June 08, 2018

Master Thesis

Family control and cash holding: the roles of multinationality,

country-level governance and trust

June 08, 2018

Student: Yan Zhu

Student number: S3262413 Study Programme: MSc IFM Supervisor: Dr. Halit Gonenc

Co-assessor:Dr. Melsa Ararat Merrell

Abstract

This thesis investigates the effect of family control on cash holding and the moderating effect of the level of multinationality, the country-level governance and trust on the correlation between family control and cash holding. Using a sample of 278798 firm-years observations representing from 52 countries between 2000 and 2016, the regression results show that the positive correlation between family control and cash holding becomes weaker with the level of multinationality. Furthermore, the results show that family firms hold less cash in countries with weaker governance and lower trust, which implies that family control plays different roles in different countries with different levels of country governance and country trust, and that family control provides a governance mechanism in countries with weak governance.

1

1. Introduction

Family firms are prevalent all over the world, and they are also an important component of the commercial field (Wei et al, 2011; Anderson and Reeb, 2003). Existing literature has compared family firms with non-family ones and found that they differ in terms of decision-making processes on key issues. One of the most crucial parts of corporate finance is cash holding, and the cash holding of family enterprises has attracted significant attention.

There are three main explanations for corporate cash holding: the transaction cost motive, the precautionary motive and agency issue. The first two rely on the cost of external finance and investment opportunities (Opler et al., 1999), while the last argues that cash holding is mainly influenced by the agency conflict because cash is easier to transfer into private benefits (Myers and Rajan, 1998). Strong corporate governance could reduce agency conflicts and thus reduce cash holding. However, the role of family ownership in corporate governance remains unclear. On one hand, controlling families could have a greater ability to monitor managers and bring strong governance (Wei et al., 2011). On the other hand, Liu et al. (2015) study Chinses family firms and find that controlling families, as dominant shareholders, have incentives to seek personal benefits and hence may exploit minority shareholders. Furthermore, from the view of the precautionary motive to hold cash, Anderson and Hamadi (2016) study Belgian listed firms and proposed their doubts about the managerial rent extraction of cash holding in the context of a control-oriented financial system. They find family firms who highly value control hold more precautionary cash. However, Croci et al. (2011) use sample of European firms and show that family firms invest less in high-risk investments, such as R&D projects, than non-family ones. With less risky investment, family firms prefer to hold less cash. Thus, the effect of family ownership on cash holding is mixed and the motivation of family firms to hold cash is unclear in previous studies. And most previous studies of family firms use sample from one region instead of using an international sample.

2

international companies (Calabro et al., 2014). However, it is not clear how the interaction between multinationality and family control influences cash holding. Multinationality is interrelated with three major explanations concerning the cash holdings of a company, including agency problems (Tong, 2011), financial flexibility (Jang, 2017) and diversification benefits (Fernandes and Gonenc, 2016). When a family firm becomes multinational, it may face more agency problems because it has become a more complex organization. However, the revenues of family enterprises become more diversified and can thus access capital markets more easily in foreign markets. Moreover, multinationality may also strengthen the governance role of family control.

The effect of family control and multinationality on cash holding can also be influenced by a country’s institutions. Many studies from the perspective of formal institutions have found that a country’s investor protections negatively affect corporate cash holding (Dittmar et al., 2003; Al-Najjar, 2012; Seifert and Gonenc, 2018). However, from the view of countries’ informal institutions, the level of cash is positively related to trust (Dudley and Zhang, 2016; Xie and Xin, 2015). Families generally have two obvious features: the family’s concern over reputation and its interest in long-term survival (Anderson et al., 2003). Because of these unique characteristics, the role of family control may change when it is faced with different countries’ governance systems and trust. Thus, unlike most previous research focused only on countries’ formal institutions to analyze corporate cash holding, this thesis take both formal and informal institution into account.

3

economies of scale in cash management. Their revenue sources become increasingly diversified compared with purely domestic family firms. When family firms become multinational, family control plays a strong corporate governance role because of family’s preoccupation with reputation and its survival over time, which leads to less cash holding. To clarify which argument is dominant, the country-level factors are considered.

At the country level, the results suggest that the role of family control changes in countries with different levels of governance, and that family firms hold less cash in countries with weak governance than in those with strong governance. This finding suggests that family control plays a strong governance role in countries with weak governance. However, the combined effect of family control and multinationality does not change with country governance. Rather, the results show that, in countries with weaker governance, multinational family firms hold less cash than domestic family firms, and multinational family firms hold far less cash than domestic family firms in countries with stronger governance. To conclude, based on agency theory, family control can positively affect cash holding because controlling shareholders have incentives to seek private benefit and exploit minority shareholders. Furthermore, considering the family’s concern for its reputation and its interest in long-term survival, family control changes its role in different country governance scenarios and plays a strong governance role in countries with weak governance. According to the precautionary motive, the interaction between family control and the level of multinationality has a negative effect on cash holdings because the high level of multinationality can lead family firms to have better access to capital markets, to have economies of scale in cash management and to have efficient cash management, while the revenue sources can become more diversified.

The level of countries’ trust is taken into consideration to verify that the role of family control differs based on the quality of a country’s institutions. The results show that family firms hold more cash compared to non-family firms in high trust countries and hold less cash than non-family ones in low-trust counties. This evidence demonstrates that family indeed plays different roles in high- and low-trust countries, and it plays a stronger governance role in low-trust and weak governance countries.

4

drivers of the cash policies of family firms. In addition, this thesis extends the research about family firms’ cash holding to investigate how the degree of multinationality affects corporate cash holding of family firms. Meanwhile, it clearly indicates of the role of family control in corporate governance and addresses how a country’s formal and informal institutions, such as country governance and trust, influence the role of family control on cash policy. Moreover, this thesis extends prior studies by using international sample.

The remaining sections of the paper are organized as follows: the literature on previous relative studies is reviewed in section 2 to develop the main hypothesis. The data collection and methodology are presented in section 3. Empirical findings are provided in section 4, followed by the conclusion.

2. Theoretical Background and Development of Hypotheses

The theoretical background can be explained in four parts. First, theories of the components of corporate cash holdings are discussed. Second, the role of family control on a company’s financial decisions is discussed. The literature concerning the multinational characteristic’s effect on corporate cash holding is explained third, and the influence of country-level factors on the role of family control on cash holding is discussed after. This section reviews the literature and advances testable hypotheses.

2.1.The determinant of corporate cash holding

5

have focused on agency problems between managers and owners, and between controlling shareholders and minority shareholders. Agency theory suggests that weak shareholder protection leads to more cash holdings (Dittmar et al., 2003). To summaries, the trade-off model illustrates three main motivations for companies to hold cash: precautionary motivation, transaction cost motivation and agency problems.

According to Keynes (1936), the transaction cost motivation for cash holding means holding cash to save the transaction costs of transferring assets into cash for current activities. If companies hold too little cash, they raise funds from external capital markets and liquidate existing assets (Opler et al., 1999). However, these activities are costly. If external funding sources are expensive or unavailable, a company can choose to save the costs of transactions by utilizing cash to finance possible future investment s to promote sales and investment opportunities (Opler et al., 1999; Denis and Sibilkov, 2010). Opler et al. (1999) observe that firms with better investment opportunities usually hold more cash since the cost of loss of investment is high for the firm. Nonetheless, according to Miller and Orr (1966) and Mulligan (1997), economies of scale exist in cash management. Larger companies can acquire funds in a less costly way as they have better access than small companies to financial markets and have economies of scale in fundraising and cash management (Opler et al., 1999; Dittmar et al., 2003).

6

financial slack (Myers and Majluf, 1984; Han and Qiu, 2007). Almeida et al. (2004) propose that financially constrained companies usually hold more cash for precautionary motives.

The third motive of corporate cash holding is established by the agency theory. Among all the assets of a company, liquid assets are easier to transfer into private profits (Myers and Rajan, 1998). Self-interested managers are incentivized to hold more cash so that they can gain private benefits easily (Jensen, 1986). The agency theory proposes that companies with more cash have agency conflicts when the cash is not invested in profitable projects (Al-Najjar and Clark, 2017). The empirical research indicates that agency problems exist especially in companies with too much cash, while managers in such companies invest in those projects with negative net present values, which can negatively affect the shareholders’ wealth or utilize corporate funds for the sake of personal gains instead of increasing shareholders’ dividends (Pinkowitz et al., 2006; Jensen, 1986). Based on agency theory, Dittmar et al. (2003) report that companies operating in nations with low shareholder protection tend to hold more cash. Additionally, governance can influence the market value of the reserves (Dittmar and Mahrt-Smith, 2007). In more recent years, Seifert and Gonenc (2018) employ a large internal sample and found that firms with strong country- or firm-level governance hold less cash. Meanwhile, agency problems can even lead to tight control over management through concentrated ownership (Anderson and Hamadi, 2016). The management, which is more entrenched, tends to hold excessive cash to prevent the discipline of the market (Opler et al., 1999).

7 2.2. Family control and cash holding

2.2.1. Agency conflict

There are many studies concerning family firms indicating that the agency conflicts between minority shareholders and controlling shareholders intensify in family firms (Liu, 2011; Lozano and Durán, 2017). In line with agency theory, cash holding by families is a way to accumulate family wealth. Liu et al. (2015) study family firms in China and found that they hold a significant amount of cash for tunneling rather than for investment or payment to shareholders. Many other studies have also focused on family enterprises in emerging markets and determined that controlling families have a strong incentive to seek private benefit and thus are likely to exploit minority shareholders (Claessens et al., 2000; Cronqvist and Nilsson, 2003). Boubaker and Derouiche (2015) find that excess cash depreciates for family-controlled firms. Moreover, according to the entrenched management literature, entrenched managers tend to avoid debt financing (Berger et al., 1996). According to agency theory, companies tend to keep low levels of leverage and are willing to hold more cash, as lower leverage companies are less subject to the monitoring of the external market (Ferreira and Vilela, 2004). Opler et al. (1999) also suggest that more entrenched management prefers to hold more cash to avoid market discipline. Family firms, generally viewed as entrenched managed firms, hold more cash. Thus, based on agency theory, family enterprises, which imply entrenched management, have more conflicts between the controlling shareholders and minority shareholders and hold more cash.

2.2.2. Precautionary motive

8

to precautionary motivation, Myers and Majluf (1984) demonstrate that, if there is information asymmetry between company insiders and outsiders, companies tend to hold cash instead of financing from the external market. Thus, information asymmetry exists between family firm insiders and outsiders, leading to a higher cost of external finance, and controlling shareholders tend to hold more internal funds because of the precautionary motive.

Moreover, families, as a unique class of investors, have control considerations in making cash policies. Family members want to maintain control of the company and maintain it as a legacy for future generations. Controlling shareholders with high value control are anti-takeover, and cash holding serves as a defense mechanism against takeovers (Faleye, 2004). Firms with anti-takeover amendments tend to hold more cash (Opler et al., 1999). Anderson and Hamadi (2016) study Belgian listed firms between 1991 and 2006 and proposed their doubts about the managerial rent extraction of cash holding in the context of a control-oriented financial system. They indicate that controlling shareholders who highly value control have more precautionary motives to h old excess cash. Considerations of controlling shareholders attach great value to control, the cost of financing equity is high because issuing shares when the company lacks cash dissipate control rights (Anderson and Hamadi, 2016). Thus, family controlling shareholders who highly value control are expected to have a higher precautionary motive to hold more cash.

Based on analysis above, it is assumed that the level of cash holding of family firms is higher than that of non-family ones. There are two possible motivations: first, the higher agency conflict between minority stakeholders and the controlling stakeholders in the family firms, and second, family firms have more precautionary motives because of higher information asymmetry and their high value of control.

Hypothesis 1a: Family control has a positive effect on corporate cash holding.

2.2.3. Strong governance

9

families include passing the company as a continuous concern to successors instead of transferring only their fortune. Therefore, compared with non-family firms, families have two obvious characteristics: the family’s concern for reputation and its interest in long-term survival. Families confront the concern over reputation because of the sustained presence of family in the company and its influence on other organizations (Anderson et al., 2003). Family shareholders also have a longer investment horizon (Anderson and Reeb, 2003). Moreover, family shareholders attach more importance to the long-term performance of the company. Therefore, family firms have a long-term commitment to retaining reputational capital and adopting the most favourable measures to generate firm value (Anderson and Hamadi, 2016).

Family control, as an internal company governance mechanism, is likely to have greater efficiency in terms of supervision and management (Wei et.al, 2011), which can reduce or even completely eradicate the agency problem between owners and managers because shareholders in family firms possess more motive to monitor managers. Based on this argument, Anderson and Reeb (2003) and Villalonga and Amit (2006) also indicated that founding family firms perform better than non-family firms in the U.S.

2.2.4. Transaction cost

10

invest less than non-family firms in high-risk areas, such as R&D projects, and more in fixed-asset capital expenditure projects, which are low–risk (Croci et al., 2011). However, Opler et al. (1999) have found that firms with riskier activities hold more cash. Moreover, R&D-driven innovation has long been regarded as one of the determinants of growth (Lyandres and Palazzo, 2016; He and Wintoki, 2016). I expect that, if family firms have less risk investment and invest less in R&D development, they should have fewer investment and growth opportunities than non-family firms. According to the transaction cost motive, companies with better investment opportunities and higher costs of raising debts would hold more cash to save transaction costs. Thus, I expect that family firms with higher leverage and lower-risk investment have less transaction cost saving motivation to hold more cash than non-family firms.

Based on analysis above, I expect that family firms hold less cash because they have less transaction cost-saving motivation, and family control brings strong governance.

Hypothesis 1b: Family control has a negative effect on corporate cash holding.

2.3. Multinational and cash holding of the family firm 2.3.1. Agency conflict and R&D investment

11

shareholders and minority shareholders. Therefore, the interaction between family control and multinationality can be assumed to sharpen the agency problem. Thus, it is expected that multinational family firms hold more cash than non-family enterprises.

According to the internalization theory, multinational firms expand overseas markets to take full advantage of intangible assets, which are difficult to exchange in external markets, and they tend to have more R&D investments. Opler et al. (1999) observe that firms with better investment opportunities hold more cash. The market-to-book rate is usually considered as a proxy for the developmental opportunities, and so as R&D expense, since R&D-driven innovation has long been regarded as a determinant of growth (Lyandres and Palazzo, 2016; He and Wintoki, 2016). Pinkowitz et al. (2016) find that companies in the U.S. with high R&D incentives hold more cash. Furthermore, firms with high R&D investments face higher information asymmetry so that have more difficult to raising external finance (Opler et al., 1999). Thus, I expect that, when family firms become multinational, with the increasing R&D investment, they have more investment opportunities and higher information asymmetry. Multinational family firms tend to hold more cash than domestic family firms because of the higher transaction cost and precautionary motive.

In light of the analysis above, the interaction between family control and level of multinationality positively affects cash holding. The hypothesis is as follows:

Hypothesis 2a: The positive (negative) relation between family control and corporate cash holding is stronger (weaker) with the level of multinationality.

2.3.2 Diversification, financial flexibility and strong governance

12

more diversified, and it needs to hold less cash than purely domestic firms. Meanwhile, economies of scale in cash management also lead multinational firms to hold less cash (Fernandes and Gonenc, 2016). In addition, multinational firms have better access to capital markets. Jang (2017) indicates that international diversification companies tend to have better financial flexibility, while multinational companies can lend from overseas financial markets more easily and can have lower capital costs than domestic firms, even during financial crises. When family firms become multinational, the diversifications benefits can help improve their access to capital markets, have efficient cash management and have economies of scale in cash management. As a result, multinational family firms hold less precautionary cash than domestic family firms. Furthermore, the internationalization-governance relationship suggest that the level of internationalization can have positive effects on corporate governance. When family firms become multinational, family control may also play a stronger governance role because of the concern over reputation and long-term commitment.

In light of the previous analyses, the interaction between family control and the level of multinationality has a negative effect on cash holding. The hypothesis raised is as follows:

Hypothesis 2b: The positive (negative) relation between family control and corporate cash holding is weaker (stronger) with the level of multinationality.

2.4. Country governance and cash holding of family firms

13

governance, laws are enforced to protect minority shareholders, and the administration and protective investment policies are rigorous and powerful. Therefore, effective governance not only allows managers to spend money in a wise manner but also motivates them to give excess cash to stockholders through dividends (Seifert and Gonenc, 2018). Controlling shareholders in family firms tend to hold more cash to extract private benefits from cash holdings, given that minority shareholders can monitor them less frequently. If the role of family control does not change with the quality of national governance, the possibility of controlling shareholders expropriating minority shareholders increases in countries with weaker governance, since there is less protection for minority shareholders’ rights. In this case, family firms hold less cash in countries with strong governance than with weak governance.

However, I expect that family shareholders play different roles in different governance environments. On the one hand, as an internal company governance mechanism, family control leads to more efficient supervision and management and reduces the agency problem between owners and managers (Wei et.al, 2011). When a firm in a country with weak investor protection, controlling shareholders have the ability and the incentives to control managers (La Porta et al., 1998; Shleifer and Vishny, 1986, 1997).

14

other side of the dividend policy, a higher cash level in countries with relatively weak shareholder rights cannot be inferred from the findings of La Porta et al. (2000). Moreover, the hypothesis that the dividends may substitute legal protection serves as a starting point to focus on reputation’s role in family firms’ financial policy.

Controlling shareholders have a long-term commitment to retaining reputational capital, and they adopt the most favourable measures to generate firm value (Anderson and Hamadi, 2016). If firm reputation shapes financial policy when it faces different qualities of country governance, family firms, which pay more attention to reputation than the non-family firms, play different roles in different qualities of country governance, as well. In countries with weak legal protection for minority shareholders, where having the reputation of excellent shareholder treatment is valuable, I expect that, in countries with weak country governance, controlling shareholders in family firms will not expropriate minority shareholders and will even distribute excess cash to stockholders via dividends due to the firm’s concern over reputation. To the contrary, in countries with stronger governance, reputation is valueless, and the requirement for a reputational mechanism is weaker, hence family firms should hold more cash than non-family firms. In this circumstance, I expect that family control plays different roles in countries with differing governance. In other words, the effect of family control on cash holding varies with the quality of a country's governance, and family firms hold less cash in countries with weaker governance than in those with stronger governance. In line with the argument above, if family control has a dominant effect, the combined effect of family control and the level of multinationality should also change with country governance. In other words, multinational family firms hold less cash in countries with weaker governance than in countries with stronger governance.

Hypothesis 3a: Family control plays significantly different roles with country governance.

Hypothesis 3b: The combined effect of family control and the level of multinationality significantly changes with country governance.

15

development, it is more expensive for firms to raise capital in countries with weaker governance (Pinkowitz et al., 2006). Thus, family firms hold more precautionary cash in such countries. From the view of the precautionary cash holding motive, there is no prediction suggesting that the role of family control changes according to a country's governance.

2.5. Cash holding and the level of trust across countries

In addition to countries’ formal institutions, informal institutions also influence corporate cash holdings. Some researchers have recognised trust’s importance for economic transactions. Keefer and Knack (1997), Temple and Johnson (1998) and Zak and Knack (2001) suggest that trust is positively correlated with economic growth. Recently, the findings of Dudley and Zhang (2016) and Xie and Xin (2015) support the agency hypothesis and indicate a significant, positive relationship between corporate cash holdings and trust. Furthermore, Dudley and Zhang (2016) also find an interaction effect of trust and country governance on cash holding.

Agency conflicts shape corporate cash policy. According to the agency hypothesis, in high-trust countries, agents are more likely to honour their own fiduciary obligations and comply with social behaviour norms. Meanwhile, minority shareholders view controlling shareholders as trustworthy, and they are less concerned about controlling shareholders transferring corporate cash into personal wealth (Xie, and Xin, 2015; Dudley and Zhang, 2016). Firms can thus accumulate greater cash reserves. However, if minority shareholders view controlling shareholders as less trustworthy, and they are more worried about expropriation, then firms prefer that management pay cash to shareholders instead of storing it (Xie and Xin, 2015; Dudley and Zhang, 2016). According to the agency hypothesis, trust has a positive influence on corporate cash holding.

When family firms in low-trust countries, minority shareholders view controlling shareholders as less trustworthy, and family firms face more pressure to disgorge cash. A

16

countries. In this case, family control is expected to play a stronger governance role in such countries. On the other hand, in high-trust countries, where controlling shareholders are perceived to be trustworthy by minority shareholders, then reputation is valueless, and family firms can use this kind of trust to accumulate greater cash reserves.

Furthermore, the agency hypothesis projects that the level of trust in different countries may determine the country governance’s different influences on enterprises’ cash holdings. In low-trust countries, firms face more pressure to hold less cash. Thus, the country’s governance mechanism is less valuable. In contrast, in a high-trust country, minority shareholders view controlling shareholders as trustworthy, and firms may use such trust to accumulate greater cash reserves. In this case, the function of a country’s governance is valuable. Thus, the effect of country governance is stronger in high-trust countries than in low-trust countries.

Considering that the effect of country governance on cash holdings varies according to countries’ level of trust, I expect that the combined effects of family control and country governance significantly change in low- or high-trust countries. In high-trust counties, the effect of country governance on family firms’ cash holding is stronger. The positive relation between family control and cash holding is stronger with country governance. Thus, family control and country governance are complimentary in high-trust counties. In low-trust countries, the reputation for excellent treatment of shareholders is valuable, family firms tend to hold less cash, then the negative relation between family control and cash holding is weaken with country governance.

Hypothesis 4a: The role of family control is significantly different between high- and

low-trust countries.

Hypothesis 4b: The combined effect of family control and country governance

significantly changes in low- or high-trust countries.

17

are more willing to provide capital to firms because corporate insiders are less likely to behave opportunistically. Meanwhile, trust reduces the cost of gaining information and increases the quality of available information about the firm. Greater trust makes the collection and dissemination of knowledge easier (Guiso et al., 2008), and increasing the collection and dissemination of knowledge reduces the information costs of issues equity and debt (Myers and Majluf, 1984). Thus, the lower cost of external financing in high-trust countries reduces the precautionary motive for firms to hold more cash. The information asymmetry in family firms is higher, since, in the view of outsiders, family members have more information about firms, and so the cost of external financing is higher for family firms. However, in high-trust countries, given the greater trust between insiders and outsiders, the problem of information asymmetry is mitigated, and the cost of external financing is reduced. Thus, from the precautionary savings hypothesis family firms hold less cash in high-trust countries and the role of family control do not change in high or low-trust countries.

3. Data and methodology 3.4. Data collection

18 3.5. Sample distribution

Table 1 presents the distribution of the total sample and the sample of family firms. In total, 31.5% of observations were conducted from the family firms. Panel A presents the distribution of observations by years; the number increases over time. Panel B presents the distribution of observations by industry, identified by the first two digits of Standard Industry Classification Codes (SIC). Nearly 60% of the observations are from the manufacturing industry. Finally, Panel C presents the distribution of observations by country; 30% of firm-year observations are firms in the United States and Japan.

Table 1. Sample distribution

This table presents the distribution of total sample and sample of family firms by years, industry and country. In panel B, the MAFF means agriculture, forestry and fishing industry.

Panel A: Sample distribution by years

Year N Percentage Family Percentage

19 Panel B: Sample distribution by industry

Industry N Percentage Family Percentage

MAFF 2471 0.89% 689 0.78% Mining 19064 6.84% 4936 5.61% Construction 10553 3.79% 3058 3.47% Manufacturing 166718 59.80% 48815 55.46% Wholesale 14782 5.30% 4486 5.10% Retail 16494 5.92% 5862 6.66% Services 48716 17.47% 20165 22.91% Total 278798 88011

Panel C: Sample distribution by country

Country N Family firm Country N Family firm

Argentina 542 74 Malaysia 10294 2441

Australia 15734 6108 Mexico 1095 307

Bahrain 106 40 Morocco 503 157

Bangladesh 164 0 Netherlands 1694 471

Brazil 2404 440 New Zealand 889 168

Bulgaria 168 24 Nigeria 303 48 Canada 6026 1769 Norway 1641 371 Chile 1433 283 Pakistan 832 30 China 24444 3554 Peru 758 146 Colombia 207 20 Philippines 1334 150 Cyprus 255 34 Poland 3218 1506 Egypt 670 210 Qatar 69 0 Estonia 102 25 Russian 1124 126

Finland 1686 441 Saudi Arabia 759 290

France 7932 4618 Singapore 6907 3924

Germany 7533 2861 Slovenia 200 5

Hong Kong 1409 482 South Africa 2840 470

Hungary 184 24 Spain 1383 720 India 20783 8263 Sweden 4100 1105 Indonesia 3547 305 Taiwan 17758 4703 Israel 2639 1331 Thailand 5107 1936 Italy 2557 1465 Tunisia 207 48 Japan 42560 10636 Turkey 2004 163

Jordan 586 301 United Kingdom 14596 6210

Korea 11837 7025 United States 40815 11443

Kuwait 534 166 Vietnam 2326 574

20 3.6. Methodology

3.3.1. Main variables

The dependent variable is corporate cash holding (cash). Based on previous studies (Opler et al., 1999; Fernandes and Gonenc, 2016; Chen et al., 2017; Seifert and Gonenc, 2018), the corporate cash holding is measured as a nature logarithm of cash and short-range investment over the net assets. The net assets equal the book value of total assets minus cash and short-range investment. For robustness, this thesis also uses an alternative measurement of cash holding, as the nature logarithm of cash and short-term investment over total asset (Gao et al., 2013;Chen et al., 2017).

Family control (Family) serve as an independent variable, is a dummy variable. According to Astrachan et al. (2002), the definitions of family firms rely on the classification about the ownership, including the involvement of the family in management, ownership and generational transferring. According to La Porta et al. (1999), families with most of the shares of a company are likely to take part in management. Faccio and Lang (2002) consider that a block of ownership held by an unlisted firm stands for the de facto family. Meanwhile, Anderson and Reed (2003) suggest that a family enterprise concerns a founding family with the share in the enterprise or members of founding family on the board of directors. Sraer and Thesmar (2007) define a company as a family firm if the generation or founder is in control. Therefore, based on the ability of a family to influence and manage the firm, a firm is defined as family controlled when the family is the largest ultimate owner with a stake of 10% or greater (Croci et al., 2011; Croci et al., 2012). If a firm is family controlled, the variable is coded as ‘1’; otherwise, it is coded ‘0’.

3.3.2. Moderators

21

multinationality, and this standard has been widely applied in the literature on global businesses (Ramírez and Tadesse, 2009; Fernandes and Gonenc, 2016). Researchers also consider a company as multinational with a dummy variable by identifying whether the degree of a company’s foreign sales is higher than 25% as consolidated in a certain year (Fernandes and Gonenc, 2016) and whether the company has any foreign subsidiaries (Gao and Chou, 2015). This thesis aims to examine the moderating effect of the different degrees of multinationality. Thus, instead of using a dummy variable, the degree of multinationality is evaluated as foreign sales divided by total sales, following prior literature (Ramírez and Tadesse, 2009; Fernandes and Gonenc, 2016).

Following by Seifert and Gonenc (2018), the World Bank Governance Indicators (WBGovIndex) are employed to evaluate the country governance in this thesis. The WBGovIndex is a comprehensive definition of country governance evaluated by the World

Bank. Country governance is gathered by the WBGovIndex according to the six indicators, including government effectiveness, voice and accountability, rule of law, political stability, control of corruption and regulatory quality.

The level of country trust is the other moderator. Helliwell and Putnam (1995) confirm that social trust refers to the degree about whether and to what extent people trust strangers. Prior studies in trust literature (La Porta et al., 1997; Ahern et al., 2015; Dudey and Zhang, 2016) evaluates social trust according to the following question from the World Values Survey (WVS):

Generally speaking, would you say that most people can be trusted, or that you need to be very careful in dealing with people?

22 3.3.3. Control variables

Firm level: There is an extensive range of company features presented in previous literature (Opler et al., 1999; Dittmar et al., 2003; Harford et al., 2008; Seifert and Gonenc, 2018) that can affect the corporate cash holdings. Following the classical cash holding research by Opler et al. (1999), this study controls the firm size, net working capital, leverage, cash flow, capital expenditure, R&D intensity and whether a firm pays a dividend.

Larger enterprises are expected to gain funding more easily as they have less demand for cash because of the economy of scale (Dittmar et al., 2003). Therefore, this thesis controls the firm size, measured as the natural logarithm of total assets (Opler et al., 1999). Net working capital (NWC), considered a substitute for cash, can influence cash holdings negatively. Net working capital equals the current asset minus the current liability minus cash equivalent and cash scaled by total asset. Leverage can influence cash holdings negatively because the interest payments would support the debt and reduce the cash holdings. Leverage (LEV) refers to the financial debt over total assets. Cash flow is added to cash and thus affects cash holdings positively. Cash flows (CF) equal net income plus depreciation scaled by the book value of total assets. Capital expenditures (Capex) ceteris paribus should reduce cash balances as they represent a cash outlay; they are calculated as the rate of capital expenditures to total assets. Firm tend to hold more cash to support R&D activity,R&D intensity (RD) is evaluated as the rate of R&D expenditures scaled by total assets. The amount of cash reserves decreases if a company pays a dividend. Payer is a dummy variable and equals ‘1’ if a dividend per share greater than 0 and ‘0’otherwise.

23

growth is used as a proxy for corporate development, and it can be evaluated as the annual development ratio of a company’s sales. On the other hand, following Fernandes and Gonenc (2016), Tobin’s Q is also used as a proxy for corporate investment opportunities. Equity issuance (EquityIssue) and debt issuance (DebtIssue) are closely related to the financing constraints of the company. Firms with equity and debt issuance hold higher cash reserves. Following Fernandes and Gonenc (2016), the equity and debt issuance is controlled.

Country level: the implications of several country-level variables have been documented as important determinants of cash holdings. Dittmar et al. (2003) control private credit as a percent of GDP. High levels of private credit can influence the cash holdings negatively because the companies can find credit easily when needed and therefore do not have to stockpile it (Seifert and Gonenc, 2018). In addition, economic size and growth have been viewed as important country-level factors, and they can affect corporate cash holding. Based on the studies of Kalcheva and Lins (2007) and Seifert and Gonenc (2018), this thesis controls current annual GDP growth, which is a proxy of economic growth.

3.3.4. Regression analysis

One of the key issues this thesis is to explore the correlation between corporate cash holding and family control and how the level of multinationality, country governance and trust of country moderate this correlation or not. The fixed effects of the country are included in the control of the unobserved country-level determinants of corporate cash policy. The year and industry fixed effect is also included. OLS regression is applied in this paper, and the regression model is composed as follows:

𝐼𝑛(𝐶𝑎𝑠ℎ)it= 𝛽0+ 𝛽1𝐹𝑎𝑚𝑖𝑙𝑦𝑖𝑡+ 𝛽2𝑀𝑁𝐶𝑖𝑡+ 𝛽3𝑊𝐵𝐺𝑜𝑣𝐼𝑛𝑑𝑒𝑥𝑘𝑡+ 𝛽4𝐹𝑎𝑚𝑖𝑙𝑦𝑖𝑡∗ 𝑀𝑁𝐶𝑖𝑡+ 𝛽5𝐹𝑎𝑚𝑖𝑙𝑦𝑖𝑡 ∗ 𝑊𝐵𝐺𝑜𝑣𝐼𝑛𝑑𝑒𝑥𝑘𝑡+ 𝛽6𝑀𝑁𝐶𝑖𝑡∗ 𝑊𝐵𝐺𝑜𝑣𝐼𝑛𝑑𝑒𝑥𝑘𝑡+ 𝛽7𝐹𝑎𝑚𝑖𝑙𝑦𝑖𝑡∗ 𝑀𝑁𝐶𝑖𝑡 ∗ 𝑊𝐵𝐺𝑜𝑣𝐼𝑛𝑑𝑒𝑥𝑘𝑡+ 𝛽₈𝑠𝑖𝑧𝑒𝑖𝑡+ 𝛽₉𝑁𝑊𝐶𝑖𝑡+ 𝛽₁₀𝐿𝐸𝑉𝑖𝑡+ 𝛽₁₁𝐶𝐹𝑖𝑡+ 𝛽₁₂𝐶𝑎𝑝𝑒𝑥_𝑖𝑡 + 𝛽₁₃𝑅&𝐷𝑖𝑡+ 𝛽14𝑃𝑎𝑦𝑒𝑟𝑖𝑡+ 𝛽15𝑆𝑎𝑙𝑒𝑠𝑔𝑖𝑡+ 𝛽16𝐷𝑏𝑡𝑒𝐼𝑠𝑠𝑢𝑒 𝑖𝑡+ 𝛽17𝐸𝑞𝑢𝑖𝑡𝑦𝐼𝑠𝑠𝑢𝑒𝑖𝑡 + 𝛽₁₈𝐶𝐹 𝑣𝑎𝑟𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦 _𝑖𝑡+ 𝛽₁₉ 𝐴𝑞_𝑖𝑡+ 𝛽₂₀𝐺𝐷𝑃𝑔 _𝑖𝑡+ 𝛽₂₁𝑃𝑟𝑖𝑐𝑎𝑡𝑒 𝑐𝑟𝑒𝑑𝑖𝑡 𝑖𝑡+ 𝐼𝑛𝑑𝑗+ 𝑌𝑟𝑡 + 𝐶𝑛𝑡𝑟𝑦_𝑘+ 𝜀𝑖𝑡

Where 𝐼𝑛𝑑𝑗 is a set of industry dummies, 𝑌𝑟𝑡is a set of yearly dummies and 𝐶𝑛𝑡𝑟𝑦𝑘 is the

24

Table 2. Measurement and source of variables

Variables Symbol Measurement Database

In (Cash holding)

In(Cash) Natural logarithm of cash and short-range investment to the net asset. Net asset equal to book value of total assets minus cash and short-range investment.

Compustat

Family control

Family Dummy variable taking the value 1 if family or an individual is the largest ultimate owner with a stake of 10 percent or greater in the firm, otherwise 0.

Datastream

Multinational MNC The ratio of foreign sales to total sales. Datastream Firsm size Size The natural logarithm of book value of total assets in USD. Datastream Net working

capital

NWC The current asset minus the current liability minus cash scaled by book value of total asset.

Compustat Net debt

issuance

DebtIssue Net debt issuance (Long-term debt issuance minus long-term debt reduction) divided by book value of total assets.

Compustat Net equity

issuance

EquityIssue Net equity issuance (Sale of common and preferred stock minus purchase of common and preferred stock) divided by book value of assets.

Compustat

Leverage LEV The ratio of book value of total long-term debt plus short-term debt to book value of total assets.

Compustat Cash flow CF The ratio of net income add depreciation to book value of

total assets.

Compustat Cash flow

volatility

CF volatility Standard deviation of cash flow to the book value assets over the last three years.

Compustat Capital

expendture

Capexp The ratio of capital expenditures to book value of total assets.

Compustat R&D

intensify

RD The ratio of Research & Development expenditures to book value of total assets.

Compustat Acquisitions Aq Acquisition to book value of total assets. Compustat Pay dividend Payer Dummy variable taking the value of 1 if dividends per share

greater than zero, otherwise 0.

Datastream Sale Growth Sale Percentage change in sales from t -1 to t. Compustat Tobin’s Q TQ Tobin’s Q ratio is defined as companies’ market value of

assets over the assets’ book value. The market value of assets as the book value of assets plus market capitulation minus the common stock’s book value.

Compustat and Datastream Trust level of

country

Trust The percentage of people who responded that most people can be trusted based on the WVS question: “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?”

World Values Survey

World Bank governance index

WBGovIndex WBGovIndex is measured by the weight average of six World Bank governance indicators, which are voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption.

World Bank

GDP growth GDP Current annual GDP growth. World Bank

Private credit (%GDP)

Private credit Private credit by deposit money banks and other financial institutions to GDP. Credit from deposit taking financial institutions to the private sector as a percentage of GDP.

25

4. Empirical results and analysis 4.1. Descriptive statistics

Table 3 presents the descriptive statistics for the major firm-level variables and country-level variables by country, and Table 4 provides descriptive statistics generated by all samples. In terms of trust statistics, the overall trust of countries ranges from 5.8% (Philippines) to 73.7% (Norway) with a mean of 34.7%. As for the statistics of country governance, Finland (1.87) has the strongest country governance, while Nigeria (-1.11) has the weakest. The mean of the country governance of all the samples is 0.81.

Table 4 presents the mean, median, minimum, maximum and standard deviation values of the dependent variable cash holding, the independent variable family control and a series of control variables. Panel A presents the descriptive statistics of all the samples. The observations in panel B are classified as family firms and non-family ones. The third column in Panel B indicates whether the characteristics of family firms significantly differ from those of non-family firms. Panel C presents a correlation matrix of firm-level variables, and, finally, Panel D displays a correlation matrix of country-level variables.

26

Table 3. Sample country and selected variables

This table summarizes major variables by country. Cash is cash and short-range investment to the net asset. The definition of other variables is shown in Table 2 in detail.

Country N Cash MNC Trust WBGov

27 Slovenia _{200 0.07 0.24 0.187 0.942 1.102 0.678} South Africa _{2840 0.135 0.124 0.174 0.31 2.876 1.411} Spain _{1383 0.096 0.35 0.238 0.963 1.59 1.392} Sweden _{4100 0.174 0.323 0.63 1.761 2.266 1.16} Taiwan _{17758 0.211 0.215 0.273 0.941 .} Thailand _{5107 0.115 0.083 0.367 -0.206 3.717 1.213} Tunisia _{207 0.124 0.018 0.155 -0.214 2.436 0.74} Turkey _{2004 0.111 0.052 0.117 -0.128 5.217 0.443} United Kingdom _{14596 0.178 0.333 0.3 1.457 1.761 1.539} United States _{40815 0.221 0.21 0.365 1.308 1.966 1.842} Vietnam _{2326 0.146 0.003 0.448 -0.491 5.958 1.043} Total _{278798 0.176 0.172 0.347 0.813 3.388 1.29}

Family firms have both a lower mean of capital expenditure and a lower mean of R&D intensity compared to non-family firms. However, the difference between capital expenditure is smaller than R&D intensity. To some extent, this result is consistent with the findings of Croci et al. (2011), which show that family enterprises invest less than non-family ones in high-risk sectors, such as R&D projects, but more in fixed-asset capital expenditure projects, which are low–risk. The mean of acquisitions for the total sample is 0.8%. However, the mean of acquisitions for family firms (0.75%) is lower than both non-family firms and the total sample. Family firms have higher mean equity issuance but lower mean debt issuance than non-family firms. The mean of Tobin’s Q for the total sample is 1.82, while for family firms, it is lower (1.76) than for non-family ones (1.85). However, family firms (0.14) have a higher mean of sale growth compared to non-family firms (0.13). The mean of payer of family firms (0.52) is much lower than for non-family firms (0.57), indicating that family firms are more reluctant to pay dividends to their shareholders than non-family firms. Additionally, family firms have lower means of cash flow but higher cash flow volatility than non-family firms.

For descriptive statistics of the country-level variables, the mean level of country trust is 34.7%, and the mean of the level of country governance quality is 0.81. The differences between country-level variables show that family firms are more likely to be located in countries with better governance and with slower economic growth.

28

Table 4. Sample statistics

This table summarizes the mean, median, min, max and standard deviation values of dependent variable cash holding the independent variables family control and a series of control variables. Panel A presents the descriptive statistics of total sample. Panel B classifies the observations as family firms and non-family ones and show the differences. Cash is cash and short-range investment to the net asset. The definition of other variables is shown in Table 2 in detail. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Panel A. Total Sample

Variable N Mean Median Min Max Std. Dev.

Firm-level variable Cash 278,798 0.180 0.120 0.002 0.904 0.183 Family 278,798 0.316 0.000 0.000 1.000 0.465 MNC 278,798 0.170 0.000 0.000 1.000 0.286 Size 278,798 5.119 5.069 -0.329 10.079 1.955 NWC 278,798 0.001 0.032 -3.118 0.537 0.343 LEV 278,798 0.209 0.171 0.000 1.000 0.201 Capex 278,798 0.048 0.029 0.000 0.361 0.060 R&D 278,798 0.021 0.000 0.000 0.427 0.060 Aq 278,798 0.008 0.000 0.000 0.220 0.031 TQ 278,798 1.818 1.198 0.305 21.445 2.230 CF 278,798 0.036 0.061 -0.426 0.206 0.133 EquityIssue 278,798 0.041 0.000 -0.066 1.061 0.148 DebtIssue 278,798 0.001 0.000 -0.097 0.173 0.025 Sale 278,798 0.134 0.040 -0.950 4.342 0.587 CF volatility 278,798 0.062 0.019 0.000 1.604 0.166 Payer 278,798 0.556 1.000 0.000 1.000 0.497 Country-level variable Trust 278,798 0.347 0.365 0.058 0.737 0.122 WBGovIndex 278,798 0.809 1.177 -1.265 1.970 0.765 Government effective 278,798 1.112 1.396 -1.215 2.437 0.717 Regulatory quality 278,798 0.937 1.132 -1.352 2.261 0.765 Political stability 278,798 0.317 0.518 -2.810 1.760 0.794 Control of corruption 278,798 0.915 1.271 -1.434 2.465 0.931 Rule of law 278,798 0.962 1.312 -1.407 2.100 0.812 Voice and accountability 278,798 0.610 0.988 -1.907 1.784 0.925

GDP 261,029 3.434 2.807 -14.724 33.736 3.292

29

Table 4. (Continue)

Panel B. Comparison of family sample and non-family ones.

Family Non-Family Difference

Variable Mean Median Mean Median Mean Median

30

Table 4. (Continue)

Panel C correlation matrix of frim-level variables

1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 Family 1.000 MNC -0.037*** 1.000 Size -0.254*** 0.264*** 1.000 NWC 0.029*** 0.015*** 0.023*** 1.000 Leverage -0.009*** -0.020*** 0.147*** -0.305*** 1.000 Capex -0.008*** -0.006*** 0.015*** -0.045*** 0.045*** 1.000 RD -0.002 0.109*** -0.150*** -0.093*** -0.098*** -0.080*** 1.000 Aq -0.010*** 0.083*** 0.087*** -0.002 0.021*** -0.034*** 0.010*** 1.000 TQ -0.017*** 0.018*** -0.111*** -0.467*** -0.037*** 0.046*** 0.221*** 0.023*** 1.000 CF -0.028*** 0.056*** 0.342*** 0.233*** -0.129*** 0.067*** -0.340*** 0.029*** -0.115*** 1.000 EquityIssus 0.030*** -0.038*** -0.298*** -0.081*** -0.109*** 0.099*** 0.242*** 0.050*** 0.244*** -0.415*** 1.000 DebtIssues -0.012*** 0.008*** 0.039*** -0.021*** 0.097*** 0.062*** 0.032*** 0.230*** 0.028*** -0.047*** -0.007*** 1.000 Sale 0.006*** -0.011*** -0.043*** 0.012*** -0.036*** 0.094*** 0.038*** 0.106*** 0.070*** 0.006*** 0.140*** 0.046*** 1.000 CF volatility 0.015*** -0.009*** -0.255*** -0.192*** 0.012*** -0.010*** 0.239*** 0.010*** 0.209*** -0.324*** 0.279*** 0.030*** 0.088*** 1.000 Payer -0.043*** 0.043*** 0.363*** 0.155*** -0.112*** 0.018*** -0.188*** -0.025*** -0.088*** 0.440*** -0.234*** -0.012*** -0.059*** -0.250*** 1.000

Panel D correlation matrix of country -level variables

1 2 3 4

Trust 1.000

WBGovIndex 0.106*** 1.000

GDPg 0.102 *** -0.587*** 1.000

31

acquisition and networking capital are not significantly correlated. The correlation of all other control variables is significantly under 0.6. All the results indicate that there is no strong correlation with independent variables or control variables.

4.2 Regression analysis

4.2.1 Cash holdings of family firms and multinationality

This section presents the findings of the relation between family control and cash holdings, and the interaction effect between family control and the level of multinationality on cash holding at the firm level. As Table 5 shows, there are five regression results: model 1 used only control variables, while model 2 used family and control variables. In model 3, only multinationality variables and control variables were used; in model 4, both family control and multinationality variables were employed; and finally, model 5 used family control and multinationality variables and interaction. All the models include indices of country, industry and year. This approach highlights the individual impact of family control and multinationality on cash holding, and this difference appears when analyzing the interaction between those two variables.

32

Table 5 Family control, multinationality and cash holding

This tables presents the estimations of the effect of family control on cash holding and the moderating effect of multinational on cash holding of family firms. The independent variable is In (cash) which equals natural logarithm of cash and short-range investment to the book value of total asset. The definition of other variables is shown in Table 2. Robust standard errors in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

model1a model1 model2 model3 model4

Family 0.035*** 0.034*** 0.052*** [0.011] [0.011] [0.012] MNC 0.282*** 0.282*** 0.317*** [0.019] [0.019] [0.021] Family *MNC -0.110*** [0.031] Size -0.057*** -0.054*** -0.072*** -0.069*** -0.069*** [0.004] [0.004] [0.004] [0.004] [0.004] NWC -0.324*** -0.325*** -0.325*** -0.326*** -0.326*** [0.021] [0.021] [0.021] [0.021] [0.021] LEV -2.380*** -2.385*** -2.372*** -2.376*** -2.377*** [0.034] [0.034] [0.034] [0.034] [0.034] Capex -1.548*** -1.551*** -1.550*** -1.553*** -1.554*** [0.076] [0.076] [0.076] [0.076] [0.076] RD 4.088*** 4.094*** 3.985*** 3.990*** 3.992*** [0.113] [0.113] [0.114] [0.114] [0.114] Aq -3.214*** -3.213*** -3.237*** -3.237*** -3.239*** [0.104] [0.104] [0.103] [0.103] [0.103] TQ 0.039*** 0.038*** 0.037*** 0.037*** 0.037*** [0.003] [0.003] [0.003] [0.003] [0.003] CF 0.266*** 0.260*** 0.256*** 0.250*** 0.249*** [0.041] [0.041] [0.041] [0.041] [0.041] EquityIssue 1.683*** 1.684*** 1.696*** 1.697*** 1.697*** [0.029] [0.029] [0.029] [0.029] [0.029] DebtIssue 1.876*** 1.878*** 1.913*** 1.915*** 1.917*** [0.133] [0.133] [0.132] [0.132] [0.132] Sale 0.036*** 0.035*** 0.037*** 0.037*** 0.037*** [0.005] [0.005] [0.005] [0.005] [0.005] CF volatility 0.305*** 0.308*** 0.294*** 0.297*** 0.297*** [0.029] [0.029] [0.029] [0.029] [0.029] Payer 0.170*** 0.168*** 0.173*** 0.171*** 0.171*** [0.012] [0.012] [0.012] [0.012] [0.012] Constant -2.552*** -2.566*** -2.492*** -2.506*** -2.507*** [0.144] [0.144] [0.143] [0.143] [0.142]

Year fixed Yes Yes Yes Yes Yes

Industry fixed Yes Yes Yes Yes Yes

Country fixed Yes Yes Yes Yes Yes

R-squared 0.385 0.385 0.387 0.387 0.387

Observations 278798 278798 278798 278798 278798

33

and multinationality can negatively impact cash holding, which supports Hypothesis 2b. Which motivations mentioned above plays dominant effect is analyzed in part 4.2.2.

The coefficients of multinationality are positive, and this result differs from that of Fernandes and Gonenc (2016), which shows a negative relation between both geographic and industrial diversification and cash holding.1 _{The coefficients of the control variables are}

consistent with prior findings (Opler et al., 1999; Dittmar et al., 2003; Harford et al., 2008; Seifert and Gonenc, 2018). Larger firms with higher net working capital, higher leverage, higher capital expenditure and more acquisition activities hold lower cash reserves. The characteristics of the company, such as higher cash flow, equity and debt issues, and greater investment opportunity, lead the company to have more cash holding. All models show that firms with higher R&D intensity hold more cash, and firms with better sale growth and higher cash flow volatility hold more cash.

4.2.2 Family cash holding and country governance

This section analyses the regression results concerning how the role of family control varies across firms located in countries with different governance. There are four regression results. Only family and control variables were used in model 1, while in model 2, family control, multinational variables and interaction were used; in model 3, the interaction between multinational and country governance was employed; and finally, model 4 included the interaction between family control, multinationality and country governance.

As shown in table 6, country governance is negative and significantly related to corporate cash holding in all models.The level of multinationality is positive and significantly related to corporate cash holding in all models. However, the effect of family control differs across models. In model 1, the coefficient of family is positive (0.056) and significant at 1%, which indicates that the effect of family control is positively related to cash holding. However, in model 2, the coefficient of family becomes negative (-0.046), which is significant at 1%. The interaction between family control and country governance is positive (0.129). Such results indicate that the negative effect of family control on cash holding becomes less negative in

34

Table 6. Country governance and cash holding of family firms.

This tables presents the estimations of the moderating effect of country governance on the relation between family control and cash holding. The independent variable is In (cash) which equals natural logarithm of cash and short-range investment to the book value of net asset. The definition of other variables is shown in Table 2. Robust standard errors in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Model1 Model2 Model3 Model4

35

countries with stronger governance. Family firms in high-governance countries hold more cash than non-family firms, whereas they hold less cash than non-family firms in weak-governance countries. The different signs of estimated coefficients of the family control show that the role of family control change with levels of country governance. Family control may play strong governance in weak-governance countries. Family control can play a stronger internal governance role when external governance is weak. Hypothesis 3a is supported. As argued in Hypothesis 3a, since families are more concerned with reputation, in countries with weaker governance, controlling shareholders in family firms do not expropriate minority shareholders and even distribute excess cash to stockholders via dividends. Family firms thus hold less cash than non-family firms in weak-governance countries. In contrast, reputation is valued less in countries with stronger governance, and the controlling shareholders in family firms still hold more cash than in non-family firms. The positive sign of the interaction between family control and country governance supports this argument.

As shown in model 3, the coefficient of the interaction between multinational and country governance is negative, which indicates that multinational firms hold less cash in countries with strong governance than in countries with weak governance. This result aligns with the study of Fernandes and Gonenc (2016); Chen and Chen (2012) have shown that the diversification strategies of these companies is more efficient in reducing cash holdings in countries with better investor protections and institutional development. Multinational firms from countries with stronger governance have better access to capital markets, and therefore can effectively utilize diversification benefits. Model 4 shows the regression result of the two-way and three-way interactions. The coefficient of the interaction between family control and multinationality is negative; however, it is not significant. The interaction between family control, multinationality and country governance is negatively related to cash holding, which is at 10% significance level. These results demonstrate that multinational family firms hold less cash in countries with strong governance than in countries with weak governance. The

Year fixed Yes Yes Yes Yes

Industry fixed Yes Yes Yes Yes

Country fixed Yes Yes Yes Yes

R-squared 0.384 0.384 0.384 0.385

36

combined effect of family control and multinationality does not change with the quality of country governance. Therefore, Hypothesis 3b is rejected. Considering the analysis of the results in models 2, 3 and 4 together, it can be concluded that, supporting the agency problem motive, the role of family control differs in countries with different levels of governance, and the governance role of family control is stronger in countries with weaker governance. However, when the family becomes multinational, multinationality then plays a significant, important role in cash holding. Multinationality brings diversification benefits, which can lead family enterprises to have better access to capital markets and to economies of scale in cash management, which supports the view of the precautionary motive for holding cash.

4.2.3 Trust and cash holdings of family firms

To explore how the level of country trust affects the cash holding of family firms, the observations are classified as located in high-trust countries (Panel A) or low-trust countries (Panel B). The critical value is the median level of country trust. Two reasons support why the subsample should be adopted to conduct such analysis. On one hand, trust is a time-invariant variable, so country-fixed effect is not available, but country-fixed effect need be used. On the other hand, it is easy to demonstrate the difference between the estimated coefficient of family control and the interactions in high- or low-trust countries through this approach. There are four regression results in both panels: in model 1, only family and control variables were used; in model 2, family control, multinational variables and interaction were used; in model 3, the interaction between multinational and country governance was employed; and finally, model 4 concerned the interaction between family control, multinationality and country governance.

37

country governance is negatively related to cash holding in either high- or low-trust countries. However, the effect of country governance is not significant in low-trust counties. This result shows that country governance does not pay a significant role in low-trust counties.

The results of model 2 in two subsamples are more interesting. The coefficient of family control is positive (0.160) in Panel A, model 2, and the coefficient of family control is negative in Panel B, model 2. However, the sign of interaction between family control and country governance in the both subsamples is positive. These results show that in high trust counties, the positive correlation between family control and cash holding becomes stronger in countries with strong governance, family control and country governance are complimentary; in low trust countries, the effect of family control changes from negative to positive when country governance is stronger, the negative correlation between family control and cash holding becomes weaker in countries with strong governance. Thus, the combine effect of family control and country governance plays significant different role in low trust countries. To some content, the combine effect of family control and country governance plays a strong governance role in low-trust and weaker governance countries.

38

Table 7. Country trust and cash holding of family firms.

This tables presents the estimations of the moderating effect of country governance on the relation between family control and cash holding in high-trust countries (Panel A) and low-trust countries (Panel B). The independent variable is In (cash) which equals natural logarithm of cash and short-range investment to the book value of net asset. The definition of other variables is shown in Table 2. Robust standard errors in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively.

Panel A. High Trust

Model 1 Model 2 Model 3 Model 4

39

Constant -1.720*** -1.719*** -1.723*** -1.722***

[0.192] [0.191] [0.191] [0.191]

Year fixed Yes Yes Yes Yes

Industry fixed Yes Yes Yes Yes

Country fixed Yes Yes Yes Yes

R-squared 0.38 0.381 0.38 0.381

Observations 114349 114349 114349 114349

Panel B. Low Trust

Model 1 Model 2 Model 3 Model 4