIVIDEND PAYOUT and FAM are weakly positively correlated. The direction is as expected. DIVIDEND PAYOUT and WEDGE has a similar correlation, although somewhat stronger. DIVIDEND PAYOUT is weakly negatively correlated with 2NDBH and 2NDFAM, although the former correlation is not significant. There is a weak negative relation between DIVIDEND PAYOUT and to-equity ratio (D/E), following the expectations. This means that firms that are more debt-financed have lower dividends. There is a moderate positive correlation between DIVIDEND PAYOUT and net income (NI), also following expectations. This means that more profitable firms have higher dividends. The correlation between DIVIDEND PAYOUT and firm size (SIZE) is weakly negative, which is as expected. This implies that large firms have lower dividends payouts relative their firm size than smaller firms. The correlation between DIVIDEND PAYOUT and Tobin’s q (TQ) is moderately positive, following expectations. This means that firms with profitable investment opportunities pay higher dividends.

The strongest correlation is between 2NDBH and 2NDFAM (0.848), which do not exceed but is close to the accepted level of ±0.9. However, as 2NDBH and 2NDFAM are not included in the same regression model this does not constitute a problem. Neither of the remaining values in the table are above the accepted level, implying no indications of multicollinearity in the models. VIF-tests are conducted as additional checks, as these tests could reveal problems with multicollinearity which may not be evident in a correlation test (Pallant, 2016). The results of the VIF-tests indicate no problems with multicollinearity.19

(32)

31 Table 5. Pearson correlation test between the test variables

DIVIDEND

PAYOUT FAM 2NDBH 2NDFAM WEDGE D/E NI SIZE SR TQ

DIVIDEND PAYOUT 1 FAM 0.061** 1 2NDBH -0.032 0.096*** 1 2NDFAM -0.045* 0.135*** 0.848*** 1 WEDGE 0.091*** 0.312*** 0.125*** 0.188*** 1 D/E -0.219*** 0.063** 0.078*** 0.076*** 0.083*** 1 NI 0.358*** 0.056** 0.069** 0.048* 0.098*** -0.077*** 1 SIZE -0.072*** 0.111*** -0.089*** -0.087*** 0.233*** 0.305*** 0.248*** 1 SR 0.002 0.013 -0.023 -0.022 0.011 -0.031 0.057** 0.110*** 1 TQ 0.333*** -0.129*** -0.025 -0.075*** -0.099*** -0.336*** 0.017 -0.323*** -0.055** 1

(33)

32 Table 6. Descriptive statistics

Panel A: Whole sample

Variable Mean Median SD Min Max

DIVIDEND PAYOUT 0.030 0.016 0.048 0.000 0.455 D/E 58.76 43.34 64.55 0.000 616.2 NI 0.028 0.056 0.179 -1.716 0.581 SIZE 6.447 6.379 0.829 4.442 8.664 SR 0.104 0.000 0.306 0.000 1.000 TQ 1.677 0.968 2.280 0.013 28.66

Panel B: Family firms vs. non-family firms

Family firms (N = 713) Non-family firms (N = 650)

Variable Mean Mean

DIVIDEND PAYOUT 0.032 0.027** 2NDBH 0.502 0.406*** 2NDFAM 0.439 0.308*** WEDGE 0.633 0.320*** D/E 62.65 54.50** NI 0.037 0.017** SIZE 6.535 6.351*** SR 0.108 0.100 TQ 1.396 1.985***

Notes. The sample consists of publicly listed firms on Nasdaq OMX Stockholm during the years 2013–2018 (N = 1,363). Family firms and non-family firms are classified following the main family firm definition. See Table 3 for a definition of the variables.

*, ** and * denote significance on 10%, 5% and 1% level using a two-sample t-test of means.

Table 6 provides the descriptive statistics of the whole sample (Panel A) and family firms versus non-family firms (Panel B). As shown in Panel A, the statistics for the control variables debt-to-equity ratio (D/E) and Tobin’s q (TQ) reveal that the distribution is strongly positively skewed, even though the variable D/E has been trimmed for extreme values20. The mean for the control variable share repurchases (SR) shows that ca. 10% of all observations have done a share repurchase during the examined years.

(34)

33 As shown in Panel B, family firms (FAM) have, on average, higher dividend payouts than non-family firms. The difference in means is significant at the 5% level. This provides initial support for the study’s first hypothesis that family firms have higher dividend payouts than non-family firms. This result follows prior studies (Pindado et al., 2012; Isakov & Weisskopf, 2015). However, caution should be taken when interpreting this result as the test does not account for other factors that may influence the dividend payout policy. A second blockholder (2NDBH) is more often present in family firms than in non-family firms. Also, family second blockholders (2NDFAM) are more common in family firms than in non-family firms. Both 2NDBH and 2NDFAM are significantly different at the 1% level. Family firms more often have a wedge between voting rights and cash-flow rights (WEDGE) than non-family firms. The difference in means is significant at the 1% level. This follows the finding of Cronqvist and Nilsson (2003), also examining Swedish firms.

(35)

34

4.2 Regression results

Table 7 presents the results of regression models 1–4. These models were performed to test the four hypotheses of the study to investigate whether family ownership impacts firms’ dividend payout policies.

Table 7. Regression results

Dep. var.: DIVIDEND PAYOUT (1) (2) (3) (4) FAM 0.008*** (0.000) 2NDBH _-0.009*** (0.003) 2NDFAM _-0.007** (0.022) WEDGE _0.013*** (0.000) D/E -0.000** -0.000** -0.000*** -0.000*** (0.015) (0.013) (0.007) (0.005) NI _0.090*** _0.125*** _0.124*** _0.119*** (0.000) (0.000) (0.000) (0.000) SIZE _-0.005*** _-0.005** _-0.004** _-0.006*** (0.003) (0.015) (0.025) (0.003) SR _0.001 _0.004 _0.004 _0.007 (0.735) (0.413) (0.355) (0.174) TQ _0.007*** _0.007*** _0.007*** _0.007*** (0.000) (0.000) (0.000) Constant _0.050*** _0.059*** _0.055*** _0.053*** (0.000) (0.000) (0.000) (0.000) N _1,363 ₇₁₃ ₇₁₃ ₇₁₃

Year & Industry _YES _YES _YES _YES

Adjusted R2 0.274 0.313 0.309 0.322

Notes. The sample in Table 7 consists of publicly listed firms on Nasdaq OMX Stockholm during the years 2013–2018. The variable DIVIDEND PAYOUT is the dependent variable in all regressions. The variable family firm (FAM) is a dummy taking the value 1 if the firm is ultimately owned by a shareholder that holds at least 20% of the firm’s total voting rights and is a family or an individual. FAM is the independent variable in model 1 (N = 1,363). The variable second blockholder (2NDBH) is a dummy taking the value 1 if the second largest shareholder holds at least 10% of the firms total voting rights. 2NDBH is the independent variable in model 2 (N = 713). The variable family second blockholder (2NDFAM) is a dummy taking the value 1 if the second largest shareholder holds at least 10% of the firms total voting rights, and is a family or an individual. 2NDFAM is the independent variable in model 3 (N = 713). The variable wedge (WEDGE) is a dummy taking the value 1 if the firm has a wedge between voting rights and cash-flow rights. WEDGE is the independent variable in model 4 (N = 713). For a definition of the control variables, please see Table 3.

The impact of family ownership on dividend payout policy