The effects of the origin of shareholders on cash dividends

(1)

The effects of the origin of

shareholders on cash dividends

Is there a home bias in France and the United Kingdom?

Abstract

This study investigates the effects of the origin of a firm’s largest shareholder on its dividend policy. It studies whether a home bias exists regarding the payout. This study compares firms from a common law country (United Kingdom) and a civil law country (France). The main finding is that firms from France and the UK with a largest domestic shareholder are more likely to pay out dividend; moreover, total cash dividend paid when the largest shareholder is domestic is significantly lower in France, but not in the UK. A robustness check with the power of the largest shareholder also indicated that domestic shareholders are more inclined to pay dividend.

Key words: cash dividends, home bias, ownership concentration, payout policy Hessel Kersting

s1534742

MSc International Business & Management – International Financial Management University of Groningen, Faculty of Economics and Business

Uppsala Universitet, Faculty of Economics & Business

February 2011

1st supervisor: Dr. J.H. von Eije

(2)

-2-

1. Introduction

This thesis studies the effects whether the origin of the largest shareholder of a firm influences the dividend policy. To be more precise, it will compare the effects of domestic and foreign largest shareholders on 1) the decision whether to pay dividends or not, and 2) the level of cash dividends paid by a paying firm. In order to investigate this, data are gathered from firms listed in France on the Euronext Paris Stock Exchange and firms listed in the United Kingdom, listed on the London Stock Exchange. The effects will be tested for the time period from 2002 until 2009.

Dividends have received widespread attention since the dividend irrelevance theory by Modigliani and Miller (1958) was developed. More than four decades after the dividend puzzle was coined, Allen and Michaely (2003) conclude that after years of research, the effects of certain factors on dividends are still unclear. Firstly, agency theory has two implied interactions with dividends. First of all, Easterbrook (1984) argues that agency costs could decline due to higher dividend payments that make it necessary to resort to external financing, which in turn generates additional monitoring by external capital markets. Moreover, it is argued that dividends can reduce agency conflict because dividend payments lower the funds that are at managers’ discretion (Jensen, 1986). More recently, studies have investigated the effects of ownership and control in combination with agency costs and dividends. In 1997, it was argued by Shleifer and Vishny that owners that have nearly full control of a firm, prefer to redistribute private benefits that are not shared with minority shareholders. This is classified as expropriation of minority shareholders. Earlier theories, however, stated the opposite, arguing that the presence of large shareholders could limit the agency conflicts between shareholders and management because of monitoring abilities that large shareholders have (Shleifer and Vishny, 1986).

This study goes beyond the previously mentioned studies and tries to link dividends to another puzzle that is present within the field of finance, the equity home bias puzzle. French and Poterba (1991) were among the first authors that observed the equity home bias puzzle. It describes the fact that investors in most countries only hold limited amounts of foreign equity. This notion conflicts with existing literature that posits that international diversification could substantially improve the returns of national portfolios.

(4)

-4- would lead to higher risks, caused by for example exchange rate exposure, higher costs or information asymmetries (Amadi, 2004). Therefore, it is argued later on that foreign investors want to be compensated for their ‘riskier’ investments. Paying out more dividends could do this.

The following news article from Industry Week1 is also characterizing for international

firms and can help to explain a possible dividend home bias.

“General Motors Europe announced to unions on Jan. 21 that it is to close its Opel auto production plant in Antwerp, in a blow that Belgian industry said would cost 5,000 jobs all told.”

This news article describes the fact that a Belgium factory was forced to close because a foreign investor (Opel/General Motors) thought it wasn’t profitable enough. In Germany, no factory was closed, but some employees were laid off. This study argues that this behaviour might be the reason why dividend pay-outs are home biased. International firms seem to be more concerned with their own country. Therefore, they might put more effort in trying to ‘save’ jobs and income for their home country. Later on, I argue that this is exactly why firms with domestic shareholders are more likely to be a dividend-paying firm, but also pay out lower dividends.

This study focuses on the effects of the origin of a firm’s largest shareholder on dividends. First of all, it will investigate whether there exists a home bias regarding the likelihood to pay out dividend by companies. Moreover, the effects of home and foreign owners on the amount of dividends paid will be investigated.

The effects will be tested for a civil law country (France) and a common law country (United Kingdom). La Porta et al. (2002) argue that firms in common law countries generally pay out more dividends, due to better minority shareholder protection. Furthermore, France will be investigated because it can be seen as the first example of a civil law system, also known as Napoleonic law. The United Kingdom is chosen because it represents the major common law country within Europe.

The existence of a home bias with regards to the dividend payout has, to my best knowledge, never been investigated before. So far, only Bena and Hanousek (2006) looked into the origin of shareholders of firms, using a sample of Czech firms. They conclude that for firms with foreign owners, the target dividend payout ratio is a lot higher (0.46 for foreign owners versus 0.12 for Czech owned firms). They also find that

1_{The full article can be found at:}

(5)

-5- foreigners use dividends more often to distribute profits. However, the authors think this is mainly caused by the fact that the country has a relatively weak corporate governance environment. In this study, it will be investigated for a country with weak minority shareholder protection (France) and strong minority shareholder protection (United Kingdom).

The main finding of this study is that there indeed exists a dividend home bias. I find that for both French and UK firms, the presence of a domestic largest shareholder has a positive influence on the decision whether to pay dividends or not. Also when this is tested in combination with the percentage of share held, I find evidence for both countries that larger domestic shareholders have a positive impact on the likelihood to pay out. With respect to the level of dividends that are paid out, I find that for French firms where the largest shareholder is domestic, total cash dividends paid is significantly lower than for French firms with a foreign largest shareholder. This finding cannot be observed in the UK.

This study adds to the existing literature by a research that, to the best of my knowledge, never has been investigated before. Namely, the investigation of the relation between the possible home bias and the dividend payout of firms.

(6)

-6-

2. Literature review

The literature is reviewed in five parts. The first part deals with the relation between agency costs and dividends. The second part focuses on the link between ownership concentration and the effects on a firm’s dividend policy. Thirdly, the theory on the equity home bias is introduced. In the fourth section, the theory on dividends and the home bias are linked. Next, I focus on the differences between the two countries that are investigated, France and the United Kingdom. Lastly, the hypotheses that combine the theory are developed.

2.1. Agency costs & Dividends

Dividends have gained widespread attention in the literature since the dividend irrelevance theory was proposed by Modigliani and Miller (1958). They argued and discovered that that value of a firm is unaffected by its capital structure. The value of the firm is influenced by the ability to make profits and the risks on the investments. Some authors argue that dividend policy doesn’t affect the stock price, whereas others argue the opposite. Numerous authors have tried to come up with various answers to the dividend puzzle (e.g. Black, 1976). Many of these answers involve the agency theory, which is centered on the traditional agency relations between a firm’s stakeholders and dividend payments. Originally, it is believed that the main duty of a manager is to maximize the shareholders’ wealth; however, another incentive might be to maximize their personal wealth. Jensen and Meckling (1976) described the consequences that agency problems could have on firms. They argue that managers might tend to expropriate a firm when the link between control and ownership loosens.2

Since the theory proposed by Jensen and Meckling (1976), several authors have tried to link the agency theory to dividend policies. Rozeff (1982), Easterbrook (1984) and Jensen (1986) were among the first who linked agency theory to dividends. These authors argue that the interests of managers and shareholders are in conflict because managers act in their own interest, at the expense of the shareholders. Rozeff (1982) finds a negative relationship between dividend payout and his proxies for agency problems; percentage of shares held by insiders and the total number of shareholders. He argues that dispersed ownership causes monitoring to become too costly for a single shareholder. Therefore, they make use of the capital markets to monitor managers and force them to pay high dividends. Easterbrook (1984) finds that a periodic payment of

2_{Jensen and Meckling (1976) also link the agency theory to debt holders and shareholders. This study}

(7)

-7- dividends causes companies to borrow more funds on the capital markets that are necessary for their investments, so borrowing is used to compensate for the dividends paid. Borrowing involves more information disclosure, thus acts as a restraint of opportunistic behavior of a manager. Jensen (1986) extended this topic by introducing the free cash flow hypothesis. He suggests that if a firm has an excessive amount of cash available, it might be better to distribute this to the stockholders in the form of dividends. Otherwise, managers might derive benefits for themselves, causing them to invest this money in possible negative net present value projects. All in all, this explanation of the agency problem is based on the vision that the interests of managers and shareholders conflict because managers act in their own best interest, instead of the best interests for the firm as a whole. Mancinelli and Ozkan (2006) note that this explanation of the agency problem most likely predominates when managers have considerable resources at their discretion. This could differ per country, in countries where concentrated ownership dominates; larger shareholders are better able to monitor and correct managers, which lowers this agency problem. It leads to conflict in firms with large controlling shareholders and minority shareholders. Hence, regarding the countries investigated in this study, monitoring by shareholders is more likely in France than in the UK due to the fact that blockholder ownership is more predominant in France.

2.1.1. Effects of ownership concentration

Past studies have identified two different effects that ownership concentration can have on dividends. Some authors argue that ownership concentration has a positive incentive effect on the dividend payout whereas others argue that it has an opposing effect.

(8)

-8- interest to pursue value maximizing activities, which is due to their high ownershi p percentage (Claessens et al., 2002).

The opposite effect, where ownership concentration has a negative effect was introduced by Rozeff (1982). He found a negative relationship between the dividend payout ratio and ownership concentration. It is argued that blockholders might even strengthen the agency problem because they prefer benefits that are described as private benefits of control. Morck et al. (1988) characterize these benefits as side payments, to increase their own wealth. In this way, these cash flows are not paid out as dividends, so minority shareholders cannot benefit from them. Morck et al. also argue that these private benefits of control are only paid out after a certain threshold of ownership has been met so that the large owners almost gain full control. This effect can be called the negative entrenchment effect and occurs when large shareholders are in full control and enrich themselves at the cost of the minority shareholders. Johnson et al. (2000) focused on the expropriation of minority shareholders and call it tunneling. The authors describe this phenomenon as the transfer of resources out of the firm to the controlling stockholder. Studies in the past have shown mixed results regarding the negative entrenchment effect. Faccio et al. (2001) and Claessens et al. (2002) conclude that this adverse effect is stronger when the control rights of the largest owner are strengthened by the use of pyramid structures, cross-shareholding or super voting rights. In these cases the majority owners have extra possibilities for control, apart from their cash flow rights. In turn, this gives the largest owners inducements to expropriate smaller shareholders.

(9)

-9- when ownership concentration increases. Lastly, Harada and Nguyen (2006) study the effects of ownership concentration for Japan. Measured as the total percentage of ownership held by the five largest shareholders, they also conclude that this ownership concentration has a negative effect on dividend payout.

2.2. Home bias

The presence of a home bias has gained widespread attention within the finance literature. The term ‘equity home bias puzzle’ has been given to the observation that both institutions and individuals hold a limited amount of foreign equity in their portfolios. This observation seems to contradict the benefits of internationally diversified portfolios. This puzzle has first been documented by French and Poterba (1991) and was later confirmed by Tesar and Werner (1998).

Grubel (1968) and Levy and Sarnat (1970) already found a low correlation of equity returns in industrial countries and reached the conclusion that firms can gain substantially from international diversification instead of holding just a domestic equity portfolio. It was until 1991 when French and Poterba demonstrated that investors do not put in practice what had been theorized about risk diversification. The authors showed that investors focus on putting their wealth in home assets, rather than also investing in international portfolio, and thus reducing their risk. French and Poterba mention for example that Japanese investors only had 1.9% of their equity in foreign stocks and U.S. investors owned about 6.2% of foreign equity. Tesar and Werner (1998) show that the home equity bias is less present in smaller countries. Moreover, they suggest that the bias declined in the 1990’s because in the mid-nineties, already 10 per cent of U.S. equity was invested overseas.

Multiple explanations have been suggested why the home bias is still present. One explanation involves institutional factors, such as high taxes on foreign investment, transaction costs and other limits on foreign investment (Bera and Kim, 2002). However, Bera and Kim argue that most other researchers reject this explanation, suggesting that their suboptimal diversification is the investor’s choice. Another explanation is given by Portes and Rey (2005); they suggest that the home bias could be explained by the lack of information, which increases the riskiness of international investments.

(10)

-10- investor is from the same country as the firm. The reasons why this phenomenon is expected are explained in the following section.

Previous research on the effects of the country of origin of the major shareholders has hardly been done. To my best knowledge, only Bena and Hanousek (2005) have taken into account the origin of major shareholders in trying to explain the level of dividends paid by Czech firms. They find that firms with foreign ownership have a higher dividend payout ratio. However, they argue that this is due to the weak corporate governance system in the Czech Republic. Moreover, they find a confirmation of the findings by Gugler and Yurtoglu (2003) and prove that the second largest shareholder has a positive effect on the level of dividends paid out, as well as on the frequency of paying out. Also, they argue that large majority shareholders often expropriate the Czech firms by extracting rents from the firm. They find that this effect is stronger for firms with a domestic owner and claim that this is due to the lack of regulation because of being an emerging economy.

2.3. Explanations for the link between home bias and dividends

This section explains why it is expected that a home bias also exists with the dividend policy of firms. One of the reasons why this relation is suspected was given in the introduction, namely the fact that a German/American firm closed down an entire factory in Belgium and laid off workers in Germany and elsewhere in Europe. One could conclude that the respective firm was more concerned about its own country and tried to avoid factory closings in Germany and the US. This example captures the general sense of concern that firms might have about their direct environment. Largest shareholders generally have a large influence on the management of a firm. Therefore, this phenomenon is more likely to occur when the shareholders are from the same country as the firm itself.

2.3.1. Corporate social responsibility

(11)

-11- voluntary and charitable activities that run within the local community. Second, and more important, it is argued that it is important for firms to assist in projects that improve a community’s quality of life. Later on, the author introduces three moral types of management and links them with the orientation towards the local community. First of all, immoral management is introduced, which presumes that in this case, the management exploits the community to its full extent and thereby disregards the needs of the community. The firm takes advantage of the community to the maximum without giving anything in return. Moreover, amoral management is introduced, meaning that the management does not take the community into account when making decisions. In other words, the community is just seen as a production factor. Lastly, moral management finds the community very important and tries to be a leading citizen. The management finds company and community goals interdependent. The first, and to some extent the second type, will most likely be linked to expropriation of the company. It is however likely that a local domestic owner induces the management to be more involved with the community, thus the country.

On the one hand, corporate social responsibility is directly linked to an ethical way of doing business and to act in the best interest to a society as a whole. However, one the other hand, corporate social responsibility is thus linked to a firm’s direct community. One might argue that the host country of the firm can be seen as the local community in which a firm operates. It is reasonable to assume that large shareholders that are from the same country as the firm itself are more involved with the local community than large shareholders from abroad.

2.3.2. Family firms

(12)

-12- Acting in the best interest of a community could be in the form of charitable foundations but also involves in sustainability of the respective firm. This could have two implications when this topic is linked to dividends. First of all, when trying to behave in a responsible way, companies pay dividends to their investors in order to reward them. More importantly, it can be seen as a reward to the local shareholders, who are in general more likely to hold local shares. In this case, it is expected that firms with a domestic largest shareholder are more probable to pay out dividend.

Additionally, the effects of the origin of the largest shareholder on the level of dividend paid out will be investigated. Based on the same ideas about corporate social responsibility that have been introduced before, one could argue that the level of dividends that will be paid out will be a fair amount, being not too high. Paying out excessive dividends could harm the company in the long run. Based on the assumption that a largest shareholder that is from the same country as the respective firm is more involved with the firm and its community, or in short, the social sustainability of the firm, it is less likely that these firms will pay out excessive amount of dividends. Consequently, the amount of cash dividend paid is thought to be lower when the firm and largest shareholder are from the same country.

Moreover, foreign firms face certain risks when investing abroad. This could range from exchange rate fluctuations to differences in culture, regulations and taxation (Coval and Moskowitz, 1999). Therefore, the home bias within the equity markets is still present, even though diversification of equity portfolios could provide substantial benefits to investors. Because of the extra risk they bear, it is expected they expect compensation in the form of dividends in exchange. Therefore, I expect that companies where the largest shareholder is foreign pay out more dividends than those with a domestic largest shareholder.

(13)

-13-

2.4. Differences between France and the United Kingdom

This section focuses more on the differences between the two countries that are investigated in this study, France and the United Kingdom. Two effects seem to have an influence on the dividend payout and the possible home bias: 1) Corporate Governance systems and 2) The number of family firms.

Corporate governance is thought to influence the dividend payout of firms. According to Shleifer & Vishny (1997), two major systems of corporate governance exist: the Continental European system and the Anglo-Saxon system. Among the various differences, the differences regarding ownership are most important for this study. The origin of these differences can be found in the legal systems in these regions. La Porta et al. (1999) show that countries with a civil law system (such as France and Germany) have weak shareholder protection. Countries with a common law system (United Kingdom and United States) often provide better protection of minority shareholders. La Porta et al. (2000) find that dividend policies differ in countries with dissimilar legal systems and are greatly affected by the shareholder protection within these countries. They conclude that firms from civil law countries generally pay out lower dividends compared to firms from common law countries. Legal protection is generally higher in the latter. Cuervo (2002) has compared the two corporate governance systems. He finds that control in Continental-European firms is mostly characterized by large-shareholder control; the Anglo-Saxon mechanism is largely determined by market-control. To be more precise, the Continental-European system is mainly characterized by the following: 1) Concentrated ownership; 2) Control exercised by large shareholders; 3) The board of director is often controlled by internal directors or external directors, linked with the large shareholders; 4) Capital markets are relatively illiquid. The Anglo-Saxon mechanism is distinguished by: 1) Diffuse ownership; 2) Control is in the hands of the board of directors; 3) External directors play an important role; 4) Capital markets are liquid (Cuervo, 2002).

(14)

-14- minority protection is weak. This causes blockholders to increase their private benefits, at the stake of the minority shareholders.

La Porta et al. (2002) provide empirical support for the above-mentioned differences regarding shareholder protection and ownership dispersion. It was found that the firm value (Tobin’s Q) for firms in civil law countries was often lower, most likely due to the weak shareholder protection in combination with blockholder ownership.

Taking into account these differences, shareholders in the UK have fewer ways to influence the management of a firm compared to their French counterparts. Moreover, in combination with the facts that France has a higher percentage of listed family firms (65 per cent) compared to the United Kingdom (24 per cent) and that corporate social responsibility might be higher for family firms could lead to stronger results for France.

2.5. Hypotheses

In this section, I will summarize the different theories that have been introduced before and develop the hypotheses regarding the effects of the origin of a firm’s largest shareholder on the respective firm’s dividend policy.

First of all, it has been argued that agency problems within firms can be linked to dividends. This link has been widely used in trying to explain dividend policies of firms.

(15)

-15- domestic shareholder has on the decision whether to pay out dividends or not. With higher corporate social responsibility, the large local owner assumes that there is also a home bias in holding the shares. Dividends (to local shareholders) will then satisfy local owners and thereby increase the local reputation of the large local owner.

H1: A domestic largest shareholder has a positive impact on the decision to pay

dividends.

Moreover, the effects of the home or foreign ownership on the likelihood to pay dividends will be tested in combination with the ownership percentage. This will thus test the power of the largest shareholder. I expect that firms with more powerful largest shareholders (measured by their percentages of shares held) are more likely to show strong results. Therefore, with regards to the case when the largest shareholder is domestic, I expect that shareholders with a higher stake in a firm result in stronger positive outcomes. Therefore, I expect again a positive impact on the likelihood to pay dividends for domestic largest shareholders. Moreover, this test can be seen as a robustness check on the first hypothesis. This will be explained in the methodology section.

H2: The percentage of shares held by the largest domestic shareholder is positively

related to the decision to pay dividends.

Additionally, the effect of the presence of a largest domestic shareholder on the total amount of cash dividend paid is tested. As is argued in the sections above, it is more likely that foreign shareholders take the risk to expropriate a firm because they are less involved with the community of the firm. Domestic shareholders would care more about possible job losses and the chances of bankruptcy. Therefore, the dividend pay outs of firms with domestic owners will be moderate compared to those with a foreign largest shareholder.

H3: A domestic largest shareholder has a negative impact on the dividend amount paid

by a paying firm.

(16)

-16- more power. Therefore, I hypothesize a negative relationship between larger domestic shareholders and cash dividend paid by paying firms.

H4: The percentage of shares held by the largest domestic shareholder is negatively

related to the dividend amount paid by a paying firm.

For each of the above-mentioned hypotheses, I develop a null hypothesis, stating that no relationship can be observed.

(17)

-17-

3. Data

This section describes the data that are used in this study. First, the sample will be explained, including how they were derived and the sources wherefrom. Section 3.2 provides information on the variables that are being used and the final section explains the descriptive statistics.

3.1. Sample

This research uses data from 2002 to 2009 for the dependent variables and ownership data. The data for most of the remaining independent and control variables is lagged by one year, except the data used for risk, which is calculated by taking data for the five years before the respective year, thus starting in 1997. The time-period starts in 2002 because data on firm ownership were not available in the years before 2002. Moreover, no dividend data were available after 2009. The data were gathered by using two databases, Amadeus was used for the ownership data and Thompson Datastream Worldscope was used for dividend data and the other financial variables.

The sample of this study comprises all firms that are listed on the Euronext Paris and London Stock Exchange, originate from France or the UK respectively and had data available on Datastream. Next to that, the ISIN codes of these firms were matched with Amadeus and only the firms that had ownership data available for all years were used in the final sample. Unlike Gugler and Yurtoglu (2003), this study does not assume that ownership concentration is stable. Therefore, information on the shareholders of a firm is collected for all years in the sample period. This includes both the ownership percentages, as well as the country of origin of the shareholders. Next, all financial institutions were removed from the sample to avoid confounding effects from financial regulations. Moreover, different variables are used to control for other effects that may have an influence on dividend policy. If the data for one of these values was missing for one year, I removed the firm year from the final sample. The final sample consists of 3805 unbalanced observations, 1272 of which are from France and 2533 from the UK.

3.2. Variables

3.2.1. Dependent variables

(18)

-18- Moreover, the second dependent variable for the second hypothesis is the log of the total cash dividend (LCDIV) that the firms paid out. The cash dividend that Datastream reports for for example 2009 is the total amount that a company paid out in 2009, which is based on the company profit of 2008. Therefore, all other variables, except the ownership data and risk are lagged by one year. Because the dividends are based on the profits of the previous year and declared by the shareholders in the same year as they are paid out, the ownership percentages of the major stockholders are not lagged.

3.2.2. Independent variables

I gathered the country of the shareholders. Together with the country of the firms from my sample, I created the dummy home, taking the value 1 if the firm and its largest shareholder are from the same country and 0 otherwise. The independent variable homeper is created by multiplying the home dummy by the ownership percentage of the largest shareholder. In this way, the effects of the ownership percentage in combination with the origin of the largest shareholder can be studied.

3.2.1. Control variables

For the control variables, I firstly gathered data on ownership. I take the percentages of the two major owners of a firm, the ownership concentration, separated, called owner1 and owner2. Amadeus reports two percentages, ‘direct’ and ‘total’ ownership. Total ownership describes the fact that firm A owns a certain amount of firm B. Direct ownership means that firm B is wholly owned by A. If firm A thus owns firm B in full and if firm B owns 80 per cent of firm C, then, firm A thus also owns 80 per cent of firm C indirectly. In order to reach to the ownership percentages, both the direct and total percentages are used. In Amadeus, in most cases, direct ownership is 0 when the firm has a listed shareholder in the total percentage and vice versa. Moreover, Amadeus sometimes reports a term instead of a percentage. I replaced these terms with the corresponding minimal value. ‘Wholly owned’ (WO) corresponds with a minimum ownership percentage of 98 per cent, ‘Majority Owned’ (MO) indicates that 50.01 per cent belongs to the shareholder. ‘Jointly Owned’ (JO) means that ownership is split between two shareholders, thus 50 per cent. Lastly, ‘Negligible’ (NG) indicates that less than 0.01 per cent is owned by the shareholder. Thus, these terms are replaced with the minimum percentage belonging to the category.

(19)

-19- significant impact on dividends3. The control variables that are used are (i) Profitability

(profit), (ii) Size (size), (iii) Leverage (lev), (iv) Risk (risk), (v) Current growth (cg) and (vi) growth opportunities (go).

The first control variable that is included is the profitability (profit) of the company. This study measures profitability in the same way as Gugler and Yurtoglu (2003) by the ratio of net income (Worldscope item WC01250) to total assets (Worldscope item WC02999), thus the return on total assets. Profitable firms generally have more funds available to distribute to their shareholders; therefore, a positive relation regarding the dependent variable is expected. Second, Fama and French (2001) show, among other authors, that size (size) does play a role when determining dividend payouts. Some well-known proxies for size are (i) the total revenues of the firm, (ii) the book value of assets, (iii) the market value of equity and (iv) the number of employees. In line with Chae et al. (2009), this study proxies size by taking the natural logarithm of sales (Worldscope item WC01001). A positive effect on the dependent variable is expected, given the analysis of Fama and French (2001) who found that dividend payers are generally larger than non-dividend payers. Thirdly, leverage (lev) is included as a control variable. Following Jensen’s (1986) free cash flow hypothesis, debt can be seen as a substitute of dividends when controlling agency problems of free cash flow. This study uses the same proxy as Mancinelli and Ozkan (2006), who proxy leverage as the ratio of the book value of debt (Worldscope item WC03255) to the book value of assets (Worldscope item WC02999). Leverage is expected to be negatively associated with the dependent variable. Fourthly, Risk (risk) is proven to have a negative influence on dividend since firms experiencing high risk tend to hold more cash reserves. Chay and Suh (2009) and Dekker (2010) measure risk by the volatility of the stock returns. They claim that the volatility of stocks reflects cash flow uncertainty because stock prices fluctuate more in case of unpredictable cash flows. This study uses a proxy for risk as was used by Von Eije and Megginson (2009). These authors the standard deviation of the revenues (Worldscope item WC01001) of the past five years, divided by total assets (Worldscope item WC02999) of the same year. Because risk takes into account the revenues of the five previous years, it is not needed to lag this variable an extra year. Moreover, two variables regarding the firms’ growth are included as control variables. Mitton (2005) and Dekker (2010) included a proxy for current growth (cg), which is measured by the relative change in total assets (Worldscope item WC02999), compared to the previous year. The relation with dividends is expected to be negative, since growth might require investments at the expense of dividend payout. Dekker (2010) argues that the growth in

(20)

-20- assets is a way to measure current investment and expects it to hold for future investment as well. Finally, I include growth opportunities (go). Fama and French (2001) found that growth opportunities have an impact on dividend payouts. They argue that firms that have never paid out dividends show the best growth opportunities. Hence, a negative relationship with the dependent variable is expected. Mancinelli and Ozkan (2006) proxy the growth opportunities by the market-to-book ratio since it could give an indication of the under- or overvaluation of a firm. I will use this approximation as well. The market-to-book ratio is calculated by the sum of total debt (Worldscope item WC03255) and market capitalization (Worldscope item WC08001) divided by total assets (Worldscope item WC02999). Appendix A provides a detailed summary of the dependent and independent variables.

3.2.2. Descriptive statistics

[INSERT TABLE 1 ABOUT HERE]

Table 1 reports the descriptive statistics for the total sample, as well as for the subsample firms in France (n=1272) and firms in the United Kingdom (n=2533). About 78 per cent of the firms in both countries pay dividends. The mean of the natural log of the cash dividends is slightly different and higher in France, 9.43 versus 9.03 in the UK. This conflicts with La Porta et al. (2000) who predicted that dividends would be higher in common law countries due to better investor protection. Moreover, the number of observations where the largest shareholder is domestic does not differ much; about 54 per cent of the French firms have a domestic largest shareholder whereas this percentage is 59 for the UK. With regards to the variable that measures the power of the largest domestic owners, homeper, it can be seen that the largest shareholder, when domestic, owns about 28 per cent of the total stock in France, whereas this is only 10 per cent in the UK. Civil law countries are known for blockholder ownership. This can also be seen from the table since the average ownership of the largest owner is about 51 per cent in France and 18 per cent in the UK. The same observation can be seen for the second largest owner. This ownership percentage is about 20 per cent in France, whereas an average second owner owns about 10 per cent of the stock in a British firm.

(21)

-21- are used in the same regressions, multicollinearity between the independent variables is not likely.

Table 2 provides the descriptive statistics for all variables again, and more importantly, also for the cases when the largest shareholder is domestic (home=1) or foreign (home=0). Table 2 shows that it is more likely that if the firm and shareholder are from the same country, dividend will be paid. However, it is reasonable to assume that the amount of dividends paid will be lower, given a lower LCDIV for home=1. With regards to the ownership percentages, the means are slightly different for the largest home and foreign owner. A largest foreign owner (home=0) owns about 30 per cent of the shares whereas a largest domestic owner owns about 28 per cent of the shares. The average percentage of shares held for the second owner is the same, namely 14 per cent. The profit measure has exactly the same mean. It seems that firms with investors from abroad are slightly larger in size and also somewhat more leveraged. Moreover, firms with a largest shareholder from abroad seem to encounter more risk and to have higher growth parameters.

(22)

-22-

4. Methodology

Because of conflicting characteristics of ownership (relatively fast changing) and cash dividends (relatively stable) I do not use fixed effects panel data to estimate the effect of ownership. Instead I take a structured approach by estimating cross-section effects in ownership on dividends. Because I have observations for a relatively long period of time (2002-2009) I use the approach of Fama and MacBeth (1973) here. When a dated panel structure would be used in order to test the hypotheses, it will not provide reliable results. The dividend changes in a dated panel would be too slow to take into account the changes in ownership. Therefore, a cross-section approach will be used. Fama and MacBeth (1973) developed a two-step model to estimate parameters with multiple subsections across time. The steps taken are as follows: First, regressions were run, one for each year within the sample. Each regression provides an estimation for the coefficients for every variable. Next, this resulted in a mean for the time-series of the coefficients. These means were tested for a significant difference from zero using a t-test.

The above mentioned approach is used for both the logit regressions, in order to test the effect that domestic shareholders have on the likelihood to pay dividends, and the ordinary least squares regressions, to study the effects of the largest shareholder on the amount of dividend paid by paying firms. For both cases, the steps are done for the total sample, as well as the French and UK subsamples.

The first hypothesis focuses on the effects of the country of origin of the first shareholder on the question whether to pay dividends or not. This question will be answered using a cross-sectional Fama MacBeth logit regression. The dependent variable in this regression will be a dummy (pay) that takes the value 1 if the firm is a dividend payer and the value of 0 otherwise. A dummy (home) that takes the value 1 if the firm and the largest shareholder are from the same country, and 0 if this is not the case, will be the primary independent variable. The regression equation is as follows:

(1)

(23)

-23- shareholder can exercise. The effect that owner1 has on the dependent variable, pay, is already captured by the homeper variable, therefore owner1 mostly measures the effect for the observations that are not included in homeper, thus when the dummy home=0. Therefore, by testing the power of the largest shareholder in combination with its origin, this regression can be seen as a robustness check on the previous regression to check whether the outcome of the first regression still holds.

Since the dependent variable is still a dummy variable, I will use the Fama MacBeth approach for a logit regression.

(2)

The third question studies the effect of the level of dividends paid out by paying firms, using the natural logarithm of the total cash dividends (LCDIV) as dependent variable, in relation to the origin of the largest shareholder (home). This will be tested by ordinary least squares analysis using the Fama MacBeth cross-sectional approach on the dividend payers.

(3)

Finally, the last regression will test the effect of the origin of the largest shareholder multiplied with the respective ownership percentages on the dividends amount paid by a paying firm, thus adding homeper to the equation. Like the second regression, this will test for the effects that largest shareholders can exercise, by measuring their power by their ownership percentage. Also similar to the second regression, owner1 measures the impact that foreign largest shareholders have on the dividends paid by a dividend paying firm, because the effects when a largest domestic owner is present, are captured by the homeper variable. Moreover, this regression will be used as a robustness check for the results of the previous regression.

It will be tested using the cross-sectional Fama MacBeth method, using an ordinary least squares regression as follows:

(4)

(24)

-24-

5. Empirical results

This section shows the results for the four regressions. Tables 3 and 4 display the logit regressions on the effects on the question whether to pay dividends or not. Tables 5 and 6 display the results of the ordinary least squares regressions on the dividend amounts.

5.1.1. Logit regressions

The first regression deals with the effects of the origin of the largest shareholder and the question whether to pay dividend or not. The independent variable home is positively significant for both the total sample and France at a 1 per cent significance level and for the UK at a 5 per cent significance level, meaning that firms where the largest shareholder is domestic are more likely to pay out dividend. Taking into account the coefficients for home, it can be seen that the effect on the likelihood to pay is stronger in France than in the UK.

(25)

-25- Hence, regarding the first research question, one could conclude that the presence of a largest shareholder from the same country as the firm itself leads to a higher likelihood that dividend will be paid out. This is the case in both France and the UK, though the effect is stronger in France, given the higher coefficient. I therefore accept the hypothesis 1, meaning that there exists a positive relation between the domestic origin of the largest shareholder and the likelihood to pay out dividends.

The second logit regression takes the ownership percentage in combination with the home dummy, creating a variable called homeper which is used as the independent variable. In this way, it can be tested whether larger local shareholders have a stronger effect on the likelihood to pay out. First of all, the homeper variable is thought to have a positive effect on the dependent variable. The variable is significant and positive in all cases, indicating that firms with a more powerful largest shareholder that is domestic are more likely to pay out dividend. Like in the previous regression, the effect is stronger in France than in the UK. It can thus be concluded that the percentage of ownership has a positive effect on the likelihood to pay out. As was mentioned in the Methodology section, owner1 can be seen as a substitute for measuring the power of foreign largest shareholders. The coefficients of this variable are negative for all subsamples, indicating that firms where the largest shareholder is foreign, are less likely to pay dividend. The coefficient of the homeper variable is positively significant, confirming the result from the previous regression. It can therefore be seen as robust.

The second owner is, like in the first regression, negatively related to the dependent variable for the total sample and the UK and is non-significantly positively related in France. Moreover, the control variables relating to profitability, size and leverage are all significant with the predicted signs. Risk has the expected negative coefficient regarding the dividend payment as well and is significant in all cases. The measures for growth seem to be contradicting to the predicted signs in some cases; however, they are not significant. They were all thought to have a negative influence on the likelihood to pay out.

The explanatory power of these regressions is about the same as the first regression; about 39 per cent of the observations are explained by the model.

(26)

-26- France and the UK. Therefore, I accept the second hypothesis and conclude that the power of the largest shareholder in combination with its origin also influences the likelihood to pay out. Larger domestic shareholders positively influence the likelihood to pay out, whereas largest foreign shareholders have a negative influence.

5.1.2. Ordinary least squares regressions

Table 5 shows the outcomes for the ordinary least squares regressions on the amount of dividend paid out. All other control variables are similar to the first regression in this study. The home coefficient is negative in all cases. This implies that for firms where the largest shareholder is from the same country as the respective firm, the dividend payout is generally lower. This only holds for France at a 5 per cent significance level. Home is not significant for the total sample and the UK, indicating that the presence of a largest domestic shareholder has no effect on the dividend amount paid in the UK.

The control variable for the ownership percentage of the largest holder behaves as expected, as it has a negative influence on the overall payout. However, it is only significant for the total sample and France, not for the UK. The influence of the second owner is not clear. The variable is positive for both France and the UK, but not significant. However, on the total sample, the variable is negative at a 10 per cent level. Again, the variables for profitability, size and leverage are all significant and behave as expected. Risk is also negatively related to the dividend payout, but never significant. Current growth also has a negative influence for all samples and behaves as predicted. Growth opportunities, on the other hand, has an unexplained positive coefficient and is significant at a 5 per cent level in all cases.

The explanatory power of this model is much higher compared to the previous logit regression, about 80 per cent.

(27)

-27- [INSERT TABLE 6 ABOUT HERE]

Finally, the fourth regression measures the influence of the largest domestic owner (measured by the ownership percentage) on the level of cash dividend paid out and acts as a robustness check for the third regression. The homeper variable is negative for the total sample, as well as for France and the UK. However, the coefficient is only significant for the total sample and France, not for the UK. Firms with a larger first shareholder that is from the same country as the firm seem to pay out lower dividends compared to smaller first shareholders that are from the same country as the firm. This is in line with the prediction that larger shareholders generally pay out lower cash dividends. This effect is negative for the UK as well; however, it is not significant. The largest owner can in this regression again be seen as a substitute for the effect that foreign largest shareholders have on the dependent variable. The owner1 coefficient is less than the homeper coefficient for the total sample, this indicates that larger foreign shareholders (measured by the substitute owner1) have a stronger negative effect than larger domestic firms. The effects of foreign largest shareholders for France and the UK are negative, but never significant so no conclusion can be drawn from this. Because the coefficient for owner1 is less than homeper, this robustness check fails to confirm the (insignificant) finding of the third regression.

The effect of the second largest owner is again not clear. The proxy is positive but insignificant for both France and the UK, but negative and significant at a 10 per cent level for the total sample. The control variables for profit, size and leverage are all significant and move along with the model as predicted. Risk is negative in all cases; however, it is never significant. Like the previous regressions, the measures for growth are less consistent.

The explanatory power of the models is about 80 per cent.

(28)

(29)

-29-

6. Conclusion

This study examines the relations between the origin of the largest shareholder of French and UK firms and a firm’s dividend policy. Whereas many authors have successfully tried to explain a relation between ownership concentration and dividend policy, this study aims to investigate whether a home bias exists in the dividend policy of firms. It was hypothesized that firms with a domestic largest shareholders are more likely to pay out dividends, but would most likely pay out lower amounts. This is tested using cross-section analysis for a sample of French and UK firms, over the period 2002 to 2009.

First of all, I test the effect of the influence of the origin of a firm’s largest shareholder, on the likelihood to pay dividends. Based on the results, one can conclude that for both France as well as the UK, a largest shareholder from the same country as a firm, generally is more likely to force firms to pay out dividends. Thus, this leads to a conclusion that a home bias that is not only present in the equity market, but is probably also present within the dividend policy of firms. This conclusion can be drawn for both France as well is the UK. Moreover, the effects of the presence of a largest domestic shareholder are tested along with the influence that the shareholders have within a firm, measured by their ownership percentage. This leads to positive significant coefficients for both France and the UK, indicating that the larger the stake that the largest domestic shareholder has in a firm, the more likely it is that firm will pay out dividend. This is in stark contrast to many previous studies, e.g. Gugler and Yurtoglu (2003) and Mancinelli and Ozkan (2006), who found that payouts decrease when ownership concentration increases. Moreover, the second regression made clear that the opposite is the case for firms where the largest shareholder is foreign; they are less likely to pay out dividend. The effects are stronger for the French subsample, as was suspected because of the blockholder ownership that is often present in France.

(30)

-30- percentage of the shareholders. Here I also find a negative relationship for the total sample as well as for France, meaning that the larger the stake of the domestic shareholder in the firm, the lower the total cash dividend paid. This is consistent with earlier studies (e.g. Gugler and Yurtoglu, 2003 and Mancinelli and Ozkan, 2006) that found a negative relation between the ownership percentage and the dividends paid out. However, from the results it can be interpreted that the effect of the percentage of shares held by the largest investor, is stronger negative than when this shareholder would be domestic.

As has been argued above, the observation that it is more likely for firms where the largest shareholder is domestic to pay dividends, but to pay out less (in the case of France) could very well be present due to a combination of two facts. 1) Firms with domestic shareholders are probably more concerned about the direct consequences that their actions might have on the local and countrywide scale. Expropriation of the firms by paying out high dividends could lead to the risk of bankruptcy. In the end, this could lead to job losses and lower revenues, causing harm to the direct community of the firm, the country. 2) A higher percentage of French firms is family owned compared to UK firms. It was argued by Dyer and Whetten (2006) that family firms are more likely to engage in a corporate social responsible way of doing business. It is remarkable that largest domestic shareholders don’t have a significant influence for firms in the UK. One potential reason might be the fact that only one third of the listed firms in the UK is family owned, whereas this number is about two third for French listed firms. Moreover, the difference in outcome between France and the UK might also exist due to the different corporate governance systems that the countries have. In France, blockholder ownership is relatively common.

6.1. Suggestions for future research

(31)

(32)

-32-

7. References

Allen, F., & Michaely, R. 2003. Payout policy. Handbook of the Economics of Finance, 1: 337-429.

Amadi, A. 2004. Does familiarity breed investment? An empirical analysis of foreign equity holdings. papers.ssrn.com.

Bena, J., & Hanousek, J. 2006. Rent extraction by large shareholders: Evidence using dividend policy in the Czech Republic. papers.ssrn.com.

Bera, A., & Kim, S. 2002. Testing constancy of correlation and other specifications of the BGARCH model with an application to international equity returns. Journal of

Empirical Finance, 9(2): 171-195.

Black, F. 1998. The dividend puzzle. Streetwise: The Best of the Journal of Portfolio.

Carroll, A. B. 1991. The pyramid of corporate social responsibility: Toward the moral management of organizational stakeholders. Business horizons.

Chae, J., Kim, S., & Lee, E. 2009. How corporate governance affects payout policy under agency problems and external financing constraints. Journal of Banking & Finance, 33(11): 2093-2101.

Claessens, S., Djankov, S., Fan, J., & Lang, L. 2002. Disentangling the incentive and

entrenchment effects of large shareholdings. The Journal of Finance, 57(6): 2741-2771.

Coval, J., & Moskowitz, T. 1999. Home bias at home: Local equity preference in domestic portfolios. The Journal of Finance, 54(6): 2045-2073.

Cuervo, A. 2002. Corporate governance mechanisms: A plea for less code of good governance and more market control. Corporate Governance: An International

Review, 10(2): 84-93.

Dahlquist, M., Pinkowitz, L., & Stulz, R. 2003. Corporate governance and the home bias.

(33)

-33- Dekker, J. 2010. The effect of ownership concentration on payout decisions. Evidence from the UK and Germany. Master thesis, University of Groningen.

Dyer Jr, W., & Whetten, D. 2006. Family firms and social responsibility: Preliminary evidence from the S&P 500. Entrepreneurship Theory and Practice, 30(6): 785-802.

Easterbrook, F. 1984. Two agency-cost explanations of dividends. The American

Economic Review, 74(4): 650-659.

Faccio, M., Lang, L., & Young, L. 2001. Dividends and expropriation. American Economic

Review, 91(1): 54-78.

Fama, E., & MacBeth, J. 1973. Risk, return, and equilibrium: Empirical tests. The Journal

of Political Economy, 81(3): 607-636.

Fluck, Z., 1999. The Dynamics of the Manager-Shareholder Conflict. Review of Financial

Studies, 12(2): 379-404

.

Grubel, H. G. 1968. Internationally diversified portfolios: Welfare gains and capital flows.

The American Economic Review.

Gugler, K., & Yurtoglu, B. 2003. Corporate governance and dividend pay-out policy in Germany. European Economic Review, 47(4): 731-758.

Harada, K., & Nguyen, P. 2006. Ownership concentration, agency conflicts, and dividend policy in Japan. papers.ssrn.com.

Holder, M., Langrehr, F., & Hexter, J. 1998. Dividend policy determinants: An

investigation of the influences of stakeholder theory. Financial Management, 27(3): 73-82.

Jensen, M. 1986. Agency costs of free cash flow, corporate finance, and takeovers. The

American Economic Review, 76(2): 323-329.

Jensen, M., & Meckling, W. 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4): 305-360.

Johnson, S., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. 2000. Tunneling. American

(34)

-34- La Porta, R., Lopez-de- Silanes, F., & Shleifer, A. 1999. Corporate ownership around the world. The Journal of Finance, 54(2): 471-517.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. 2000. Investor protection and corporate governance. Journal of Financial Economics, 58(1-2): 3-27.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. 2002. Investor protection and corporate valuation. Journal of Finance, 57(3): 1147-1170.

Levy, H., & Sarnat, M. 1970. International diversification of investment portfolios. The

American Economic Review, 60(4): 668-675.

Lintner, J. 1956. Distribution of incomes of corporations among dividends, retained earnings, and taxes. The American Economic Review, 46(2): 97-113.

Mancinelli, L., & Ozkan, A. 2006. Ownership structure and dividend policy: Evidence from Italian firms. The European Journal of Finance, 12(3): 265-282.

Modigliani, F., & Miller, M. 1958. The cost of capital, corporation finance and the theory of investment. The American Economic Review, 48(3): 261-297.

Morck, R., Shleifer, A., & Vishny, R. 1988. Management ownership and market valuation:: An empirical analysis. Journal of Financial Economics, 20: 293-315.

Mueller, H., & Philippon, T. 2006. Family firms, paternalism, and labor relations.

papers.ssrn.com.

Portes, R., & Rey, H. 1999. The determinants of cross-border equity flows. cepr.org.

Poterba, J. M. 1991. Investor diversification and international equity markets. The

American Economic Review.

Rozeff, M. 1982. Growth, beta and agency costs as determinants of dividend payout ratios. Journal of Financial Research, 5(3): 249-259.

Schooley, D., & Barney Jr, L. 1994. Using dividend policy and managerial ownership to reduce agency costs. Journal of Financial Research, 17(3): 363.

Shleifer, A., & Vishny, R. 1986. Large shareholders and corporate control. The Journal of

(35)

-35- Shleifer, A., & Vishny, R. 1997. A survey of corporate governance. Journal of Finance, 52(2): 737-783.

Sraer, D., & Thesmar, D. 2007. Performance and behavior of family firms: Evidence from the French stock market. Journal of the European Economic Association, 5(4): 709-751.

Tesar, L. & Werner, I. 1998. The internationalization of securities markets since the

1987 Crash. Brookings-Wharton Papers on Financial Services, The Brookings

(36)

-36-

Tables

Table 1: Descriptive statistics for Total sample, France and UK

Total sample Panel A: France Panel B: UK

Mean Median Max Min Std. Dev. Mean Median Max Min Std. Dev. Mean Median Max Min Std. Dev.

Pay 0.78 1.00 1.00 0.00 0.42 0.78 1.00 1.00 0.00 0.41 0.78 1.00 1.00 0.00 0.42 LCDIV 9.18 8.96 15.84 0.37 2.24 9.43 9.18 15.44 0.69 2.35 9.03 8.85 15.84 0.37 2.15 Home 0.57 1.00 1.00 0.00 0.49 0.54 1.00 1.00 0.00 0.50 0.59 1.00 1.00 0.00 0.49 Homeper 0.16 0.08 0.99 0.00 0.21 0.28 0.25 0.99 0.00 0.29 0.10 0.08 0.81 0.00 0.12 Owner1 0.29 0.20 0.99 0.00 0.21 0.51 0.50 0.99 0.03 0.18 0.18 0.14 0.90 0.00 0.12 Owner2 0.14 0.11 0.50 0.00 0.09 0.20 0.19 0.50 0.00 0.11 0.10 0.10 0.43 0.00 0.06 Profit 0.06 0.07 0.78 -1.77 0.12 0.06 0.05 0.78 -0.47 0.07 0.06 0.08 0.57 -1.77 0.14 Size 5.50 5.38 8.28 2.38 0.96 5.79 5.66 8.14 3.28 0.96 5.36 5.28 8.28 2.38 0.92 Lev 0.20 0.18 1.99 0.00 0.17 0.23 0.22 1.55 0.00 0.16 0.19 0.16 1.99 0.00 0.18 Risk 0.13 0.03 33.02 0.00 0.83 0.06 0.03 6.61 0.00 0.28 0.16 0.04 33.02 0.00 0.99 CG 1.16 1.05 70.13 0.07 1.65 1.16 1.04 70.13 0.19 1.99 1.16 1.05 65.49 0.07 1.45 GO 1.16 0.90 13.89 0.06 0.98 0.97 0.79 0.76 0.09 0.70 1.26 0.97 13.89 0.06 1.08

(37)

-37-

Table 2: Descriptive statistics for Total sample, Home=0 and Home=1

Total sample Home=0 Home=1

Mean Median Max Min Std. Dev. Mean Median Max Min Std. Dev. Mean Median Max Min Std. Dev. Pay 0.78 1.00 1.00 0.00 0.42 0.74 1.00 1.00 0.00 0.44 0.80 1.00 1.00 0.00 0.40 LCDIV 9.18 8.96 15.84 0.37 2.24 9.66 9.42 15.84 1.10 2.43 8.84 8.71 15.63 0.37 2.03 Home 0.57 1.00 1.00 0.00 0.49 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00 0.00 Homeper 0.16 0.08 0.99 0.00 0.21 0.00 0.00 0.00 0.00 0.00 0.28 0.19 0.99 0.00 0.22 Owner1 0.29 0.20 0.99 0.00 0.21 0.30 0.25 0.98 0.00 0.21 0.28 0.19 0.99 0.00 0.22 Owner2 0.14 0.11 0.50 0.00 0.09 0.14 0.11 0.50 0.00 0.09 0.14 0.11 0.48 0.00 0.09 Profit 0.06 0.07 0.78 -1.77 0.12 0.06 0.07 0.78 -1.69 0.12 0.06 0.07 0.38 -1.77 0.12 Size 5.50 5.38 8.28 2.38 0.96 5.62 5.47 8.28 2.45 1.08 5.42 5.33 8.24 2.38 0.84 Lev 0.20 0.18 1.99 0.00 0.17 0.21 0.19 1.99 0.00 0.18 0.19 0.17 1.61 0.00 0.16 Risk 0.13 0.03 33.02 0.00 0.83 0.14 0.04 19.69 0.00 0.78 0.12 0.03 33.02 0.00 0.85 CG 1.16 1.05 70.13 0.07 1.65 1.19 1.05 70.13 0.07 1.93 1.14 1.05 65.49 0.14 1.42 GO 1.16 0.90 13.89 0.06 0.98 1.20 0.93 13.89 0.13 0.99 1.14 0.88 10.80 0.06 0.97

(38)

-38-

Table 3: Fama MacBeth logit regressions testing the home bias related to the likelihood to pay out dividends, tested for the total sample, France and UK over the period 2002-2009.

_{Dependent variable: Pay}

Total France UK Independent variable Home 0.458*** 1.044*** 0.278** (5.236) (5.725) (2.682) Control variables Owner1 -1.199*** -1.690** -0.928 (-6.702) (-2.558) (-1.794) Owner2 -0.978 0.676 -3.613* (-1.162) (0.419) (-2.006) Profit 15.179*** 16.589*** 17.040*** (7.419) (4.992) (5.834) Size 1.004*** 1.326*** 0.805*** (15.643) (12.413) (7.669) Lev -2.905*** -3.826*** -3.047** (-5.112) (-4.229) (-2.492) Risk -4.927*** -11.464*** -4.845** (-5.352) (-4.351) (-3.224) CG -0.104 0.391 -0.191 (-0.308) (0.395) (-0.786) GO 0.129 -0.209 0.350 (0.702) (-1.137) (1.205) Constant -3.373*** -5.224*** -2.068** (-6.598) (-4.249) (-2.847) Average observations per year with Pay=0 98 35 69 Average observations per year with Pay=1 376 125 247

McFadden R-squared 0.38 0.40 0.41

(39)

-39-

Table 4: Fama MacBeth logit regressions with the home bias and effect of the ownership percentage related to the likelihood to pay out dividends, tested for the total sample, France and UK over the period 2002-2009.

_{Dependent variable: Pay}

_Total _France _UK Independent variable Homeper 1.373*** 1.798*** 1.154** (5.303) (6.678) (2.716) Control variables Owner1 -2.016*** -2.559*** -1.476** (-7.697) (-3.550) (-2.699) Owner2 -0.874 0.817 -3.700* (-1.002) (0.489) (-2.042) Profit 15.185*** 16.681*** 17.037*** (7.539) (5.092) (5.802) Size 1.013*** 1.309*** 0.808*** (15.074) (12.198) (7.672) Lev -2.946*** -3.725*** -3.046** (-5.201) (-3.991) (-2.520) Risk -4.868*** -11.260*** -4.856** (-5.409) (-4.440) (-3.248) CG -0.100 0.440 -0.210 (-0.296) (0.432) (-0.833) GO 0.137 -0.233 0.360 (0.754) (-1.249) (1.201) Constant -3.175*** -4.742*** -1.918** (-6.471) (-3.673) (-2.629) Average observations per year with Pay=0 103 35 69 Average observations per year with Pay=1 371 125 247

McFadden R-squared 0.37 0.40 0.41

(40)

-40-

Table 5: Fama MacBeth ordinary least squares regressions testting the home bias related to the cash dividends by dividend payers, tested for the total sample, France and UK over the period 2002-2009.

_{Dependent variable: LCDIV}

_Total _France _UK Independent variable Home -0.059 -0.164** -0.006 (-1.634) (-2.503) (-0.177) Control variables Owner1 -0.736*** -0.249** -0.161 (-9.668) (-2.511) (-1.223) Owner2 -0.290* 0.083 0.246 (-2.242) (0.309) (0.849) Profit 5.898*** 7.626*** 5.032*** (8.875) (9.181) (6.596) Size 2.312*** 2.346*** 2.361*** (156.537) (87.678) (469.367) Lev -1.235*** -2.373*** -0.711*** (-7.698) (-6.932) (-5.604) Risk -0.256 -1.295 -0.252 (-1.772) (-1.802) (-1.722) CG -0.324** -0.060 -0.387*** (-3.450) (-0.419) (-3.678) GO 0.226** 0.219* 0.207** (3.121) (3.358) (2.435) Constant -3.840*** -4.514*** -4.166*** (-20.239) (-20.594) (-32.440) Average number of observations per year 371 125 247

Average R-squared 0.80 0.79 0.82

(41)

-41-

Table 6: Fama MacBeth ordinary least squares regressions with the home bias and effect of the ownership percentage related to the cash dividends paid by dividend payers, tested for the total sample, France and UK over the period 2002-2009.

_{Dependent variable: LCDIV}

_Total _France _UK Independent variable Homeper -0.327*** -0.367** -0.230 (-3.670) (-3.110) (-1.778) Control variables Owner1 -0.536*** -0.036 -0.065 (-6.301) (-0.269) (-0.400) Owner2 -0.288* 0.057 0.242 (-2.347) (0.208) (0.817) Profit 5.896*** 7.560*** 5.030*** (8.798) (9.162) (6.576) Size 2.307*** 2.344*** 2.358*** (153.434) (99.977) (393.249) Lev -1.220*** -2.378*** -0.702*** (-7.739) (-6.978) (-5.696) Risk -0.269 -1.432 -0.234 (-1.874) (-1.847) (-1.562) CG -0.329*** -0.061 -0.390*** (-3.613) (-0.409) (-3.778) GO 0.222** 0.215** 0.204** (3.091) (3.312) (2.431) Constant -3.839*** -4.574*** -4.142*** (-21.859) (-23.175) (-33.994) Average number of observations per year 371 125 247

Average R-squared 0.80 0.79 0.82

The effects of the origin of shareholders on cash dividends