The value of family control, family ownership and
family management of European firms.
An agency problem approach
Author:
Gert-Jan Veltkamp
S1323709
January 2009
Rijksuniversiteit Groningen
Faculty of Economics and Business
MscBA Finance
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
1 Abstract
This study investigates the performance of European family controlled firms in the period 2005-2007 with respect to their governance variables. If members of one family own more than 25% of the voting rights of a company, then a firm is considered as a family controlled firm in this study. Using data of 250 firms from 10 countries, I find that family controlled firms do not significantly outperform nonfamily firms. The relationship between the share of ownership rights owned by a controlling family and the market to book value is significantly positive. This is due to the fact that a family owning a large fraction of the shares of a company has strong incentives to monitor the management. A wedge between voting rights and cashflow rights will decrease the value of the firm because a wedge provokes the conflict of interests between the controlling family and minority shareholders. I find that family involvement in the management board of family controlled companies on its own does hardly influence the valuation of these firms. The highest valued type of family firms are the ones that have no wedge between voting rights and cashflow rights and that have family involved in the management board.
Keywords: corporate governance, family firms, agency costs, management
Table of contents
Abstract………..………...1
Table of contents………..………....2
1. Introduction………...3
2. Theoretical background on family firms and firm value………...5
2.1 Agency problem 1……….6
2.2 Agency problem 2……….9
2.3 Resource based view………10
2.4 Hypotheses……….11
3. Review of empirical literature………...………..12
4. Method and data……….………17
4.1 Definition of variables………17
4.2 Data collection………17
4.3 Methodology………...18
5. Descriptive statistics………..………..20
6. Results……….……….25
6.1 Family control, ownership and performance………..25
6.2 Agency costs and founder (descendant) effects………...27
6.3 Family management and performance………...29
6.4 Agency problems and performance………31
6.5 Summary of results……… ………...34
7. Conclusion……….………..36
References………..……….39
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
3
1. Introduction
Family firms are very prominent in European countries. Table 1 shows their contribution to the GNP of a country. Family controlled firms have different characteristics compared to nonfamily firms. On the one hand, family firms can have positive features. For example, family firms are considered to have a long term investment horizon. On the other hand, family controlled firms may have negative characteristics. For example, the interests of family shareholders may conflict with the interests of minority shareholders.
This study investigates the performance of European family controlled companies. This study focuses on three governance characteristics of family controlled firms: management, voting rights and cashflow rights. I investigate whether and to what degree these governance variables affect the performance of family controlled firms.
Family involvement in the management board may influence the performance of family controlled firms. It is believed that founder managed family companies possess a competitive advantage relative to other family and nonfamily firms. The reason for this is that a founder as the entrepreneur whose vision, innovation, and ability to see opportunities causes creative destruction in the marketplace and enables the firm to capture extraordinary profits (Schumpeter, 1943). If a family controlled company is managed by a founder’s descendant, then this will have both positive and negative implications for its value. On the one hand, family connections inspire loyalty and family members are therefore willing to work long hours and be highly flexible in their work roles and assignments in order to help the firm succeed. However, a family member may not be most suitable candidate to fulfill an important management function. Academists study whether family firms that are managed by a founder perform differently than family firms that are managed by a descendant or an outsider. Barontini and Caprio (2006) and Villalonga and Amit (2006) find that the best performers are the family firms in which the founder holds a management position.
An increase of the voting rights and cashflow rights owned by family shareholders may affect the performance of family firms. First of all, due to an increase of cashflow rights or voting rights possessed by the controlling family, the controlling family will be more inclined to monitor the management (Burkart, Crom and Panunzi, 1997). This may mitigate the conflict of interests between management and shareholders. However, an increase in the voting rights owned by a family may induce the controlling family to expropriate the minority shareholders. Due to an increase in voting rights, a family will be better able to extract private benefits from the company. Furthermore, an increase of cashflow rights owned by a controlling family induces the controlling family to invest in projects that yield high free cashflows. Thus, an increase of the cashflow rights owned by a controlling family reduces the conflict of interests between minority and majority shareholders.
inclined to invest in projects that yield high private benefits, instead of investing in high free cashflow projects.
First of all, this study investigates whether the market to book value of European family controlled firms differ from the market to book value of nonfamily firms. Secondly, I study the relation between the degree of ownership rights hold by a controlling family and the market to book value. A large part of this study is devoted to the issue whether family management influences the market to book value of family controlled firms. Beside this, I study whether different theories can explain the valuation across firms. Finally, I categorize family controlled firms into four categories. This categorization is based on the fact whether family is involved in the management board of a family controlled company and on basis of a presence of a discrepancy between voting rights and cashflow rights owned by family shareholders. Due to this categorization of family controlled firms, I can study which type of family firm is the best performer and this helps me to separate the effects caused by family management and the wedge between voting and cashflow rights. This is the first study in Europe that makes this categorization of family firms.
There is no real consensus across previous studies about the performance of family controlled firms. The majority of these studies analyze the performance of family firms from the United States of America. Anderson and Reeb (2003) and McConaughy, Matthews and Fialko (2001) find that family firms outperform nonfamily firms, while Miller et al. (2007) and Villalonga and Amit (2006) do not find evidence that family firms outperform nonfamily firms. Barontini and Caprio (2006) and Maury (2006) study the performance of European family firms. They find that European family firms outperform nonfamily firms. Beside this, the majority of the studies conclude that founder managed family firms outperform other (non-) family firms (Barontini and Caprio, 2006; Anderson and Reeb, 2003; Villalonga and Amit, 2006; Miller et al., 2007). Control enhancing instruments reduce firm value. Barontini and Caprio (2006) and Villalonga and Amit (2006) find that a wedge between voting rights and cashflow rights reduce firm value. Villalonga and Amit (2006) find that family firms in which family members are present in the management board and in which no wedge exists between the voting rights and cashflow rights owned by a controlling family, are the best performers.
The sample of this study contains of 250 firms. The firms are selected on basis of the ownership data of the firms at 31-12-2005. If members of one family own more than 25% of the voting rights at 31-12-2005, then a firm is considered to be a family controlled firm. The selected firms are from 10 European countries. These countries are Austria, Belgium, Denmark, Germany, Holland, Italy, Norway, Spain and Sweden. Data about voting rights are from Bureau van Dijk’s Amadeus, CNMV (Spain) and Consob (Italy). Data about cashflow rights are retrieved from Bureau van Dijk’s Amadeus, annual reports and company websites. Data about the management of companies are retrieved from annual reports and websites of the companies.
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
5 between family possession of ownership rights and the market to book value. An increase of the voting rights owned by the controlling family will increase the market to book value. However, when the voting rights owned by a family increase with a higher portion than the cashflow rights owned by a family, then this will reduce the value of the firm. I find that family involvement in the management board of family controlled firm does not influence the value of the firm. I find evidence that the highest valued type of family firms are the ones that have no wedge between control rights and cashflow rights and that have family involved in the management board.
The rest of this thesis is organized as followed. In section 2 I will give a comprehensive overview about the theory behind firm value and family ownership, control and management. In section 3 I give a review of important empirical literature. In section 4 I will give an outline of the data and the methods used in this study. I will present the descriptive statistics in section 5. The results will be presented in section 6 and the conclusion is presented in section 7.
Table 1
The contribution of family firms to the GNP
(Table is from Groupement Européen des Entreprises Familiales, GEEF Benchmark study 2005.)
2. Theoretical background on family firms and firm value
This section deals with the theory behind the relation between family firms and firm value. There are various theories that analyze whether family control, ownership and/or management influence the value of a firm. The first theory I consider is the agency problem 1. This problem is the result of the conflict in interests between shareholders and managers. Hereinafter, I will outline the agency problem 2. This problem is the result of a discrepancy in interests between blockholders and minority shareholders. Finally, I consider the resource based view. This view suggests that family firms may create a competitive (dis-)advantage because of their unique organizational characteristics.
Country Contribution of family firms to the Gross National Product in 2001
Belgium 55%
Germany 55%
Holland 54%
France 60%
2.1 Agency problem 1
The agency theory deals with the conflict in interests between an agent and a principal. Berle and Means (1932) is the first study that recognizes the agency problem between shareholders and managers. They argue that a separation of ownership and control brings along a condition where the interests of the owner and manager may not be aligned and that discretionary power of managers exists. Berle and Means (1932) state: “have we any justification for assuming that those in control of the modern company will also choose to operate in the interest of the owners”. Therefore, they assume that managers want to maximize their own wealth, power and prestige while keeping a good reputation. This phenomenon is called moral hazard. Moral hazard by the management comes at the expenses of the shareholders, who want the value of their assets to be maximized. These costs are called agency costs, they reduce shareholder value.
The agency problem 1 is the result of the conflict in interests between shareholders and managers. If family controlled firms are not managed by members of the family, agency problem 1 will come to light. In this setting, managers might be inclined to take actions to satisfy their own private benefits at the costs of the family and other shareholders. However, a large fraction of family controlled firms are managed by members of the family. In the dataset of this study 63% of the family controlled firms have CEO’s with family relations. Jensen and Meckling (1976) argue that the agency problem will disappear when the agent and the principal have the same interests. Accordingly, family controlled firms that are managed by members of the family firms will not bear any agency problem 1 costs.
Other family firms are managed by outsiders, i.e. managers that are no relatives of the family. The agency problem 1 shows up in this case. A large blockholder, for example family shareholders, may influence this agency problem. To show the effect of family ownership on the agency problem 1 and the value of the firm, I will use the model of Burkart, Crom and Panunzi (1997). I show the implications of this model in the case a family controlled company is managed by a risk-neutral outside manager, owning no shares in the company.
The Burkart, Crom and Panunzi model assumes that a fraction α shares is held by a large family, while (1- α) are dispersed among small shareholders. Furthermore the shareholders are risk neutral and both the minority shareholders and the family have congruent interests. As a result, there is no wedge between voting rights and cashflow rights in this setting. The firm faces N projects, i∈ {1,2,3, …, N}, yielding verifiable security benefits ∏ i to the shareholders and non-verifiable private benefits bi to the manager. Project 1 is known and verifiable to both the shareholders and the managers. The payoffs of this project are ∏ i
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
7 Table 2
Payoffs from investment projects
Project N-1 Project N Probability
{∏, b} {∏,b} λ
{∏, 0} {0,b} 1-λ
Where∏>0, b>0 and ∏>b
The sum of both the private and the security benefits of all N+1 are smaller than zero. So the shareholders and managers will choose project 1 in the case they don’t verify the other projects.
At date 1 the manager chooses to exert a nonverifiable effort e∈[0,1] at cost e2/2. This effort makes it possible to learn the payoffs of all projects with probability e. At date 1 the family shareholders can choose to exert a nonverifiable effort E∈[0,1] at a cost E2/2. If the manager becomes informed at date 2, the shareholders become informed with probability E. If the manager is uninformed at date 2, then the family is also uninformed. At date 3 the payoffs are realized.
If nobody is informed, project 1 will be chosen, because the sum of both the private and the security benefits of all projects are smaller than zero. If the manager is informed and the family shareholders are uninformed at date 2, shareholders will follow the recommendation of the manager. If λ<1, in this case project N will be chosen. If λ<1 and both the manager and the shareholders are informed, project N-1 will be chosen.
Given E, the manager’s payoff (mp):
2
/
]
)
1
(
0
)
1
(
[
0
)
1
(
e
e
E
b
E
E
b
e
2mp
=
−
•
+
•
λ
+
−
λ
+
−
−
(1)Given e, the blockholder’s payoff (bp):
2
/
]
)
1
(
[
0
)
1
(
e
e
E
E
E
2sp
=
−
•
+
α
Π
+
−
αλ
Π
−
(2)Taking the first derivative of mp with respect to e:
e
b
E
b
E
e
mp
−
−
+
=
∂
∂
)
1
(
λ
(3)Reaction curve managers:
Taking the first derivative of sp with respect to E:
e
e
E
sp
Π
−
Π
=
∂
∂
αλ
α
(5)Reaction curve shareholders:
E=
α
Π
(
1
−
λ
)
e
(6)Now E and e can be solved.
)
)
1
(
(
1
)
1
(
)
(
2λ
α
λ
α
α
−
Π
+
−
Π
=
b
b
E
(7))
)
1
(
(
1
)
(
2λ
α
α
−
Π
+
=
b
b
e
(8)The total equity value V is the expected security benefits less the monitoring cost e2/2.
2
]
)
1
(
[
2E
E
e
V
=
λ
+
−
λ
Π
−
(9)The derivative of equation 9 with respect to α:
Π
−
+
∂
∂
+
−
Π
−
∂
∂
=
∂
∂
]
)
1
(
[(
]
)
1
(
[
e
E
e
E
E
V
λ
λ
α
λ
α
α
(10)From equation 10 it can be derived that an increase in ownership has both positive and negative consequences for firm value. The first term is positive, because
α
∂
∂E
>01. Due to an
increase in ownership, shareholders are more inclined to monitor the firm. Due to monitoring, information asymmetries between managers and shareholders decrease. Therefore the more valuable project, in term of total equity value, will be chosen. However
α
∂
∂e
<02, therefore the
second term in (10) has a negative influence on firms value. Shareholders are more inclined to monitor the firm when their fraction of shares increases. This reduces the manager’s initiative to exert a nonverifiable effort e.
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
9 To summarize, in the case the company is managed by the family, there are no agency problem 1 costs. Therefore agency problem 1 is mitigated in family managed family firms. If family controlled firms are runned by outsiders, the effect of a family blockholder can both be value enhancing and decreasing from agency problem 1 perspective.
2.2 Agency problem 2
The agency problem 1 is mitigated when a member of the family shareholder is present in the management board. Next to it, it was showed in the model of Burkart, Crom and Panunzi (1997) that a large blockholder, such as a family, can both mitigate and enhance the agency problem 1. However, agency problem 2 may appear when a family is the controlling shareholder in a company. Agency problem 2 shows up when the controlling family uses its controlling position to extract private benefits at the expenses of the minority shareholders. The controlling family may have incentives to extract private benefits that come at the expenses of the minority shareholders. Control enhancing instruments facilitate the controlling family with more power for minority expropriation.
I use the theory of Bebchuk et al. (2000) to show the effect of voting rights, cashflow rights and control enhancing instruments owned by a controlling family on the value of the firms. Consider the case that a family controlled firm can choose between two investment projects. The payoffs of these investment projects are shown in table 3. Vx is the total value of project x, which consists at the cashflow available for the shareholders (Sx) and the private benefits of control (Bx). Vy is the total value of project y, which consists at the cashflow available for the shareholders (Sy) and the private benefits of control (By). Suppose further that Bx>By and Vy>Vx, therefore the efficient project choice is project Y. The controlling family has a cashflow stake of α.
Table 3
Payoffs from investment projects
Project X Project Y
{Vx= Sx+Bx} {Vy=Sy+By}
The controlling family chooses project Y only if: (11)
For example, in family company ABC it is known that ∆B=0,03Vx and α=0,5. Therefore family
ABC will choose project y when the excess value of project y over project x is more than 3%3.
However, if α=0, then the controlling family ABC will choose project Y if the excess value of
project y over project x is more than 27%4. Thus, when the cashflow stake of the controlling
family increases, then the probability that the family will choose the efficient project is larger. When both the cashflow stake and the voting rights of the family in a family firm rise, this will have a two side effect on its value. First of all, it appears that a higher cashflow stake will lead to a higher firm value. However, when the voting rights owned by a controlling family increase too, then it will be easier for the family to extract private benefits from the investment projects. Therefore the probability that the family will choose the inefficient project is larger, which reduces value.
A wedge between voting rights and cashflow rights reduces the value of the company.
For example, if company ABC issues new shares at the stock market without voting rights, then the family is still able to extract the same private benefits from the projects, because their control power remains the same. However, its cashflow stake α will decrease. Therefore the probability that the family will choose the inefficient project is larger. Thus a wedge between voting and capital rights will have a negative impact on the value of family firms.
2.3 Resource-based view
The resource based view suggests that firms with assets that are valuable, rare and non-imitable are able to create a sustainable competitive advantage and shareholder value. In relation to family firms, they study whether family firms bring with them unique assets. Human capital is an example of a resource that can give a company a distinctive competence (Dyer, 2006). The management board of a company is an important governance variable and is a part of the human capital of the company.
Family firms can be managed by the founder, a founder’s descendant or outsiders. It is common believed that the founder of a firm, i.e. the entrepreneur, has unique competences and talents. Schumpeter (1934) argues that an entrepreneur has competences that help a company to grow and succeed. The basic competences of an entrepreneur are his entrepreneurial spirit, his unique know-how and skills. Schumpeter argues that one of the most valuable commodities for a company is the entrepreneur whose vision, innovation, and ability to see opportunities to cause creative destruction in the marketplace and to enable the
firm to capture extraordinary profits. Therefore, founder managed companies possess a
competitive advantage relative to other family and nonfamily firms.
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
11 When a company is managed by a founder’s descendant this can have both positive and negative implications for the value of the company. At first, Rosenblatt et al. (1985) argue that family members will be more motivated and committed to the business, because they are more naturally part of the company. Such family connections inspire loyalty and family members are therefore willing to work long hours and be highly flexible in their work roles and assignments in order to help the firm succeed. Second, family members have often been socialized at a very early age to understand the nature of the business, its customers, and its competitors, and have received hands-on training from family leaders who are knowledgeable and highly skilled (Dyer, 1992). However, the family may not be able to supply the firm with enough talented employees to manage the key operations. The restricted nature of the human resource pool supplied by the family means that the family may not have enough qualified personnel to operate a business successfully unless they recruit nonfamily employees to fill key positions. Dyer (1989) argues that family members who are incompetent may be placed in key positions. This will have negative implications on firm value.
2.4 Hypotheses
Figure 1 illustrates the relation between different theories, governance characteristics and the market to book value of family firms. This figure is based on the theory mentioned in section 2.1, 2.2 and 2.3. From these theories a number of hypotheses can be deduced.
Figure 1
Family Firms Compared to Widely Dispersed Firms
This figure shows whether the agency theory and the resource based view predict whether a governance characteristic should be positively (+) or be negatively (-) valued.
Hypothesis 1: A family firm may have a higher or lower M/B-ratio than a widely dispersed firm.
Hypothesis 3: A wedge between voting and capital rights will have a negative impact on the performance of family firms.
Hypothesis 4: A family firm in which the founder functions as a CEO will be positively valued.
Hypothesis 5: A family firm with a descendant functioning as the CEO may be positively or negatively valued.
Hypothesis 6: A family firm with a founder functioning as a management board member (not CEO) may be positively or negatively valued. Hypothesis 7: A family firm with a founder’s descendant functioning as a
management board member (not CEO) may be positively or negatively valued.
I grouped family firms into four types, according to the presence or absence of agency problems in each type. Hypotheses 8, 9, 10 and 11 relate to these categories of family firms. Type I family firms are firms that have a wedge between voting rights and cashflow rights and these firms have a family member present in the management board. These firms are dealing with agency problem 1, but not with agency problem 2. Type II family firms have a wedge between voting rights and cashflow rights and do not have a family member in the management board. These firms deal with both agency problems. Type III family firms do not have a wedge between voting rights and cashflow rights and have a family member present in the management board. These firms do not cope with agency problems. Type IV family firms do not have a wedge between voting rights and cashflow rights and do not have a family member present in the management board.
Hypothesis 8: Type I family firms will have a lower M/B ratio than type III family firms.
Hypothesis 9: Type II family firms will have a lower M/B ratio than type III family firms.
Hypothesis 10: Type IV family firms will have a lower M/B ratio than type III family firms.
Hypothesis 11: Nonfamily firms will have a lower M/B ratio than type III family firms.
3. Review of empirical literature
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
13 of the ultimate voting rights. Only family firms worth more than €300 million are included in their sample. The dataset of their study includes firms from Belgium, Denmark, France, Germany Italy, The Netherlands, Spain, Switzerland, Finland, Norway and Sweden. Using a sample of 675 publicly traded firms, Barontini and Caprio find that the effect of family control is positive. However, they find that part of this positive effect is wasted by the use of control enhancing instruments. Barontini and Caprio find that founder managed family companies outperform other firms. They do not find evidence that family controlled companies that are managed by founder descendants perform differently than nonfamily firms.
Maury (2006) also takes 10% of the ultimate voting rights as the under limit for the existence of family control. Using a sample of 1672 non-financial firms from 13 Western European countries, Maury concludes that family control is associated with positive valuations and higher profitability as compared to nonfamily firms. Family controlled firms in which the family holds one or more top officer positions are not valued higher than other family firms.
Sraer and Thesmar (2004) label firms as a family firm, if a family controls at least 20% of the shares. Using a sample of 470 nonfinancial publicly listed firms, they conclude that family controlled firms significantly outperform widely dispersed firms. Furthermore, they find that founder managed companies are very profitable. They find that both descendant managed and outside managed family firms outperform nonfamily firms.
Although there is little evidence about the performance of European family firms, more evidence exists about the performance of American family firms. There is no real consensus in the literature from the USA whether the complete group of family firms outperforms nonfamily firms. According to the theory, as outlined in the previous paragraph, this lack of consensus should not be a surprise. Anderson and Reeb (2003) and McConaughy, Matthews and Fialko (2001) found that the whole group of family firms outperform nonfamily firms, while Miller et al. (2007) and Villalonga and Amit (2004) did not find evidence that the complete group of family firms outperformed nonfamily firms. However, no one concluded that there is a negative relation between family firms and value.
Researchers in the USA also study the performance of family firms along different governance variables. Miller et al. (2007) and Villalonga and Amit (2004) conclude that founder managed and controlled firms significantly outperform other family firms and non-family firms. This in line with the theory as is outlined in figure 1. This framework suggests that both the agency theories and resource based view predict that founder managed family firms should outperform all other (non-) family firms. In table 4 the findings of similar studies are summarized.
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
17
4. Method and data
4.1 Definition of variables
There is not a universal definition of a family controlled firm. For example, Villalonga and Amit (2006) define a family firm as a firm that has the founder of the firm or his descendant working as a manager or a blockholder. But Sraer and Thesmar define a family firm as a firm in which the family controls more than 20% of the shares. In this study I will label a company as a family controlled firm when the family controls more than 25% of the voting rights. In various regressions I will further categorize family controlled companies in founder managed family companies, Type I or as type II family firms et cetera.
The market to book value is a valuation ratio. In this study the market to book value is the market value of equity divided by the book value of equity at the end of a year. A higher market to book value implies that investors expect that the firm will create more value from a given set of assets. Market to book values may vary across industries, countries, years, size and gearing of firm. Some industries require a higher intensity of capital than others. Manufactures of cars will usually trade at market to book values lower than, for example, consulting firs. For this reason, one has to control for these variables when measuring family firm effects. The industry is in this study defined as the two digit NACE code in which the company operates. The size of a company is the market value of total assets at the end of each year. Gearing is defined as the book value of long term financial debt divided by the book value of equity at the end of each year.
The cashflow rights are defined as the dividend rights owned by the family divided by the total dividend rights. There may be a discrepancy between cashflow right and voting rights. A wedge may be caused by the issue of dual class shares or by a pyramidal ownership structure. For example, Heineken Holding possesses 50,01% of the shares of Heineken N.V. At her turn, ms. Carvalho-Heineken possesses 58,81% of the shares of the Heineken Holding. The voting rights of ms. Carvalho-Heineken are 50,01% in Heineken N.V., that are the voting rights held in the weakest link of the chain. The cashflow rights of ms. Carvalho-Heineken in Heineken N.V. are 29,405%, that is a multiplication of the whole ownership chain.
4.2 Data collection
There are four main stages in the data collection process in this study:
• The selection of family firms and nonfamily firms and collection of ownership and control data. In this study a company is considered as a family firm if the family holds at least 25% of the control rights in 2005. The family is here defined as the founder of the company or as the founder’s descendants. Founder descendants must have the same family name as the founder to be considered as a relative. To search for family firms within Amadeus, individuals or a family should hold more than 25% of the control rights. To select the family firms in the 10 European countries, I used the ‘selection by shareholder characteristics’ tool in Bureau van Dijk’s Amadeus. I selected companies that are for 25% or more controlled by an individual or a family. A company is considered as a nonfamily firm if a firm is by less than 15% controlled by an individual or a family and by less than 15% controlled by the government and institutional investors. Therefore the nonfamily sample consists mainly at widely dispersed firms. I do not want that the nonfamily firm sample contains large blockholders, for example the government, that may influence the agency problems. • Verification of ownership and control data. After the selection stage I verified in
Amadeus, annual reports, company websites, CNMV (Spain) and CONSOB (Italy) if the family really holds more than 25% of the voting rights. If it appeared from one source that the family holds less than 25% of the voting rights, this firm was skipped from the family firm sample.
• The collection of management data. Data about the management of firms is retrieved
from company websites and annual reports. A founder’s descendant CEO or other manager should have the same family name as the founder to be considered as a relative.
• The collection of financial data. The market to book value and size of companies are
retrieved from Tomson’s Datastream. NACE industry codes and gearing are retrieved from Amadeus.
4.3 Methodology
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
19 urge the share of ownership that a family holds in a company. So, a family may increase its ownership share in a family controlled firm when the firm performs well. By using the lagged values of the ownership variables, the endogeneity problem is shrinked.
An observation of the market to book value is identified as an outlier if it exceeds the value of the 99th percentile observation, or if it is below the value of the 1st percentile observation. Then, the values of the large outliers are changed into the value of the 99th percentile observation and the values of small outliers are changed into the value of the 1st percentile observation.
I use an Ordinary Least Squares regression model to analyze the effects of family controlled firms, family ownership, family management and the control variables on the market to book value. In various regressions family controlled firms will be further categorized according to the presence of agency problems in these firms. Industry and country effects are assumed to be fixed effects. The fixed effects are dummy variables of nine countries and dummy variables of 36 two-digit NACE codes. Due to perfect multicollinearity, it is impossible to include all countries and all two-digit NACE codes as dummy variables. The general form of the regression equation I employ in the multivariate analysis of this study takes the form as follows:
Market_to_Booki=α0+ α1Austria+ …+ α8Spain+α11NACE15+…+α47NACE92+β0Family_firmi,2005
+β1Governance_variablesi,2005+β2Gearingi+β3LN(size)i
+β4(LN)agei+εit (13)
Market_to_Booki The average market to book value of a company over the years 2006 and 2007.
Austria, Belgium etc.
Country dummy variables that are equal to one if a company is settled in the corresponding country. These variables capture country fixed effects.
NACE15, NACE16 etc.
Industry dummy variables. These variables equal to one if a company operates in corresponding industry. These variables capture fixed industry effects.
Family_firmi
Dummy variable that equals to one when a company is family controlled at 31-12-2005. In various regressions more dummies will be used to address the different categories of family firms.
Governance_variablesit
In various regressions the %ownership, the wedge between voting and capital rights and management characteristics of a company are included as independent variable.
Gearing i The average of the long term financial debt divided by the book value of equity of a company over the years 2006 and 2007.
LN(size) i The natural logarithm of the average market value of a company over the years 2006 and 2007.
5. Descriptive statistics
Table 5 shows the number of family and nonfamily firms by NACE code. It shows that the presence of family firms across the different industries is not uniformly distributed. Family firms are in the sample not present in the air transport, manufacture of basic metals, post and telecommunications and other service industries. On the other hand, there are nine industries that are entirely composed of nonfamily firms. Because family and nonfamily firms are not uniformly distributed across industries, it is important to control for industry effects in the regressions that are presented in paragraph 6.
Table 5
Number of Nonfamily and Family Firms by Two-Digit NACE Code
This table classifies the number and percentage of family firms in the sample by two-digit NACE industry classification code. Family firms are defined as those in which one or more family members hold more than 25% of the voting rights. The sample consists of 250 firms that are listed in Austria, Belgium, Denmark, France, Germany, Holland, Norway, Italy, Spain and Sweden during 2005-2007.
NACE
Code Industry Description
All firms Family firms Nonfamily firms Family firms in Industry (%)
15 Manufacture of food products and beverages
8 6 2 75,00%
17 Manufacture of textiles 3 3 0 100,00% 18 Manufacture of wearing apparel; dressing
and dyeing of fur
7 6 1 85,71%
21 Manufacture of pulp, paper and paper products
2 2 0 100,00%
22 Publishing, printing and reproduction of recorded media
4 4 0 100,00%
24 Manufacture of chemicals and chemical products
13 5 8 38,46%
25 Manufacture of rubber and plastic products 9 7 2 77,78% 26 Manufacture of other non-metallic mineral
products
4 3 1 75,00%
27 Manufacture of basic metals 4 0 4 0,00% 28 Manufacture of fabricated metal products,
except machinery and equipment
4 3 1 75,00%
29 Manufacture of machinery and equipment 12 8 4 66,67% 30 Manufacture of office machinery and
computers
2 1 1 50,00%
31 Manufacture of electrical machinery and apparatus
5 3 2 60,00%
32 Manufacture of radio, television and communication equipment and apparatus
9 4 5 44,44%
33 Manufacture of medical, precision and optical instruments, watches and clocks
6 3 3 50,00%
34 Manufacture of motor vehicles, trailers and semi-trailers
5 3 2 60,00%
35 Manufacture of other transport equipment 1 1 0 100,00% 36 Manufacture of furniture; manufacturing 5 3 2 60,00% 40 Electricity, gas, steam and hot water
supply
3 1 2 33,33%
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
21 Table 5 (continue)
Table 6 provides statistics about the sample, split up in family and nonfamily firms. Family firms represent 60% of the total sample. In the studies of Barontini (2006) and Caprio and Maury (2006) family firms represent respectively 51% and 70% of the total sample. It appears from table 6 that family firms have a slightly higher market to book value than nonfamily firms: 2,80 of family firms to 2,76 of nonfamily firms. However, this difference is not statistical significant.
Family firms are significantly smaller than nonfamily firms. The finding that family firms are smaller than nonfamily firms is confirmed by other studies, like Villalonga and Amit (2006) and Barontini and Caprio (2006). Nonfamily firms are higher, although not statistical significant, leveraged than nonfamily firms. Furthermore, family firms are somewhat younger than nonfamily firms. Younger firms are in general higher valued than older firms, this is the so called young firm effect. Therefore, the natural logarithm of the firm’s age is included as control variable in the regressions.
NACE Code
Industry Description All
firms Family firms Nonfamily firms Family firms in Industry (%)
50 Sale, maintenance and repair of motor vehicles and motorcycles; retail sale of automotive fuel
1 1 0 100,00%
51 Wholesale trade and commission trade, except of motor vehicles and motorcycles
29 18 11 62,07% 52 Retail trade, except of motor vehicles and
motorcycles; repair of personal and household goods
6 4 2 66,67%
55 Hotels and restaurants 5 5 0 100,00%
61 Telecommunications 4 3 1 75,00%
62 Air transport 1 0 1 0,00%
63 Supporting and auxiliary transport activities; activities of travel agencies
1 1 0 100,00%
64 Post and telecommunications 2 0 2 0,00% 71 Renting of machinery and equipment
without operator and of personal and household goods
5 5 0 100,00%
72 Computer and related activities 36 20 16 55,56% 73 Research and development 4 1 3 25,00% 74 Other business activities 31 14 17 45,16% 75 Public administration and defence 1 1 0 100,00% 85 Health and social work 2 1 1 50,00% 90 Sewage and refuse disposal, sanitation
and similar activities
1 1 0 100,00%
92 Recreational, cultural and sporting activities
7 4 3 57,14%
93 Other service activities 1 0 1 0,00%
Table 7 presents statistics about family firms that are categorized by the composition of their executive board. It seems that most family controlled firms have the founder as CEO: 62 out of the 151 family firms. It appears that these firms have a slightly higher market to book value than nonfamily firms. Furthermore, family controlled firms in which the founder functions as the CEO are the smallest, in terms of total assets, category of family firms. This is not a surprising result, because these firms are often in the constructive phase of their lifecycle. Obviously, these firms are the youngest companies. To control for the young firm effect, the natural logarithm of the age of the company will be included as control variable.
Family controlled companies that have a founder descendant as CEO are very common. 33 out of the 151 family firms in this sample have a founder descendant as CEO. These firms have a somewhat lower market to book value than nonfamily firms. They are smaller and older than nonfamily firms. Family controlled firms in which the founder fulfills another management function are very rare. Hence, this group can not be described reliably. Family controlled firms in which the executive board is entirely composed of outsiders, are the largest family firms in terms of total assets. Because the size of the company differs across different categories of family firms, it is important to include the natural logarithm of the market value of a company as control variable.
Table 6
Descriptive statistics of the sample
Means, standard deviations and test of differences between family and nonfamily firms. Family firms are defined as those in which one or more family members hold more than 25% of the voting rights. The sample consists of 250 firms listed in Austria, Belgium, Denmark, France, Germany, Holland, Norway, Italy, Spain and Sweden during 2005-2007. Asterisks denote statistical significance ate the 1% (***), 5% (**) or 10% (*) level.
[a] All firms [b] Family firms [c] Nonfamily firms Diff. in Means Mean Std Dev Mean Std Dev Mean Std Dev [b-c] t-stat
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
25
6. Results
6.1 Family control, ownership and performance
In this paragraph I try to unravel whether family controlled firms are differently valued than nonfamily firms. In table 8 the results of four different regressions are shown. The independent variables family firms, family voting rights and wedge that are presented in this table are of the main interest. In regressions 1 and 3 the wedge between voting and capital rights is not included among the independent variables. Regressions 2 and 4 include the wedge as an independent variable. The theory section of this study claims that a wedge between voting and cashflow rights should be value decreasing. For this reason, the coefficients of the variables family firms in regression 2 and family voting rights in regression 4 represent the effect of family control net of the effect caused by the wedge. Thus, the coefficients of the variables family firms and family voting rights in regression 2 and 4 are not influenced by the use of control enhancing instruments.
Table 8 shows that family control has a slightly positive outcome on the market to book value of firms. The coefficients of the variables family firms and family voting rights are all positive. There is no statistical significant evidence that the entire group of family controlled firms outperform nonfamily firms because the coefficients of family firms are statistically insignificant. However, there is significant evidence that the coefficient of family voting rights in regression 4 is positive. That is the effect of family voting rights, separated from the effect of control enhancing instruments. This means that an increase of voting rights owned by the controlling family is associated with a higher market to book value, on the condition that the wedge between voting rights and cashflow rights owned by a controlling family does not increase.
Regressions 2 and 4 show that the use of control enhancing instruments reduces firm value. The average wedge of family controlled firm is 4%. Therefore, the market to book value of an average family controlled firm is reduced by 0,2216 points5, due to the use of control enhancing instruments. It can be derived from the results presented in table 8 that an increase of 1% of the family ownership of both the capital rights and voting rights, will
increase market to book value by 0,01 point6. However, when the ownership of voting rights
increase by 1% and the ownership of capital rights remain the same, the market to book value will decrease by 0,05 points7.
family in management can in some instances be value decreasing. If there is no family involved in the board of family controlled firms, then family control will enhance monitoring of the management by the family. Due to monitoring, information asymmetries between managers and shareholders decrease. Therefore moral hazard of the manager will decrease. This has positive consequences for the market to book value. However, because managers are intensively monitored by the shareholders, they have fewer incentives to verify the most profitable investment project. This has negative consequences for the market to book value. Family controlled firms are a group of firms that consist of firms with different governance characteristics, what could be the explanation for the fact that there is no statistical evidence that the entire group of family controlled companies outperform nonfamily firms.
The finding that the coefficient of family voting rights is positive can be explained by agency problem 1. As already mentioned, if it is assumed that there is no wedge between voting and cashflow rights, then an increase of voting rights will induce the family shareholders to monitor more intensively the firm. The model presented in paragraph 2.1 argues that monitoring can both enhance and decrease firm value. I find that if the voting rights possessed by the family increase and there is no wedge between voting and cashflow rights in a family firm, then this will have a positive effect on the market to book value. Therefore, I can conclude that positive effects of family monitoring are larger than the negative effects of family monitoring.
Table 8
Performance and Family Control In the table the results of the regressions are shown, with as dependent variable the Market to Book value. Family Firm is a dummy variable, equals one in the case the family owns more than 25% of the voting rights. Family voting rights is the percentage of voting shares owned by the controlling family in a family controlled firm. Wedge is the difference between the share of voting rights and cashflow rights hold by the family. LN(size) equals to natural logarithm of the firm’s market value. Gearing is the book value of debt divided by the book value of equity. LN(age) is the natural logarithm of the age of a company. Among the regressors the effects of the NACE code, countries and years are included.
Asterisks denote statistical significance at the 1% (***), 5% (**) or 10% (*) level.
Dependent variable Market to Book
(1) (2) (3) (4)
Family Firm 0,23 (0,62)
0,43 (1,14)
Family voting rights 0,71
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
27 Agency problem 2 states that control enhancing instruments enlarge the chance that firms invest in less efficient projects. By the use of such instruments, a controlling family will invest in projects that deliver higher private benefits. The findings in this study confirm the assumption that control enhancing instruments reduce firm value. The negative effect of the wedge on the market to book value is in absolute terms larger than the positive effect of the family ownership of voting rights. This may imply that the conflict of interests between majority an minority shareholders is more value destructive than the conflict of interests between managers and shareholders. In paragraph 6.2 and 6.4 I will further investigate whether this is the case.
The results of the regressions shown in table 8 are in line with existing empirical literature. Villalonga and Amit (2006) and Barontini and Amit (2006) find statistical evidence that family firms outperform nonfamily firms. And others, like Miller et al. (2007) do not find statistically significant evidence that family firms have a higher market to book value. Claessens et al. (2002), Villalonga and Amit (2006) and Barontini and Caprio (2006) find that a wedge between voting and capital rights decreases the value of a firm.
6.2 Agency costs and founder (descendant) effects
In section 6.1 I focus on the effect of family control on the market to book value. The effects of family control were not explicitly distinguished between agency theory 1, agency theory 2 and the resource based view. In this paragraph I study explicitly whether agency theory 1, agency theory 2 and founder (descendant) effects can explain the valuation across firms.
I explain in section 2 that three theories analyze the performance of family controlled firms, these theories are the resource based view and the agency theories. The resource based view states that a founder has unique entrepreneurial skills. Therefore, a founder holding a CEO position may enhance firm value. On the other hand, a founder’s descendant may not be the best candidate to fulfill the CEO position. Agency problem 1 is the conflict of interests between managers and shareholders. If a member of the controlling family is present in the management board, then this problem is mitigated. Agency problem 2 refers to the conflict of interests between the controlling family and minority shareholders. A family controlled firm faces agency problem 2, if there is a wedge between the voting rights and cashflow rights owned by the controlling family.
statistical evidence that these kinds of family firms are superior performers may lay in the fact that the sample of this study is not large enough.
There is no statistical evidence from both regression 1 and 3 that a family controlled firm that is managed by a founder descendant is differently valued than other firms. The signs of the coefficients of both the dummy variable agency problem 1 and the dummy variable agency problem 2 are negative. These results imply that agency problems reduce the value of firms. However, only the coefficients of agency problem 2 in regression 2 and regression 3 are statistical significant. The coefficients of agency problem 2 are far more negative than the coefficients of agency problem 1. Therefore, it can be concluded from these results that agency problem 2 is more value decreasing than agency problem 1.
I add the values of the variance inflation factor of the different variables in regression 3, to show that there is a strong degree of multicollinearity across the independent variables in this regression. This multicollinearity exists because if a family controlled firm that has a founder or a founder descendant operating as a CEO, this automatically implies that it does not face agency problem 1. Therefore it is hard to separate the founder (descendant) effect and agency 1 costs.
Table 9
Agency 1 costs, agency 2 costs and founder effects across family controlled firms In the table the results of the regressions are shown, with as dependent variable the Market to Book value. Founder CEO is a dummy variable, equals one in the case that the family firm is managed by its founder. Founder descendant CEO is a dummy variable, equals one in the case that the family firm is managed by a founder’s descendant. Agency problem 1 is a dummy variable, equals one if a family controlled is managed by outsiders. Agency problem 2 is a dummy variable, equals one if there is a wedge between the voting and cashflow rights of a controlling family in a family controlled firm. LN(total assets) equals to natural logarithm of the firm’s market value. Gearing is the book value of debt divided by the book value of equity. LN(age) is the natural logarithm of the age of a company. Among the regressors the effects of the NACE code, countries and years are included. Asterisks denote statistical significance at the 1% (***), 5% (**) or 10% (*) level. Market to Book (1) (2) (3) VIF Founder CEO 0,40 (0,94) 0,39 (0,56) 4,38 Founder descendant CEO -0,31
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
29 Summarizing, family controlled firms hardly increase the value of their firm by adding family in the management board of their firm. However, a family controlled firm does increase its value when it removes the wedge between control and cashflow rights. Is shall be noted that there are different types of family involvement in the management board. Therefore I will further examine the effects of family management in the next paragraph.
6.3 Family management and performance
In paragraph 6.2 I study the founder effect, the founder descendant effect and agency costs of family controlled firms. In this paragraph I will further examine whether the management characteristics influence the value of family controlled firms relative to nonfamily firms. I categorize family firms into five groups: family controlled firms in which the founder is CEO, family controlled firms in which the founder descendant is CEO, family controlled firms in which the founder is a member of the executive board, family controlled firms in which the founder descendant is a member of the executive board and family controlled firms in which no family is present in the board. I study which categories of family firms outperform nonfamily firms.
Agency theory 1 and the resource based view predict that the effect of a founder CEO should be positive. If the founder of a family controlled firm is both CEO as blockholder, agency problem 1 is mitigated. Therefore agency costs are avoided, which will have a positive effect on the valuation of these type family firms. However, as appeared from previous paragraph, agency 1 costs are not very high. Thus, the higher valuation of this type of family controlled firms will be mainly caused by the argument that entrepreneurs, i.e. the founders of a company, have a unique set of competences (Schumpeter, 1934). The finding that this type of firms should be higher valued is confirmed by the study of Barontini and Caprio (2006). There is no significant statistical evidence that family controlled firms in which founder descendant acts as a CEO are differently valued. This is in line with the theory and the results of other empirical studies. The theory states that that founder descendant’s acting as a CEO
Table 10
Performance, Family Control and management
In the table the results of the regressions are shown, with as dependent variable the Market to Book value. Founder CEO is a dummy variable, equals one in the case that the family firm is managed by its founder. Founder descendant CEO is a dummy variable, equals one in the case that the family firm is managed by a founder’s descendant. Founder is a member of the executive board, but not as a CEO is a dummy variable that equals to one in the case the family firm is managed by an outside CEO, although the founder is a member of the management board. A founder descendant is a member of
the executive board, but not as aCEO is a dummy variable that equals to one in the case
the family firm is managed by an outside CEO, although the founder’s descendant is a member of the management board. No family on board is a dummy, equals to one if there is no family present in the management board. Wedge is the difference between the share of voting rights and cashflow rights hold by the family. LN(total assets) equals to natural logarithm of the firm’s market value. Gearing is the book value of debt divided by the book value of equity. LN(age) is the natural logarithm of the age of a company. Among the regressors the effects of the NACE code, countries and years are included. Among the regressors the effects of the NACE code, countries and years are included.
Asterisks denote statistical significance at the 1% (***), 5% (**) or 10% (*) level.
Dependent variable Market to Book
(1) (2)
Founder CEO 0,43
(0,94)
0,55 (1,19) Founder descendant CEO -0,26
(-0,45)
0,09 (0,15) Founder is a member of the executive
board, but not as a CEO
-1,31 (-0,38)
-0,92 (-0,27) A founder descendant is a member of
the executive board, but not as a CEO
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
31 could both enhance and decrease the market to book value. First, the agency problem 1 is mitigated in this case. This will enhance the value of a company. However, the resource based view states that there may be more talented candidates than a founder descendant’s to fulfill the CEO function. Like this study, Barontini and Caprio (2006) find no significant evidence that family controlled companies in which the founder descendant acts as a CEO are differently valued. The other categories do also not deliver any statistical significant results. It appears that family controlled firms that have no family in the board, do not perform very differently than nonfamily firms. Barontini and Caprio (2006) also find that this category of family controlled firms is not differently valued than nonfamily firms.
6.4 Agency problems and performance
As presented in previous paragraphs, there is evidence that a wedge between voting rights and cashflow rights is value destructive. In paragraphs 6.2 I show that family controlled firms without family in the management board, hardly bear any agency 1 costs. In paragraph 6.3 I argue that family involvement in the management board of family controlled companies barely influence the valuation of these firms. To further examine the effects of agency problem 1 and agency problem 2 in family controlled firms, family controlled firms are in this paragraph typed into four categories on basis of the presence of agency problems in these firms. Therefore I can investigate which type of firm does outperform and which agency problem is most costly.
As appears from table 11, type III family firms have indeed the highest average market to book value. The market to book value of type III family firms is not statistical significant higher than the market to book value of other firms. In table 12 I present the results of the regressions with market to book value as dependent variable and the different type of firms, country, ln(age), ln(total assets), gearing and industries as independent variables. It seems that type III family firms are significant higher valued than type I family firms.
Type III family firms are not higher valued than type IV family firms. As already mentioned, type IV family firms deal only with agency problem 1. Because family controlled firms that have control enhancing mechanisms are significantly lower valued, it can be concluded that agency 2 costs are more value destructive than agency 1 costs. This is in line with the findings in paragraph 6.2, in which I also argue that agency 2 costs are more costly than agency 1 costs.
It is very worthwhile to note that type III family firms do not have a statistical significant higher market to book value than type II family firms in table 12. If it is noticed that type III family firms have a higher market to book value than type I family firms, one should expect that a firm that deals with both agency problem 1 and agency problem 2, will be valued lower than a firm that deals with just agency problem 2. What is the reason that this does not appear from table 12? I think that a family firm that has family on the board and that has a positive wedge between voting and capital rights is better able to expropriate minority shareholders. If there is no family in the management board, the management of a family controlled company will be more inclined to consider the interests of minority shareholders too. Therefore, the presence of the family in the executive board reinforces agency problem 2.
The coefficient of type IV family firms is slightly higher, although not statistical significant, than the coefficient of nonfamily firms. Both categories face the same agency
Table 11
Agency problems and market to book
The numbers in each cell is the average market to book value. The agency problem 1 is present if a family member is absent in the executive board. The agency problem 2 is present if there is a gap between the voting and the cashflow rights. Family firms are defined as those in which one or more family members holds more than 25% of the capital or the voting rights. The sample consists of 250 firms listed in Austria, Belgium, Denmark, France, Germany, Holland, Norway, Italy, Spain and Sweden during 2005-2007.
Agency problem 1 Differences (t-stats) Agency problem 2 No Yes
Yes Type I family firms Type II family firms I-II
2,46 2,50 -0,04
(-0,06) No Type III family firms Nonfamily firms III-Nonfamily firms
The value of family control, family ownership and family management of European firms An agency problem approach
G.J. Veltkamp
33 problem. The reason that the coefficient of type IV family firms is slightly higher than the coefficient of nonfamily firms may be due to the effect of family control. I find in paragraph 6.1 some evidence that family control is associated with a higher market to book value. Therefore this may explain the difference between the coefficients of these types of firms. However, because I do not find statistical significant evidence for this discrepancy, I am very careful to draw this conclusion.
Like this study, Villalonga and Amit (2006) conclude that type I family firms are lower valued than type III firms. Next to it, Villalonga and Amit find that type III family firms are higher valued than nonfamily firms. It should be noted that Villalonga and Amit (2006) did not correct for industry, time and firms specific effects when they studied the effect of agency problems on the market to book value.
Table 12
Agency problems and performance
In the table the results of the regressions are shown, with as dependent variable the Market to Book value. Type I is a dummy variable that equals to one if a family firm have family members present in the management board and have a gap in voting and cashflow rights. Type II is a dummy variable that equals to one if a family firm have no family present in the management board and have a gap in voting and cashflow rights. Type IV is a dummy variable that equals to one if a family firm has no family present in the management board and has no gap between voting and capital rights. Nonfamily is a dummy variable that equals to one if a firm is not family controlled or owned. Family firms are defined as those in which one or more family members holds more than 25% of the voting rights. The sample consists of 250 firms listed in Austria, Belgium, Denmark, France, Germany, Holland, Norway, Italy, Spain and Sweden during 2005-2007. Asterisks denote statistical significance at the 1% (***), 5% (**) or 10% (*) level.
