1 Introduction

This paper examines the peculiarities of corporate governance found in Italian family firms, and the connection between governance and performance.

As already introduced by Caselli and Gatti (2006) or Mazzola and Marchisio (2003), this empirical evidence is aimed to continue the analysis of Italian family firms and, in particular, tries to underline the typical peculiarities of governance structure and the connections among performance and governance, rather than to discover which variables determines or explains the governance structure adopted by firms.

This work contributes to the development of a contingency model that postulates a connection between ownership variables and family firm performance, while taking into account the composition of a firm’s governing structure. Family controlled firms, in the present study, are defined by two variables: the level of ownership, and the level of participation in management. Regarding the former, family-controlled firms are defined as those in which the founding family (also in terms of family relationship between partners) holds a sizable number of shares. The run analyses concern the relationships between family ownership and board composition, the correlation between performance and outside directors, and the connection between family ownership and whole share performance.

The evidence developed is especially interesting for purposes of contingency theory validation, and contributes to the discussion on the need to develop corporate governance systems for Italian business entities, both private and institutional.

This article is organized as follows: Sect. 2 reviews the most salient findings about the relations between performance and corporate governance, ownership and firm management. Section 3 introduces the hypotheses investigated in this study and Sect. 4 describes how data have been gathered and details the methodology of analysis. Section 5 presents the most interesting findings, while Sect. 6 concludes with comments and remarks.

2 Theoretical framework

Any study focused on corporate governance (both for family and non-family firms) must take into consideration the findings reached initially by Jensen and Meckling (1976) and Ross (1973) that underscore the inherent potential conflict between management and shareholders. This is “agency theory,” which assumes that factors allowing shareholders to monitor management or to align management’s interests with their own will affect firm performance in the long run. This means that where ownership is dispersed, corporate governance measures have a positive relation with performance, all other conditions being equal.

At the same time, there is an alternative to agency theory, “stewardship theory,” which suggests that it is appropriate that there be a prevalence of insider over outsider actors. This theory argues that managers are more reliable and are less liable to waste resources, since they act in the firm’s interests and therefore aim to achieve higher profit and greater return to shareholders. This theory is regarded as directly opposite to agency theory, as its main tenet is that control should be centralized in the hands of management (Davis et al. 1997). In fact, many researchers (e.g., Kesner 1987 or Vance 1978) find that the presence of insider actors is indeed correlated with superior firm performance.

Historically, the debate on corporate governance developed and evolved first in the US, then in the UK, but such debate has always been framed around an absolute definition of the composition of the governing structure. Only later (Davis 2002; Roberts et al. 2005) did studies emerge on a broader range of governance systems that focused on comparative analysis among a more diverse group of countries and economies. More specifically, these broader analyses have led to the development of alternative perspectives for understanding relationships among governance, firm ownership and performance (in particular, IPO performance). Consequently, in clear contrast with the two dominant theories, which postulate the existence of a standardized approach to governance structures, a new and more composite research stream has developed that holds that there is no best system of governance; on the contrary, there are better systems than others or better rules to follow than others. This approach is known as “contingency theory,” which holds that the deficiencies of the dominant theories stem from the fact that they do not simultaneously consider firm characteristics and corporate governance. For Italian firms, Caselli and Gatti (2007) verify differing agency contexts among family firms, and that corporate governance mechanisms must be evaluated from both the cost and benefit perspectives, since these can determine a firm’s final performance in different ways.

Thus, in order to define a board of directors or a system of governance, two elements must be taken into consideration: actors and context.

By actors we mean not only directors, but the entire range of stakeholders representing interests and power within the firm. Their presence guarantees, on the one hand, the creation of value and the distribution of the value created; on the other, it determines contextual factors, beginning with governance mechanisms.

By context, we mean geographical, cultural, sectoral, and firm-specific differences and variations. Among these are the degree of dispersion in and the nature of firm ownership, differences in size, lifecycle variations, including crises and the configuration of firm resources, and CEO tenure and its features and background. For purposes of this empirical evidence, a firm’s ownership structure and all variables related to CEO characteristics and/or outside/independent director attributes are more important than all other variables.

Ownership structure refers to type of ownership and to ownership variation or dispersion (Pederson and Thomsen 2003); each of these elements varies considerably by country and by sector. Generally, there are five types of firm ownership: institutional, which usually is reluctant to have direct representation on the board, and individual, corporate, state, and family, which usually do take board representation (Gedaljovik and Shapiro 1998). Consequently, the role of a governing body differs with each type of ownership structure in accordance with the type of involvement that satisfies the needs expressed by the ownership. In particular, family firms show a concentrated and enduring ownership structure (Gabriellson and Huse 2002); this is particularly clear in Italy, where firms routinely go public with high levels of family control. Such family controlled firms value control above all and are less inclined to give up the majority of votes. Moreover, when a controlling stake is relinquished, it is typically sold to another blockholder on a “friendly basis” (Rigamonti 2007).

The board’s roles also vary in relation to characteristics of the CEO, including tenure, ownership, ownership stake, and competence. We may assume that a CEO’s power tends to increase over time, but a new CEO, according to the stewardship theory, creates a mentoring problem, as he is not well enough acquainted with the business. Conversely, a more expert CEO creates a monitoring problem (Berle and Means 1932; Fama and Jensen 1983).

Apart from CEO characteristics, governance may be examined in light of the relationship between independent/outside directors and performance. Leblanc and Gillies (2003) re-examine existing findings and argue that a rather non-homogeneous classification continues to exist because what are considered “external” directors are defined in various ways (for instance, unrelated, independent, outsider), depending on the degree of independence conceived by the regulator drafting the regulations or by parties involved in a contract. The Harvard Law Review (Harvard University 2006) proposes an interesting interpretation of independent directors and board independence topics. In the authors’ opinion, regulators, commentators and courts have all used “independence” to mean different things at different times for different reasons. Through examination of underlying policies, these definitions are found to fall into three broad categories: the independent director as a “disinterested outsider,” the independent director as an “objective monitor,” and the independent director as an “unaffiliated professional.” In particular, existing empirical studies describe different roles and types of independent directors, partly because of differing variables that the various authors consider as the basis for analysis in each study.

Research on corporate boards has been dominated by a tradition in which board composition is viewed in relation to firm financial performance (e.g., Johnson et al. 1996; Zahra and Pearce 1989). The main research stream has been especially influenced by financial analysis, and by a financial research tradition that treats the board as a black box. The various roles that a board can assume are rarely taken into consideration. Johnson et al. (2000) conclude that it is difficult to find a relation between board composition and financial performance, even if it may be affirmed that, generally, a negative correlation exists between the number of members on a board and performance. Nevertheless, Bhagat and Black (2002), starting from the assumption that a “monitoring board” composed almost entirely of independent members is an important component of good corporate governance, analyze a wide sample of US firms and find no supporting evidence for this conventional wisdom. In particular, firms with more independent directors do not perform better than other firms over the long term. Belkhir (2004) investigates the relationship between board size and performance in a sample of banks and savings-and-loan holding companies. Contrary to theories predicting that smaller boards of directors are more effective, the author finds that increasing the number of directors in banking firms does not undermine performance.

In the US, family ownership is commonly seen as less profitable or less efficient compared with a dispersed ownership structure (public companies). Fama and Jensen (1983) note that the combination of ownership and control allows the taking of profits for private income. Demsetz (1983) maintains that some owners may choose non-pecuniary consumption and take resources away from profitable projects. Shleifer and Vishny (1997) observe how greater shareholder ownership and control clearly facilitate the elimination of private benefits from a firm. Furthermore, families often restrict managerial positions to family members, thus reducing access to the most talented and capable people; this is a potential competitive disadvantage. Indeed, these findings are generally supported by evidence pointing to a negative relation between family ownership and performance. However, this point of view is not universally held; for example, Demsetz and Lehn (1985) note that the combination of ownership and control can be an advantage where there are many shareholders to mitigate expropriation, while Stein (1988, 1989) shows that the presence of shareholders with long-term investment horizons can mitigate the theory on decision making by managers.

The relationship between performance and ownership structure is particularly interesting when venture capitalists are firm shareholders. Baker and Gompers (2001) stress the role of venture capitalists in the formation of boards of directors. These authors find that venture capitalists provide financial instruments and inside board members; this suggests that they occupy some of the main roles and provide certain value-added services that other actors, such as consultancies, banks, and other independent board members are not capable of providing. Similar findings are described by Hermalin and Weisbach (1998, 2003): venture capitalists select family members as CEOs only if firm reputation increases, otherwise the executive selection process is independent of family aspirations. So, it may be affirmed that venture capitalists play an instrumental role within the board of directors, in addition to their monitoring and control functions, thereby increasing firm profits (Caselli and Gatti 2006; Jaskiewicz et al. 2005).

Considering more specifically family firms, IPO performance and governance, we may affirm that, from an empirical standpoint, the relationship between IPOs and investment performance in the sphere of family owned firms is quite unique. Indeed, until now, investigations have brought inconsistencies to light. On one hand, in fact, some test results show significant underperformance in IPOs of family businesses as compared to non-family firms or to other benchmark indices (Aussenegg 1997 or Ehrhardt and Novack 2003). On the other hand, a vast body of research attests to better long-run performance for listed family firms with respect to non-family firms (e.g., Anderson and Reeb 2003; Villalonga and Amit 2004).

Dyer (2006) offers a deep review of the literature and several ways to explain the contradictory evidence regarding the performance of family owned firms, while Caselli and Gatti (2006), reporting the results of several empirical studies conducted on family firm IPO performance in Europe and the US, conclude that there are no proper conclusions to reach as the topic has been analyzed much too little. Nevertheless, all these findings require deeper analysis aimed toward understanding the instruments used by families to manage, monitor and maintain control over their firms, that is, their governance methodologies. In family firms, very often a family member holds the office of CEO or other top management positions. This entails two consequences. The first is closer alignment of family and business interests, which acts as a multiplier on the effects of ownership on performance. The second is that family members holding the position of CEO through entitlement must sustain the cost of excluding more capable and talented outsiders. Gomez Mejia (2001) further develops the latter consequence by adding the issue of CEO accountability to shareholders and directors. Schultze et al. (2002) note that it is possible to create a form of resentment on the part of non-family executives when merit and commitment are not essential requirements to the achievement of top management positions. However, a family CEO may have specific capabilities and distinctive traits that an outsider cannot have. Mork et al. (2000) refer to superior capacity for innovation and increased use of one’s own capabilities, while Davis et al. (1997) speak of a service attitude held by family members, who strongly identify with the firm and who consider such identification a reason for its success. Anderson and Reeb (2003) maintain that the effects of family ownership on business culture are particularly positive and lead to improved performance, as they generate strong incentives for management.

3 Testable hypotheses

To improve understanding of the IPO performance of family firms with reference to corporate governance, four hypotheses are set forth.

The first hypothesis considers the relationships between firm ownership and board dependence, that is, board composition and/or the presence of insider, outsider/independent, and family directors. Theory holds that if a board is to be effective, it must have outside directors; this necessity conflicts with the presence of strong family ownership within a body of shareholders.

As a substantial proportion of the universe of Italian family firms that reach the market is made up of venture backed firms (Table 4), it must be considered whether the presence of such firms could induce significant change in board composition. Considering the F-PEC data for Italian family firms, it may be supposed that venture capitalists or private equity operators reduce family board dependence. However, the significance of this ability is difficult to define for many reasons: the level of involvement and commitment of venture capitalists, the options available to venture capitalists to enforce their role in management (e.g., covenants), the different methods and/or typologies of venture capital (e.g., closed end funds, banks, other financial institutions, etc.). At the same time, board composition and family participation seem to be independent of governance structure, even in board committees that comply with the code issued by the Milan Stock Exchange.

Accordingly, the first hypothesis is as follows:

Hypothesis 1

Family participation increases the level of board dependence, independent of the presence of venture capitalists among shareholders and of the adoption of Italian code of governance criteria.

The second hypothesis stems from the observation that the participation of outside directors should positively correlate with firm performance. So:

Hypothesis 2

A board chiefly composed of outside directors ensures superior family firm performance.

The same approach can be used with regard to the presence of committees within the board. In particular, the number of committees, as an approximation of board capital, turns out to be positively correlated with performance. The third hypothesis is therefore as follows:

Hypothesis 3

The presence of committees within the board is positively correlated with family firm performance.

Lastly, the fourth hypothesis serves as the overall connecting link for this study:

Hypothesis 4

Family involvement is negatively correlated with performance: concentration of power in the hands of family members generates inferior performance.

4 Sample, data collection and methodology

The sample was determined through combination of several defining criteria based on the aims of this survey. Further, the sample is built from the universe of possible observations, constitutes a subset on which significant studies can be conducted, and whose characteristics are consistent with analysis variables.

As in the empirical study of Caselli and Gatti (2006), the initial defining stage delimits an interval of time encompassing the cases under scrutiny, all companies listed on the Milan Stock Exchange between 1999 and 2005 (136 firms). Given the overall intention to study the composition of the main governing bodies of businesses, the extent of family participation, their performance, and the possible presence of venture capitalists within the body of shareholders, the initial sample is screened according to two additional criteria: first, the elimination of banks, insurance companies and financial companies, in order to exclude firms whose governance is chiefly defined by legal criteria, and second, the elimination of municipal and privatized, formerly government owned enterprises (for which no family ownership could easily be supposed) and foreign companies (as the purpose is to study the behavior of Italian companies).

The most interesting phase, however, and the most significant for purposes of sampling, is the choice of a criterion for distinguishing “family” and “non-family” businesses; the F-PEC score is selected to make the distinction. This method, which produces a better understanding of the establishment of a firm’s family nature, was recently formulated by Astrachan et al. (2002); Caselli and Gatti (2006) and Jaskiewicz et al. (2005) have already applied it in their empirical studies. A family business (FB) may be defined as follows:

$$ {\text{FB}} \ge 0. 5, $$

where FB is the F-PEC score, which is defined by the following procedure:

$$ {\text{FB}} = ({\text{EQfam}}/{\text{EQtot}}) + ({\text{BoDfam}}/{\text{BoDtot}}) + ({\text{SBfam}}/{\text{SBtot}}) $$

The first addend of the FB formulation defines the equity share owned by the family (EQfam) over total equity (EQtot); the second defines the percentage of family members or members interconnected with the family on the board of directors (BoDfam) on the total (BoDtot); and the third addend represents the percentage of family members or members interconnected with the family (SBfam) on the whole supervisory board (SBtot).

The most important advantage of the F-PEC score is its ability to group into a single index a series of contingent dimensions suggested in the literature that constitute significant variables for the purpose of defining firm performance and behavior. This wider consideration of family status is supported by the definition of FB provided in Corbetta and Tomaselli (1996): “Family businesses are businesses where one or a few families connected by a mutual relation, close ties or a solid alliance hold a share of equity which is sufficient to ensure control over the company or to make the main strategic management decisions.”

The F-PEC score is calculated for each company that reached the Milan Stock Exchange in the mentioned period, in order to identify those that can be believed to be FB. Moreover, the F-PEC score is used to classify companies into subgroups. An F-PEC score from 0 to 0.5 indicates a non-family firm and, for this reason, firms with scores in this range are excluded. Firms with F-PEC scores from 0.5 to 1 are classified as weak family owned businesses, those with F-PEC scores from 1 to 1.5 may be considered normal family owned businesses, while firms with F-PEC scores >1.5 are strong family owned businesses.

Returning to the sampling procedure, it is emphasized that the data used for selection is drawn mainly from prospectuses, Italian Stock Exchange databases, Consob (the Italian supervisory authority for securities markets), Datastream, Worldscope, Aida and Zephir, along with various websites that furnish financial information.

Some 68 companies meet the FB criteria. However, data necessary for classification were not available for all firms; therefore, a portion (12) of this initial set is excluded. Many of the excluded companies are being wound up, are in extraordinary administration, or have actually gone bankrupt. Of the remaining 56 firms, the cluster distribution according to F-PEC yields 37 weak family owned, 16 normal family owned, and 3 strong family owned companies.

As shown in Table 1, in overall terms, it is apparent that the F-PEC mean value is 0.91, with variance equal to 0.30. The firms in the sample have, on average, considerable family involvement, especially considering that deviation from the mean value is minimal.

Table 1 F-PEC distribution
Full size table

To further define the features of this set of firms, a temporal distribution of IPOs is first defined. A very high concentration of listings is noted in 2000; some 27 companies in our sample, about 50% of the total, entered the stock exchange that year. Nevertheless, analysis of correlation between year of listing and performance shows that these two variables are not correlated, so it may be affirmed that year of listing and IPO date do not affect final results. Della Bina and Cervellati (2005), studying a sample of Italian IPOs realized between 1999 and 2001, note that IPOs and overall market trends are not similar, whatever happens the first day of trading (Table 2).

Table 2 IPO year distribution
Full size table

Inside the sample, there is broad representation of industry sectors and, for this reason, a sector bias should not be present in this empirical evidence. In particular, mechanical and electromechanical sectors are most heavily represented. It must be highlighted that in the sample, there are family firms in the high-tech or high growth sectors, such as media, telecommunications (TLC), software houses, etc., but since their share is not greater than the one of other sector less “high-growth” or “high-tech” (e.g., textile or food and beverages), any additional analyses directed toward consideration of statistical differences among sectors is redundant and not strictly necessary. In practice, according to the objectives of this empirical research, differences among firms are attributable to internal decisions and not to market/sector structure or to any other such variables, which are not controllable by the firms directly (Table 3).

Table 3 Breakdown by sector of firms in the sample
Full size table

Similar conclusions may be drawn concerning the size of firms at IPO. This situation confirms the findings of Arosio et al. (2001) that in Italy, from 1985 to 1999, all IPOs were similar and that any differences must be attributed to the privatization process. Caselli and Gatti (2007) is also vindicated in that, for the period 2001–2005, it is affirmed that differences among family firms are not statistically significant and that they cannot be caused by items related to corporate governance systems (Table 4).

Table 4 Breakdown by size and age at IPO
Full size table

The sample average age and the sample age distribution, on the contrary, are quite different from the typical situation: on average, Italian firms are founded 36 years before going public (Rigamonti 2007), but in our sample the average is about half that and is more similar to that found by Caselli (2009)—this is due to the sample selection process. Nevertheless, after a statistical test that we do not report here, it may be affirmed that there is no diversity among firms with different ages.

As a result, the entire sample is analyzed using a multivariate regression methodology. The primary interest is in investigating the relation between firm performance, corporate governance variables, and variables pertaining to family ownership, while also bearing in mind the possible involvement of the firm’s body of shareholders at the time of an IPO by a venture capital company. A step-by-step analysis is conducted for this purpose. First, a link between ownership and corporate governance features is drawn. Then, a correlation between corporate governance variables and performance is established. The analysis also incorporates variables that identify the CEO as a descendant founder, a family member, or an outsider (independent).

Two dependent variables are used in the regression models. The first variable, indicated as “mid board,” goes beyond the traditional distinction between an insider board and an outsider board by including a family board among the possible options. The argument on which this classification is based rests on the observation that firm managerialization develops less slowly than the adoption of outside directors; both a mixed board and an outsider board can fill the need for strategic guidance and support. Thus, the insider board is a transitional category between a family board and an outsider board; it is also a first step toward greater professionalization of corporate governance. The existence within the sample of the insider board is taken as the first dependent variable and is explained by independent variables pertaining to the influence of family and/or venture capitalists.

Firstly, the ratio between the number of independent and executive directors and the number of executive directors related to the owner family (Pout/Pfam) is calculated and, secondly, in order to avoid all biases due to size, the LN is employed as a dependent variable in the regression model. In particular, a director is considered an “outsider” if he can be believed to be “non executive and non familiar” according to the definitions set forth in the Italian code of governance issued by the Milan Stock Exchange.

The second dependent variable is long-term firm performance. This is calculated using the T period buy-and-hold abnormal returns (indicated as “BHAR”), which can be taken to be the simplest and most intuitive measure of the long-term performance of IPOs, and which is equivalent to an investor’s return differential for the purchase of shares on the first day of an IPO, and to those shares’ resale value at the end of the period, which is T months in length. In the empirical evidence, in order to reduce effects stemming from overall market trends and to broaden the explanatory capabilities of the model, time horizon T is set at 12 months. A positive buy-and-hold abnormal return must be interpreted as a better performance compared to the IPO with respect to the benchmark. In our case, the benchmark is the historical Mib (Milan Stock Exchange Index).

As in Caselli and Gatti (2006), BHAR is calculated as follows:

$$ {\text{BHAR}} = \frac{1}{N}\sum\limits_{i = 1}^{N} {\left[ {\left( {\prod\limits_{{t = t_{i} }}^{\eta } {\left( {1 + {\text{Rit}}} \right) - 1} } \right) - \left( {\prod\limits_{{t = t_{i} }}^{\eta } {\left( {1 + {\text{Ret}}} \right) - 1} } \right)} \right]} , $$

where Rit is the return on stock i in month t, N is the number of IPOs (56), T is the number of months (12), t i is the date of the closing price on the first day of trading and E(Ret) is the benchmark.

Even if it is obvious, it must be emphasized that this approach considers all IPOs equivalent, that is, each holds the same weight. In other words, in this empirical evidence, an equally weighted system is adopted that assumes, at the portfolio level, an equal investment in all stocks regardless of the size of the operation or of the company. The most obvious outcome of this method of calculation is that the weight of small issuances is much greater than it would be if criteria capable of weighting each issuance for its actual market value were used.

Considering the independent variables that are to be applied to the regression models, two types of parameters are used: corporate governance and corporate ownership.

The first type, as it relates to the governance system, pertains to board dependence (outsider, insider, family, independent directors), board capital (presence of committees and size of the board) and, finally, CEO power. The second type, on the contrary, relates to the body of shareholders, such as the proportion of family shares, the presence of venture capitalists, and the shares attributable to the CEO.

Given the empirical evidence and level of homogeneity with regard to the concept and definition of independence, this paper uses the definitions given in the Italian Code of Governance issued by the Milan Stock Exchange (Borsa Italiana); this code is used by listed companies to organize and define criteria for corporate governance. According to article 3.C.1 of the Italian Code of Governance, boards of directors should evaluate the independence basing the estimation “primarily on substance as opposed to form,” defining cases in which an individual can never be considered independent. Moreover, the same set of recommendations requires a thorough analysis concerning the presence and role of committees and is aimed at ensuring fair firm management.

In particular, with respect to the empirical evidence in this study, the following definitions are applicable to the entire study:

  • Insider director: an executive director who is a member of the board;

  • Outsider director: a non-executive director who is a member of the board;

  • Family director: a director appointed or selected by the family and who is a member of the board;

  • Independent director: a director not appointed or selected by the family and who is a member of the board;

  • Appointment committee, compensation committee, internal control committee, executive committee: the teams and commissions provided for by the Italian Code of Governance;

  • Directors: the total number of directors belonging to the board;

  • Family holding company: presence of a company held by an owner family;

  • Venture capitalist (or VC): presence of venture capitalist in whatever model or type. Nevertheless, this kind of institution must be member of AIFI, the Italian venture capital association;

  • Founder, familiar, hired CEO: relationship between CEO and owner family.

It may be said that the presence of venture capital does not, in itself, induce effective governance, but given the prevalence of venture capital among the shareholders of Italian family firms that reached the market, it is impossible to exclude this item from analysis. Anyway, the objective of this empirical evidence is not an understanding of the role of venture capital in family firms, but simply to understand whether its presence affects the governance structure and, as a consequence, market performance.

Tables 5 and 6 present additional descriptive statistics concerning the parameters used as independent variables.

Table 5 Data on variables relating to the corporate governance system
Full size table
Table 6 Data relating to corporate ownership: family ownership of shares before and after the IPO, presence of venture capitalists and CEO
Full size table

The first regression model explains the determination of the “Mid Board” variable, on the basis of the following six predictors: the presence of a venture capitalist, the number of directors, the presence of appointment and compensation committees, the difference in family ownership of shares before and after the IPO, and finally, the percentage of family members present in management. Formally:

$$ \begin{aligned} {\text{Mid Board}} & = b + b1\,({\text{venture capitalist}}) + b2\,({\text{number of directors}}) + b3\,({\text{appointment committee)}} \\ & \quad + b4\,({\text{compensation committee}}) + b5\,({\text{percentage family members present in management}}) \\ & \quad + b6\,({\text{difference in ownership of shares by the family before and after the IPO}}), \\ \end{aligned} $$

where

  • Mid Board: the LN of the ratio between the number of independent directors (Pout) and the number of directors related to the owner family (Pfam);

  • Venture capitalist: a dummy variable that assumes the value 1 if a venture capitalist or a private equity operator is present among shareholders, and 0 otherwise. In order to define a shareholder “venture capitalist” or “private equity operator,” the institutions must be members of AIFI, the Italian venture capital association;

  • Directors: the total number of people belonging to the board (executive and non-executive);

  • Appointment committee: a dummy variable that assumes the value 1 if an appointment committee exists, and 0 otherwise;

  • Compensation committee: a dummy variable that assumes the value 1 if an appointment committee exists, and 0 otherwise;

  • Percentage family members present in management: the ratio between the sum of managers and directors appointed or selected by the family and the total number of directors and managers;

  • Difference in ownership of shares by the family before and after the IPO: the difference in share ownership of the controlling shareholder before and after the IPO.

The second regression model is intended to establish a relation between the composition of corporate governance and particular aspects of family governance and, again, between ownership variables and performance.

In particular, the regression is conducted with a dependent variable, the BHAR, calculated for the securities being studied, 12 months after their being listed on the stock exchange.

The analysis assumes the values of the following equation:

$$ \begin{aligned} {\text{BHAR}} & = b + b1\,({\text{F-PEC}}) + b2\,({\text{executive board}}) + b3\,({\text{outsiders}}) + b4\,({\text{CEO duality}}) \\ & \quad + b5\,({\text{descendant/familial}}) + b6\,({\text{LN(LTDebt/totalassets))}}, \\ \end{aligned} $$

where

  • BHAR: buy and hold abnormal return calculated 12 months after the IPO;

  • F-PEC: the F-PEC score calculated in accordance with the formula proposed by Astrachan et al. (2002);

  • Executive board: a dummy variable that assumes the value 1 if the executive board is present, and 0 otherwise;

  • Outsiders: the number of non-executive (and non-familial) directors serving on the board;

  • CEO duality: a dummy variable that assumes the value 1 if the CEO is also a managing director, and 0 otherwise;

  • Descendent/familial: a dummy variable that assumes the value 1 if the CEO is related to the owner family;

  • LTDebt/Total assets: a sort of firm leverage, defined as the ratio between long term debts and total assets at the end of the year before the IPO. This variable is used because Italian families often exploit bank debts to increase the value of their companies and, more importantly for this empirical evidence, they tend to use debt sources to maintain familial hegemony. Thus, it might be expected that higher BHAR must be connected to greater LTDebt/total assets.

5 Results

The first regression is built so as to place board composition, especially the percentage of outside members, in relation to variables typically pertaining to the characteristics of the body of shareholders, but also to corporate governance mechanisms (Table 7).

Table 7 Determination of the “Mid Board” variable
Full size table

Overall, the model seems to support the contingency theory, as the board, whose composition is the first feature that allows its dependence level to be assessed, is negatively correlated with family share ownership to a very high level of significance. Furthermore, there is equally significant evidence that the presence of a venture capitalist in the body of shareholders can increase the level of board professionalism in a FB, increasing the number of directors who are outsiders both to the firm and to the family.

The variable “difference in shares owned by the family before and after the IPO” can be interpreted in exactly the opposite manner. According to this variable, the lower the proportion of share capital held by the family (which is guaranteed a certain minimum number of voting rights), the more the firm board will be composed of outside directors, rather than family member directors. This would represent a very clear scenario of concentrated control, fully supported by factors which are typical of, though not limited to, family businesses.

The connection between directors associated with family ownership and the percentage of outsiders on the board is less apparent, even if, in regression coefficient terms, this correlation is taken to be negative. This correlation can be explained by inverting this phenomenon’s direction, i.e., by supposing that we are dealing with a board of family members that tends to be inclined toward employing family members in management and which therefore supports the encountered correlation.

However, this paper maintains that this phenomenon actually implies that the general attitude of the same family will be equally inclined toward choosing family members as directors as it is toward selecting non-family members. The tendency to favor family members within the firm appears clear.

Final confirmation comes from the positive correlation between the size of the board and the presence of outsiders. This relationship can immediately be interpreted simply as a greater “spatial” availability of new directors, albeit outsiders; though in this case, we are led to consider the evidence as indicative of a family that is more inclined toward external support, though only to the extent of establishing the number of firm directors.

Though tested for, no significant connections are found concerning the CEO’s status as founder, despite its showing a negative correlation with the dependent variable, nor for family generation and generational transition variables, which instead seem to tend toward a positive correlation with the presence of outsiders.

The second regression yields the figures shown in Table 8.

Table 8 Determination of the BHAR variable
Full size table

The first notable result from this analysis is the negative correlation, significant at the 10% level, between firm performance and presence of outside directors on the board. This outcome supports the stewardship theory and could be related to the fact that in family firms, “interaction problems” between outsiders and family members could emerge more readily than in other situations. Another explanation could be tied to family members’ greater knowledge of the business, which allows them to manage it in the best possible way if the governing body is composed of insider members.

Actually, there emerges from this point of view a confirmation of the contingency theory: it is clear that the corporate governance structure must adapt to its firm-specific environment, and it cannot necessarily respond to rules of composition defined elsewhere.

Another particularly interesting outcome is the negative correlation of the F-PEC index with performance, confirming that family presence affects performance negatively. This begs the question of how to improve performance if both family and outsider presence lead to negative performance. It must be underlined that the answer to this question probably lies in the fact that outsiders, in this empirical evidence, means precisely “independent.” Moreover, it must be noted that in the regression model, F-PEC is a more important explanatory factor than outsiders, as it is significant at the 5% level, rather than the 10% level for outsiders.

The executive committee is a governing body that enables the board to delegate strategic issues and which therefore, in a sense, centralizes control over management in the hands of a few board members. The presence of this type of committee reduces performance, or rather, is related to performance reduction at a considerable degree of significance. This can be explained as follows: more complex decision making where there is a board and clear centralization of power in the hands of family members, which entails less control. Both conditions block communication and decision making while committees are set up for exactly the opposite purpose, which is also in accordance with the directives of the self-regulatory code.

Furthermore, negative correlation with CEO duality is observed; i.e., the CEO’s double incumbency as managing director and chairman within the board of directors, whose relation with performance is self-evidently the result of limited monitoring opportunities, since this task, as theory fully bears out, typically falls within the duties of the chairman of the board.

The negative correlation between performance and a family-member or descendant CEO is also notable. This evidence is consistent with the preceding analysis: a family-member or descendant CEO could indeed be in power only by virtue of his belonging to that family.

The last variable studied in the regression, firm LN (long-term debt/total assets), also proves very interesting, as it shows a positive correlation with performance. This outcome requires deeper interpretation. Two explanations may be supposed. First, there may be a sort of “IPO exuberance” for risky family firms that choose the stock exchange to improve their growth prospects. In this sense, the entire market rewards firms and their potential growth, at least for the first 12 months of listing. Second, and more simply, it may be assumed that greater underpricing characterizes IPOs of more risky family firms.

If the hypotheses and findings are considered together, it may be stated that:

  • Hypothesis 1, which pertains to board dependence and family participation, may be considered true, as it is establishes that the presence of a venture capitalist in the body of shareholders, a smaller number of shares owned by the family after the IPO, and the establishment of larger boards and interior committees, have a positive influence on the organization of boards composed of outsiders/independent directors;

  • Hypothesis 2, which treats the relationship between independent directors and higher share performance, can be considered only slightly true, as the statistical correlation is significant only at the 10% level;

  • Hypothesis 3, concerning committee capability to affect performance positively, may be considered false, in particular for executive committees, because this type of committee reduces performance or, rather, is related to performance reduction at a high degree of significance;

  • Hypothesis 4, pertaining to family and performance, is validated entirely. In fact, higher F-PEC scores or “family blood” related CEOs lead to inferior performance.

6 Conclusions

Using empirical evidence, relationships among performance, corporate governance and ownership structure are tested. In particular, hypotheses are formulated with respect to family firms for the purpose of verifying which of all the most important approaches (that is, agency theory, stewardship theory, and contingency theory) is the most capable in explaining trends and behaviors of Italian listed family firms.

Using data gathered from various databases, a sample of Italian IPOs issued between 1999 and 2005 is created and, with the F-PEC procedure, family firms are identified. The whole sample includes 56 companies and the sample structure (i.e., sector, size, age) is strong enough to support statistical testing. It must be underlined that all results are affected by the small sample size and the F-PEC score used to discriminate family firms from non-family firms.

Four hypotheses are investigated: the first relates to the relationships between board dependence and family participation in the firm. The second relates independent or outside directors to performance. The third hypothesis links governance structure, in particular the presence of executive committees, with share performance, while the fourth tests whether family involvement leads to inferior performance.

Results from the applied models, on the one hand, confirm the hypothesized relationship between family attendance and board dependence and performance, but on the other hand, they tend to rebut assumptions about the role of outsider directors and the ability of committees to improve performance.

With respect to corporate governance and board dependence, it can be affirmed that the presence of venture capitalists among shareholders, a significant sale of shares by the family at the IPO, and the existence of large boards or committees within them, make the establishment of a board composed of outsiders or independent directors easier. These results give the impression that families tend to hold a certain number of “seats,” whatever their role; in fact, there is a higher number of relatives acting as managers in firms, and the presence of directors imposed by the owner family is far lower. It must be noted that this situation cannot be interpreted simply as a need for power or a need to be present on the part of owner families, because if this were true, sitting on boards of directors would become increasingly more important than holding positions as managers, and no relationships would be noted. Hypothesis 1 can be regarded as true.

At the same time, Hypothesis 4 is credible and convincing: family involvement, direct and indirect, affects performance negatively. In particular, family involvement, measured as percentage of shares owned, participation in administration or supervision activities, and presence of CEO related to the owner family, leads to poorer performances. Moreover, in family firms the dual role of the CEO is not amenable to improved market results.

Further, there is a strong relation between performance and firm risk. The empirical evidence shows that the riskier the family firm, the higher the share performance. From a governance point of view, this situation can be interpreted as a lack of ability on the part of family firms to form relationships with outsiders. In fact, since the heavy use of debt is intended as a method that families pursue to keep non-family members out of the firm, results show that the greater the use of this tool, the more firms are able to maximize performance.

Considering the relationships among performance and corporate governance, it can be stated that independent advisors do not affect performance positively, as the significance level is just below 10%. So, for family firms, it cannot be said that outside directors are the best way to improve performance.

More definitive findings emerge from analysis of performance with respect to the presence of execution committees: surely, as there is a significant and negative correlation, such committees can be regarded as worsening performance, owing to the higher costs they generate.

Overall, the findings entirely support neither agency theory nor stewardship theory and consequently, are aligned to outcomes that consider these two theories too standardized to explain such a problematic issue as the relationship among ownership, corporate governance and family firm performance. As in Caselli and Gatti (2007), the results of this study suggest the existence of differing contexts among family firms, so corporate governance mechanisms cannot be applied in a standardized manner—because of their implications for final performance. In other words, for Italian family firms, a “one size fits all” solution does not exist, even in firms where ownership structure and size are similar.

These findings suggest that separation between ownership and management should be more deeply analyzed, in order to define the proper governance mechanisms for family firms. Such analysis should focus on costs and benefits that could lead to variability in market performance.

Of particular value for future investigation would be the possibility for owner families to use corporate governance and board ownership as mechanisms to manage and monitor the firm without missing opportunities that stem from shareholder base enlargement, rather than pursuing exploitation of outsider director skills.

Another topic that should be investigated, especially for Italian family firms, concerns transition and succession problems with respect to defining alternative corporate mechanisms designed to limit the value destruction that is typical of firms managed by a descendent familiar CEO.