1 Introduction

Family firms are the predominant form of business in economies around the world, and they contribute extensively to gross national products and job creation (IFERA 2003). A strategic theme in these firms is the generational transition (Morris et al. 1997). Although a long term commitment to continue the firm as a family firm is often a dream of the founder, only about one-third of these firms are successfully transferred into the second generation and barely ten percent continue into the third one (Birley 1986; Neubauer and Lank 1998). Several positive (e.g., a successful acquisition) as well as negative (e.g., failure) explanations exist for this pattern (Raskas 1998). Nevertheless, the exit or failure of a significant number of these family ventures could be avoided by implementing good functioning governance mechanisms, such as a board of directors. These mechanisms may stimulate survival among a larger group of businesses, as they enhance the cohesiveness of the family controlling the company and corporate performance (Neubauer and Lank 1998; Schulze et al. 2001).

Family firm governance and accountability differ from practices in non-family businesses. Most family business definitions emphasize the idea of family influence through ownership and management (Chua et al. 1999). Hence, governance entails avoiding conflicts between the family members’ family and business roles, while preserving unity among the family members (Lane et al. 2006). Over the past few years, academics have extensively studied the distinguishable characteristics of family firm governance systems (cf. Corbetta and Montemerlo 1999; Davis and Pett 2000; Schulze et al. 2001; Lane et al. 2006; Miller and Le Breton-Miller 2006; Van den Heuvel et al. 2006). However, fairly little is known about the relationship between the generation in charge of the company and its governance design.

As firms evolve, different forms of corporate governance may be needed (Filatotchev and Wright 2005). A number of life cycle perspectives have been provided, but these generally focus on the business system while family firms are more complex (Morris et al. 1997). The generation in charge of the family business is a central component of its life cycle as important changes in family attributes will occur (Gersick et al. 1997; Steier 2001). In the academic literature, in depth theoretical explanations of expected governance patterns over generations are just emerging and these are mainly grounded within an agency theory framework (Schulze et al. 2003; Lubatkin et al. 2005). In the appendix we provide an overview of the main empirical research findings concerning governance differences between generations. These studies have primarily examined board composition characteristics on the basis of anecdotal or exploratory evidence.

In this study, we will present an integrated governance framework which builds on (1) previous research discussing generational changes in family attributes (e.g., Raskas 1998; Davis and Harveston 2001; Steier 2001), and (2) current evolutions in the governance literature, in which the importance of aspects beyond board composition—such as board roles and tasks—has been emphasized (Forbes and Milliken 1999; Gabrielsson and Winlund 2000; Huse 2005; Van den Heuvel et al. 2006).

Our first goal is to examine generational evolutions in board task needs. Boards can perform several tasks that may contribute to the organizational value-creation process, and these tasks can have either an external or an internal orientation. Externally oriented tasks concern the firm’s relationship with external constituencies, and include legitimizing the firm in society and providing access to crucial resources (Zahra and Pearce 1989). Internally oriented tasks concern the internal functioning of the firm, and include advising and controlling the managers (Sundaramurthy and Lewis 2003). In this study, we adopt an internal perspective and focus on the board’s advice and control task. When discussing these board tasks, governance scholars make a distinction between board task needs or expectations (Grundei and Tallaulicar 2002; Huse 2005) and board task performance [i.e., the degree to which boards actually succeed in fulfilling their tasks (Forbes and Milliken 1999)]. In this article, we examine board task needs because contextual factors such as the generational phase can be argued to primarily influence board task needs, before any other governance element such as board composition is affected (Huse 2005).

The second aim of this article is to scrutinize the influence of these board task needs on the relationship between the generational phase and board composition. As theoretical and empirical studies suggest that board task needs may have a significant influence on board composition (cf. Grundei and Talaulicar 2002; Huse 2005; Audretsch and Lehmann 2006; Voordeckers et al. 2007), we hypothesize and test the mediating effect of board task needs in the generation-board composition relationship. The presented framework will advance theory development in the domain of family firm governance and allows us to explain both diverging and converging generational trends in board characteristics.

The structure of the article is as follows. In the model development section, we first discuss three family attributes that have been demonstrated to change over the generations: task conflict, family experience, and intentional trust. This discussion forms the basis for our hypotheses development part, in which we analyze the impact of the family firm’s generational evolution on board task needs and board composition. Hereafter we clarify our research method, followed by a discussion of the empirical results. Finally, in the conclusion, we summarize our main findings and formulate directions for future research.

2 Model development

2.1 Generational changes in family attributes

2.1.1 Task conflict

A first argument for the contention that board characteristics in family firms might alter over the generations, relates to the level of task conflict. Task conflict refers to disagreements about the content of tasks, such as discussions about goals and strategies (Jehn 1995; Kellermanns and Eddleston 2004). Family firms have been argued to be a ripe context for conflict due to the overlap of the family and business systems (Sharma 2004).

A number of researchers have demonstrated that the level of task conflict among the relatives involved in the family firm increases over the generations (Raskas 1998; Davis and Harveston 1999, 2001; Ensley and Pearson 2005). This is due to the fact that the likelihood of diverging views concerning the firm increases as different generations and family branches get involved (Dyer 1994; Ward and Aronoff 1994). In addition, the number of passive family shareholders typically augments from one generation to the next (Jaffe and Lane 2004). These shareholders tend to prefer short-term dividend pay-outs whereas active family members mostly emphasize long-term performance (Vilaseca 2002; Schulze et al. 2003). As will be demonstrated in Sect. 2.2, this increase over the generations in the level of task conflict can be linked to the need for advice and control by the board of directors.

2.1.2 Family experience

A second argument for the claim that the generational evolution of family firms influences board characteristics is related to the level of family experience. The concept of family experience is derived from the F-PEC scale (Astrachan et al. 2002; Klein et al. 2005), which is an acknowledged scale for assessing the extent of family influence in a business organization (cf. Sharma 2004). Family experience is one of the subscales of the F-PEC scale, and refers to the tacit organizational knowledge that families develop over time (Astrachan et al. 2002).

With regard to the operationalization of the family experience subscale, Klein et al. (2005) have demonstrated that the generational phase of the family firm is its main indicator. Anecdotal evidence indicates that families do well at passing on tacit organizational knowledge from one generation to the next (Miller and Le Breton-Miller 2006). The older generation is typically very willing to share wisdom with the next generation of family managers, and to discuss their own mistakes with them (Miller and Le Breton-Miller 2006). Furthermore, each generation adds valuable business experience and skills to the family (Corbetta and Salvato 2004; Klein et al. 2005). In Sect. 2.2, we will link this generational increase in the level of family experience to the need for board advice.

2.1.3 Intentional trust

The third argument for the contention that board characteristics in family firms may be influenced by the generational evolution relates to changes in the level of intentional trust. Intentional trust can be defined as the willingness to be vulnerable to the actions of another individual based on a perception of integrity or altruism (Mayer et al. 1995). In the family business literature, intentional trust among relatives is often referred to as one of the main advantages of this organizational form (Tagiuri and Davis 1996; Habbershon and Williams 1999). The frequent social interactions in the family system provide the opportunity to build strong trusting relationships.

Yet several researchers found empirical evidence for a decrease of intentional trust over the generations (Raskas 1998; Steier 2001). This decrease is attributable to the fewer social interactions taking place among the relatives, which limits the opportunity to develop mutual intentional trust (Mayer et al. 1995). Simply put, family members in parental firms generally have closer relationships than those in sibling partnerships, and siblings mostly have stronger ties than the members of a cousin consortium (Gersick et al. 1997; Ensley and Pearson 2005). Seeing that the perception of altruism is one of the antecedents of intentional trust, this line of reasoning can also be linked to the discussion of Schulze and colleagues (Schulze et al. 2003; Lubatkin et al. 2005). They argue that relatives belonging to a later generation have incentives to attach a greater weight to the welfare of their own nuclear household than to the welfare of their extended family (Schulze et al. 2003). Moreover, many of the altruistic attributes which make family firms theoretically distinct are often completely lost during the cousin consortium generational stage (Lubatkin et al. 2005). In the next section, we will link this decrease over the generations in the overall level of intentional trust to the need for control by the board.

2.2 Board task needs and board composition over the generations

In the above section we have provided an overview of generational changes in three important family attributes—task conflict, family experience, and intentional trust—which have been identified by family business scholars. In this section, we build on the findings of these scholars and consider the possible resultant adaptations of the family firm’s governance system. More specifically, we examine the relationship between the generational phase of the firm, board task needs concerning advice and control, and board composition.

2.2.1 Need for board advice & outside directors

The board’s advice task relates to the resource-based view, which emphasizes that directors can be used as resources for the firm through their expertise and experience (Huse 2005). One of the main contributions of active boards in family firms is the provision of advice and counsel (Ward and Handy 1988). In this section, we will link generational changes in the level of task conflict and family experience to the need for board advice. Furthermore, seeing that it are mainly outside directors who can provide valuable advice (Ward and Handy 1988; Ward and Aronoff 1994), the generational phase can also be linked to the likelihood of having outside directors on the board.

As demonstrated by Jehn (1995), moderate levels of task conflict can enhance performance through the generation of new ideas and insights. However, when task conflict concerns strongly held preferences and consensus is difficult to attain, the functioning of the family firm may be paralyzed (Miller et al. 1998). In this case, arbitration by a third party may prove imperative for the sustainability of the organizational value-creation process. The board of directors, and especially the outsiders in it, are commonly attributed this role of arbitrator in family firms because of their objectivity and expertise (Nash 1988; Ward and Aronoff 1994). As the level of task conflict has been shown to increase over the generations, board task needs concerning advice and the likelihood of having outsiders on the board can be expected to increase.

In contrast, the rise over the generations in the level of family experience may have the opposite effect. As indicated by Huse (1990), boards can augment the expertise and know-how of the management team. This may be especially valuable for family firms, where management teams are typically small and dominated by a single decision-maker (Feltham et al. 2005). However, seeing that the level of family experience increases over the generations, this need for board advice can be expected to decrease. As the level of organizational knowledge develops over the generations (e.g., Astrachan et al. 2002; Miller and Le Breton-Miller 2006), the need for complementary know-how held by outside directors should decrease.

Therefore, in order to formulate hypotheses regarding the generational evolution of advice needs and the likelihood of having outsiders on the family firm board, a more detailed depiction of the generational changes in the level of task conflict and family experience is required. Anecdotal evidence indicates that the threat of task conflict is most prevalent in third generation family firms (Neubauer and Lank 1998). This is in line with the empirical results of Davis and Harveston (1999, 2001), who found only a moderate increase in the level of task conflict from the first to the second generation, and a more substantial increase from the second to the third generation. With regard to the level of family experience, family business scholars have argued that its increase is most substantial from the first to the second generation, since first generation firms typically build up a great amount of capabilities and rituals. Second and subsequent generations tend to contribute far less to this knowledge-development process (Astrachan et al. 2002; Klein et al. 2005).

The more detailed depiction of these generational changes suggests a significant decrease in the need for complementary outside know-how from the first to the second generations, which is not likely to be compensated by the increased need for outside arbitration. From the second to the third generation, however, a substantial increase in the need for outside arbitration takes place, which is not likely to be compensated by the further decreasing need for complementary outside know-how. These trends are summarized in Fig. 1. The suggested convex generational trend in the likelihood of having outside directors may explain why Ward and Handy (1988) and Schwartz and Barnes (1991) found that third and subsequent generations firms seem more likely to have outside boards, whilst Fiegener et al. (2000b) found no such difference between firms led by founders and firms led by successors (see Appendix).

Fig. 1
figure 1

Graphical representation of hypothesis 1

Full size image

Therefore, we propose the following:

Hypothesis 1a

Advice needs decrease from the first to the second generation, and increase from the second to the third generation.

Hypothesis 1b

The likelihood of having an outside director on the board decreases from the first to the second generation, and increases from the second to the third generation.

So far, we related the generational evolution to board task needs and board composition without discussing explicitly the causal chain between these variables. However, recent studies investigating boards (cf. Grundei and Talaulicar 2002; Huse 2005; Voordeckers et al. 2007) suggest that board task needs will mediate the relationship between the generational phase of the firm and board composition. Huse (2005), for example, argues that internal contingencies—such as the family generation and the family attributes discussed above—are a central element in the delineation of board task needs. Subsequently, board composition is likely to be adapted in line with these board task needs (Huse 2005; Voordeckers et al. 2007). Analogously, Grundei and Talaulicar (2002) argued that—even though external contingencies, such as legal obligations, may influence board composition—in reality companies follow mainly their governance needs and set up boards that reflect these needs. Therefore, we hypothesize that the generational phase of the firm influences the likelihood of having outsiders on the board via its effect on the need for board advice.

Hypothesis 1c

Advice needs mediate between the generational phase and the likelihood of having outside directors.

2.2.2 Need for board control & family directors

The need for control by the board is grounded in agency theory which assumes that (1) the interests of shareholders and managers often diverge, and that (2) managers seek to advance their own interests (Eisenhardt 1989). As will be discussed below, the generational changes in the level of task conflict can be related to the first agency assumption, and those in the level of intentional trust to the second. As such, the generational phase of the family firm can be linked to the need for board control.

In the previous section, we have argued that outside directors may play a vital role as arbitrator in the case of task conflict between family members. However, a family manager may not always open up the organizational processes to other relatives, and instead address strategic issues behind the scenes (Miller et al. 1998). Therefore, as the divergence of preferences increases over the generations, the need for control by the board can also be expected to increase in order to ensure that the management team considers the preferences and interests of all family owners.

Furthermore, perceptions of integrity or altruism are the possible antecedents of intentional trust (Mayer et al. 1995). Both integrity and altruism can be viewed as non-rational sources of a manager’s behavior, which reduce his/her propensity to behave opportunistically (Nooteboom 2002). As the level of intentional trust—and thus perceptions of integrity or altruism—decrease over the generations, family managers may be viewed as being exclusively concerned with the interests of their own nuclear household rather than the interests of the entire extended family (Raskas 1998; Lubatkin et al. 2005). Consequently, the need for board control can be expected to increase. Sibling partnerships often require the installation of formal governance mechanisms to control employed siblings, and this can be considered even more important for the cousin consortium generational stage (Steier 2001; Lubatkin et al. 2005).

Family firm boards are typically dominated by family directors (Westhead et al. 2002; Voordeckers et al. 2007). Seeing that each family branch is likely to require the presence of a fully trusted relative on the board in order to represent that branch’s interests when controlling the management team, we expect a rise over the generations in the presence of family directors (cf. Fiegener et al. 2000a; Van den Berghe and Carchon 2002; Cabrera-Suárez and Santana-Martín 2003). The discussion about the mediating effect of board task needs in the previous subsection also applies for the control task. Therefore, we expect that the generational phase will influence the number of family directors via its effect on the need for board control. As a result, we propose the following three hypotheses:

Hypothesis 2a

Control needs increase from one generation to the next.

Hypothesis 2b

The number of family directors increases from one generation to the next.

Hypothesis 2c

Control needs mediate between the generational phase and the number of family directors.

The hypotheses of this study are summarized in the framework presented in Fig. 2.

Fig. 2
figure 2

Framework of hypotheses

Full size image

3 Methodology

3.1 Sample

The empirical data presented in this article are derived from a study exploring a wide range of characteristics—strategic and environmental issues, management, governance, succession, and performance—in a representative sample of Belgian family businesses. The time frame for the survey was end 2002—beginning 2003. Based on the common selected criteria of ownership and management control (Chua et al. 1999) and the CEO’s perception (Westhead and Cowling 1998), the following businesses were included in the sample: (1) at least 50% of ownership and management is controlled by the family or (2) at least 50% of ownership is controlled by the family but less than half of the managers are part of the family, or (3) less than 50% of ownership is controlled by the family, at least 50% of the managers are part of the family, and the majority of the shares is owned by an investment company or a venture capital firm. In addition, in all the cases the CEO had to perceive the firm as a family firm.

In total, 3,400 limited liability firms (Naamloze Vennootschap) were randomly selected from a family business database and a survey was mailed to the CEOs.Footnote 1 After sending a reminder, 311 surveys were returned (9.1%) of which 295 firms could be classified as family firms according to our definition used. In 246 firms (83.4%) the family owned at least 50% of the shares and at least half of the management team was part of the family. In 41 cases (13.9%) the family owned at least half of the shares, but less than 50% of the managers were part of the family. In eight companies (2.7%) the family owned less than 50% of the shares, but at least half of the managers were part of the family. All companies were privately-owned enterprises employing at least five people, and were situated in the Flemish part of Belgium.

As CEOs have a major impact within management teams (e.g., Feltham et al. 2005), we also required the CEO to be part of the family. This assured us that the management team would be substantially influenced by family members, something we implicitly assumed in the hypotheses development, even when less than 50% of the managers were part of the family. This additional requirement reduced the sample by nine cases so that 286 family firms were included in the final sample. The total sample characteristics are presented in Table 1.

Table 1 Total sample characteristics
Full size table

Potential response bias was evaluated by two separate procedures. First, we contacted a random sample of non-respondents by telephone and requested data about their firm size, board size, board composition and meeting frequency. Comparing this data for non-respondents with the respondents in our sample suggests that the total population of Flemish family firms may be characterized by slightly more ghost or rubber stamp boards of directors (cf. Lane et al. 2006) in terms of board size, outside representation and frequency of board meetings than the firms in our sample. Secondly, following the argument that late respondents are expected to be similar to non-respondents (Kanuk and Berenson 1975), we differentiated between the 20% earliest respondents and the 20% latest respondents and conducted several t-tests and Chi-square tests on the variables included in the analyses. The results showed no significant differences on any of the variables, indicating that there is no significant response bias in the results. Robustness checks with cut-off points on 10% or 30% showed exactly the same result.

3.2 Measures

3.2.1 Intervening and dependent variables

As measure for board task needs, and in line with Van den Heuvel et al. (2006), the respondents were asked in two separate questions (one for each board task) to evaluate the importance of the control task and the importance of the advice task of the board on a Likert scale from 1 (not important) to 5 (very important). They were also asked to specify the number of family directors and outside directors on their board. A family director is defined as a member of the family who has a position as director in the firm. The variable presence of an outside director was 0 if no outside directors were on the board and 1 otherwise. As no consistency exists in the literature regarding the definition of an outside director (e.g., Ward and Handy 1988; Schwartz and Barnes 1991; Westhead et al. 2002), we applied an approach that seemed most appropriate in the context of this article. Therefore we define an outsider as someone who is neither a member of the family, nor part of the management team, nor affiliated to the company (such as an attorney or accountant). As such, the outside director can provide objective and impartial advice.

3.2.2 Independent variables

To determine the generation in charge of the company, the survey included a question in which the respondents had to indicate the generation currently having the decision power in the firm. We recoded this variable in three categories: first generation; second generation; third and subsequent (a.s.) generations.

3.2.3 Control variables

In this study, we include five control variables in the analysis. First, several authors claim that the economic development of the company may have a substantial impact on its governance practices (e.g., Fiegener et al. 2000b; Lynall et al. 2003). We therefore add the size of the company (logarithm of total assets) and the phase in the business life cycle (start, growth, maturity or consolidation) to the analyses. Since no starting firms are present in the sample, phase of the business life cycle was treated as a set of categorical dummy variables consisting of three phases. This enables us to distinguish governance changes exclusively due to generational issues from alterations due to the economic evolution of the business.

Secondly, since several studies reported that CEO power seems to be a main determinant of board characteristics in SMEs (Fiegerer et al. 2000b; Voordeckers et al. 2007), we include CEO duality as a control variable. Moreover, results concerning the relationship between generation and board tasks or board composition may be driven by the possible increasing number of employed family members (Voordeckers et al. 2007) or the number of family shareholders which could be considered as a proxy for ownership dispersion. Hence, we also include these variables as control variables.

3.3 Method

To test the various hypotheses concerning the direct effects, we initially apply the most obvious and straightforward statistical analyses. Depending on the question under consideration, we used an analysis of variance (ANOVA) or a Chi-square test. With regards to the ANOVAs, we employ the F-test to determine the overall significance level, and the commonly used Tukey procedure for the pairwise comparisons. In the case of heteroscedasticity problems, we utilize the Welch-test and the Games-Howell procedure instead.

To test for mediation effects, we followed the procedure proposed by Baron and Kenny (1986). These authors propose to estimate three regression equations: (1) regressing the mediator on the independent variable, (2) regressing the dependent variable on the independent variable, and (3) regressing the dependent variable on both the independent variable and the mediator. We also regress the dependent variable on the mediator as this provides information for differentiating between mediation and direct effects (Mathieu and Taylor 2006).

4 Results and discussion

Hypothesis 1a relates to the need for board advice. Both the ANOVA (Table 2, part a) and the regression analysis (Table 4, model 1) demonstrate that the trend in this board task need is as hypothesized, although only significant at the 10% level. The advice needs decline from the first to the second generation, and increase again in third and subsequent generation firms. A statistically significant difference exists between the second generation and both the first and third or later generations. These results support hypothesis 1a.

Table 2 Results board task needs
Full size table

Tables 3 (part a) and 4 (model 2) present the results concerning the likelihood of having outside directors on the board (hypothesis 1b). The empirical results show that this likelihood follows a similar convex trend over the generations as the need for board advice. The second generation has a lower likelihood of having outside directors than the first and third or later generations, and this difference is statistically significant. This supports hypothesis 1b.

Table 3 Results board composition
Full size table
Table 4 Regression results of the hypothesized mediation effect on outside directors
Full size table

Our empirical results thus indicate a convex relationship between the generational phase and the governance elements under study. These hypotheses were the result of previous research findings concerning generational changes in two family attributes, task conflict and family experience, with opposing effects on the need for advice and the likelihood of having outside directors. The rise in the level of task conflict over the generations (e.g., Raskas 1998) was argued to augment the need for board advice and thus outside directors, whilst the rise in the level of family experience (e.g., Klein et al. 2005) was argued to have the opposite effect.

The convex trend in our results suggests—as expected and in line with the literature—that the increase in family experience is the main determinant of governance changes from the first to the second generation, and that the level of task conflict the main driver of governance changes from the second to the third generation. Moreover, our results also shed light on prior seemingly contradictory and puzzling evidence concerning the likelihood of having outside directors in family firm boards over the different generations (cf. Schwartz and Barnes 1991; Fiegener et al. 2000b; Voordeckers et al. 2007).

In order to test the mediation effect of the need for board advice (hypothesis 1c), we estimated six regression models which helped us to examine the formulated hypothesis (Table 4, models 1–4) and to verify whether the addition of control variables significantly changes the results (models 5 and 6). First, in model 1 of Table 4, we regressed the advice needs on generation. Secondly, in model 2 we regressed the likelihood of having outside directors on generation. As indicated above, in both regressions generation is statistically significant. Following the procedure of Baron and Kenny (1986), we then estimated a third regression (model 4) in which the likelihood of having outside directors is regressed on both the need for board advice (the mediator) and the generational phase. In this regression, the advice needs have a significant effect on the likelihood of outside directors while the generational phase is no longer statistically significant.

This set of results indicates that the need for board advice is a fully mediating variable for the relationship between generation and the likelihood of having outside directors on the board: when we include advice needs as independent variable, the effect of generation on the likelihood of having outside directors disappears. Model 3 already indicated that the need for board advice has a significant positive effect on the likelihood of having outside directors. Hence, our tests extend current knowledge about the relationship between the generational phase and the presence of outside directors as it illuminates the precise nature of the relationship and the causal chain. The hypothesized generational effects seem to affect the likelihood of having outside directors fully through their influence on the need for board advice.

The addition of the control variables in model 5 and 6 (Table 4) does not change the results of the mediating analysis. In these additional regressions, only two control variables were found to be significant. In line with the majority of empirical evidence in the field, CEO duality has a negative effect on the likelihood of having outside directors on the board. Furthermore and as expected, a higher number of working family members also decreases this likelihood.

Table 2 (part b) contains the analyses concerning the generational evolution in the need for board control (hypothesis 2a). We observe a rise in the need for the control task, with the largest increase occurring from the first to the second generation. The differences between the generations are, however, not statistically significant in the ANOVA set-up. The regression results (Table 5, model 1) confirm the mounting trend with only the difference between the first and the third or later generations being significant. Hypothesis 2a is thus only partially supported.

Table 5 Regression results of the hypothesized mediation effect on family directors
Full size table

The average number of family members on the board of directors (hypothesis 2b) also increases over the generations (Table 3, part b). The ANOVA test indicates that the impact of the generation-factor is significant at the 0.01 level. The Tukey pairwise comparisons indicate that the difference between the first and the second generation is significant at the 0.10 level, and the difference between the first and the third or later generations at the 0.01 level. However, the disparity between the second and the third or later generations is not found to be significant. Regression model 2 in Table 5 confirms the ANOVA results. Hence, hypothesis 2b is also partially supported.

The mediation effect of the need for control (hypothesis 2c) was tested by estimating the same set of six regression models (Table 5) as for the need for advice. As mentioned, the results of the first two regressions confirm the ANOVA results. In model 1, we estimated the effect of generation on the control needs. We find a positive effect, but only the difference between the first and the third or later generations is statistically significant. The results of model 2 show that generation also has a significant positive effect on the number of family directors. The nature of the relationship is revealed in model 3. Since the regression coefficient in this model is not significant, our tests suggest a direct effect of the generational phase on the number of family directors on the board (Mathieu and Taylor 2006). Consequently hypothesis 2c, postulating that the need for board control mediates between generation and the number of family directors, was not supported by our regression results.

Although the direct effect of generation on the number of family directors is puzzling at first sight, it might suggest that other needs or desires than the need for board control mediate this relationship. A potentially important board task that we have not included in our study is the board’s active involvement in the strategic decision-making process (cf. Pearce and Zahra 1992). As discussed above, previous research has demonstrated that the degree of task conflict tends to increase over the generations. Therefore, one might expect that over the generations more family members will demand a seat on the board in order to present their personal or the nuclear family’s strategic views of the family business during board discussions. This would also explain the significance of the control variables ‘number of family shareholders’ and ‘number of working family members’ in model 5 and 6. In brief, as more generations, family branches, and family members get involved in the company, cognitive divergence concerning strategic issues is likely to increase (cf. Ensley and Pearson 2005), and consequently more family members may want to have a say on major strategic issues in board discussions. Of course this line of reasoning will need to be tested in future research since we were unable to test it with our data. However, based on the present data, we can conclude that other needs or desires than the need for control underlie the generational increase in the number of family directors.

Our analyses indicate that the rise in the need for board control over the generations is not reflected in the number of family directors. However, these increased control needs might influence the likelihood of having outsiders on the board. Family members may rely primarily on outside directors with the required functional skills and ‘independence of mind’ (cf. Forbes and Milliken 1999; Roberts et al. 2005) to perform the board’s control task. In order to test this alternative hypothesis, we estimated a third set of regressions (not reported) in which we examined the control needs as possible mediator between the generational phase and the likelihood of having outside directors. Both generation and control needs seem to have a significant effect on this likelihood. When estimated together as independent variables, however, neither the coefficients nor the significance levels changed. Moreover, the convex generational trend found in the regression models with advice needs also shows up in these regressions. This set of results suggests that generation influences the likelihood of having outside directors primarily through changes in advice needs, and that once on the board these outsiders may also fulfill the need to control the management team.

As indicated, we measured board task needs by asking respondents to evaluate the importance of the board’s advice and control task in two separate single-item questions. No other direct measures for these board task needs were available. In order to test the robustness of our findings, we retested our hypotheses with an alternative proxy for board task needs that is closely related to our direct measures. This alternative proxy is the percentage of time the board spends on a specific board task. We propose that a board which experiences a changing need for a specific board task will first spend less or more time on that task, before this changed need is reflected in board composition (cf. Huse 2005). Spending more time on a specific board task can be executed immediately without significant constraints, whereas board composition can be expected to be less flexible. Therefore, we propose that the percentage of time spent on the advice and control task by the board (evaluated in two separate questions) is an acceptable indirect proxy for board task needs to test the robustness of our results. This retest of our statistical models showed exactly the same patterns as with our direct measures (importance of the two tasks), although the statistical significance of the effects was lower.

5 Conclusion

The purpose of this study was to scrutinize the relationship between an important dimension of a family firm’s life cycle, namely the generation in charge of the company, and board characteristics, namely two main board task needs (advice and control) and its composition. Moreover, we tested the mediating effect of board task needs in the generation-board composition relationship. This study not only contributes to the literature on governance in family firms, but also adds to the recent literature examining behavioral perspectives of corporate governance (Forbes and Milliken 1999; Huse 2005). Board task needs are theoretically distinct from board task performance, and the majority of prior board role research has concentrated on the latter. Nevertheless, these underresearched board task needs take a central position in models about understanding actual board behavior (Huse 2005).

Our results suggest that the advice needs decrease from the first generation to the second, and rise again in third and subsequent generation firms. An analogical trend is found for the likelihood of having outsiders on the board. The effect of the generational phase on the likelihood of having outside directors seems to be fully mediated by the need for board advice. This convex trend can be explained by two underlying opposing generational evolutions, namely the increase in task conflict among the family members and the rise in family experience (e.g., Davis and Harveston 1999; 2001; Klein et al. 2005).

Further, our results suggest a mounting trend in the need for board control and in the number of family directors over the generations. This evolution in the need for control may be explained by the fact that further generation companies are, on the whole, characterized by higher levels of task conflict and less intentional trust (e.g., Raskas 1998). Surprisingly, the relationship between generation and number of family directors does not seem to be mediated by the need for board control. More specifically, the results suggest a direct relationship between generation and number of family directors. This may be interpreted as an indication that family members will require a direct representation on the board based on other desires than the need to control the management team. One such desire may be to become actively involved in major strategic decisions in order to express one’s ‘diverging’ opinion. However, future research will need to examine this possibility. A further analysis of how the control needs are reflected in board composition revealed that outside directors are adopted on the board because of both advice and control needs. Yet the primary board task needs mediating the relationship between generation and the inclusion of outside directors seem to be those related to the advice task.

This study has some limitations that should be acknowledged and addressed in future research. First, we discussed generational changes in various family attributes and integrated them to derive implications for board characteristics. However, future research should incorporate direct measurements of these family attributes in the empirical models in order to get a clearer understanding of the precise impact of each of these different attributes. Secondly, the proxies for board task needs were measured by two separate single-item perspective measures. Although our results remain robust with an alternative proxy for these variables, future research may develop more detailed and multiple-item perspective measures for board task needs.

Furthermore, several challenges remain for future research. First, we concentrated on two important board tasks, namely advice and control. Nevertheless, other board tasks have been described in the literature, such as management of external dependencies and involvement in strategic decisions (Pearce and Zahra 1992; Hillman et al. 2000). Secondly, in this study we concentrated on generational differences in board task needs and board composition. A next step would be to explore generational differences in actual board task performance. However, with regard to this dependent variable a whole range of intervening process variables will need to be taken into consideration such as, for example, board cohesiveness and effort norms (Forbes and Milliken 1999; Huse 2005).

Thirdly, we examined generational differences in board characteristics by asking the respondents to indicate the generation currently having the decision power in the firm. This approach indicates the most dominant generation involved in the running of the firm. However, different hybrid forms may exist, with varying degrees of involvement by several generations. Scholars may therefore refine our model by taking a more continuous approach and examining the level of involvement of different generations in ownership, board of directors, and management.

Lastly, the family firm’s life cycle should not be viewed as a static phenomenon. As mentioned by Filatotchev and Wright (2005), this life cycle takes place “within the life cycle of the market environment in which the firm is operating and, more broadly, the institutional life cycle of the country in which the firm is located” (Filatotchev and Wright 2005, p. 9). Future studies may therefore examine to what extent governance reforms, and more specifically the recent development of governance codes specifically for non-quoted companies (e.g., Code Buysse in Belgium) influence the generational changes in the characteristics of family firm boards.