Journal of Small Business Management 2006 44(2), pp.

245267

The Relationship between Family Firms and Corporate Governance*

by Simon Bartholomeusz and George A. Tanewski

This paper contributes to the agency theory literature by identifying relations between family control and corporate governance structure. Emerging literature supports the notion that family control creates strong incentives that have potentially competing inuences on the manner in, and extent to, which internal corporate governance mechanisms are utilized. A sample of 100 listed companies (evenly divided between family and nonfamily rms) is used to test the hypotheses that corporate governance structures are different between family and nonfamily rms; and that family rms adopt optimal corporate governance structures. This research nds evidence that suggests that family rms utilize substantially different corporate governance structures from nonfamily rms and that these differences lead to performance differentials. Indeed, results suggest that family control creates, rather than negates, agency costs and future research may be well rewarded by pursuing this latter notion further.

Introduction
New research on family-controlled rms provides a novel method for reexamining a maturing economic paradigm: agency theory. The progeny of Jensen

and Mecklings (1976) ambitious attempt to expose the inner machinery of the black box called the rm, agency theory has been used by a myriad of researchers to explain how the misalignment of

Simon Bartholomeusz is a bachelor of commerce (Honors) graduate, Department of Accounting and Finance, Monash University and is currently completing his law degree, Monash University. He also works for ISS Proxy, Australia, which specializes in governance research. George Tanewski is senior lecturer, Department of Accounting and Finance, Monash University. He has been active in family business research for more than 10 years in his role as educator, researcher, author, and consultant. *The authors are grateful for the comments from the editor and two anonymous reviewers. An earlier version of this paper was presented at the International Council for Small Business (ICSB) 50th World Conference in Washington, D.C., June 1517, 2005.

BARTHOLOMEUSZ AND TANEWSKI

245

interests between the rms participants is brought into equilibrium throughout a diverse range of contexts. One context that has been represented in the literature with relative paucity is family control. The purpose of this paper is to contribute to the small but emerging body of family-rm literature and to provide a new and powerful perspective on the agency theoretic. While prior research (for example, Anderson and Reeb 2003) has focused on the manner in which the different incentives faced by family members impact on outcomes such as performance, another purpose of this paper is to examine the relation between family control and other potentially substitute or reinforcing internal control mechanisms (that is, corporate governance). The combined effect of any interaction between rm type and corporate governance is then tested on performance within a simultaneous equations framework to establish if family rms adopt a wealthmaximizing set of corporate governance mechanisms. Extant research (for example, James 1999) posits that family blockholders in family-controlled rms have different incentives to atomized shareholders in widely held companies. However, the question of whether family ownership provides incentives to reduce agency costs (through a better alignment of shareholder and managerial interests) or create it (by providing opportunities for family members to expropriate the wealth of outside shareholders) remains an open empirical issue. The strongest evidence in the emerging literature tends

to corroborate the former (Anderson and Reeb 2003; Anderson, Mansi, and Reeb 2003), whereas research by Schulze et al. (2001), conducted in the context of proprietary companies, supports the alternative. Thus, the rst purpose of this paper is to examine the nature of the relationship between family control and individual corporate governance variables. This is achieved by using a similar set of variables to that incorporated in Anderson et al. (1998) study of corporate governance in the context of rm diversication, rather than family control. Agency theory suggests that family owners experience different incentives to diversied, atomized shareholders. These distinguishing incentives should manifest themselves through differences in the utilization of other control mechanisms such as lack of external discipline and monitoring, and this paper produces evidence to support this notion. As the focus of the agency argument relates to the incentives of the rms internal participants, it follows that it is reasonable to restrict the discussion to internally set controls (that is, corporate governance mechanisms such as board composition and inside ownership). The inuence of external control mechanisms (for example, the takeover market, the labor market) is assumed to be fairly uniform across all observations in the sample.1 The second purpose of this paper is to examine the degree to which family rms adopt wealth-maximizing internal controls. The shift in focus between the rst purpose and the second is an attempt to go beyond developing

See Agrawal and Knoeber (1996) for a discussion of the distinction between internal and external controls and an enumeration of each. They suggest that, because internal controls are set by the rms participants, they are more likely to be consistent with value-maximization than, say, external control mechanisms (where the effects of the decision-makers actions on rm value may be borne by others). A novel contribution of this paper (the second purpose) is to examine the conditions whereby internal controls may not be consistent with value-maximizationthat is, family control.

246

JOURNAL OF SMALL BUSINESS MANAGEMENT

isolated relationshipsto identify the extent to which the matrix of control mechanisms operate in unison in resolving competing incentives (that is, the incentive to reduce agency costs as opposed to the incentive to expropriate wealth and create agency costs). H2 is tested on a reduced sample (excluding nonfamily rms) within a simultaneous equations framework.2 Evidence of signicant results on individual coefcients indicates the use of suboptimal corporate governance structuresthat is, either too much or too little of a particular mechanism is being utilized by family rms. For example, when Agrawal and Knoeber (1996) regress rm performance on a series of internal control mechanisms in a two-stage least squares (2SLS) framework, they nd that the proportion of outsiders on the board is negative and statistically signicant. They interpret this as implying that the number of outsiders on the board is too great to be consistent with wealth-maximization (that is, it is suboptimal). This paper applies the same process to a cross section of family rms. The remainder of this paper is partitioned as follows: (1) a review of the literature outlines the motivation and signicance of the study, which is then followed by development of several testable hypotheses; (2) review of the data collection procedures; (3) explanation of the empirical procedures; (4) results of the empirical tests; (5) discussion; and (6) conclusions.

Literature Review: Motivation and Signicance

Family Firms and Agency Theory
It was Smith (1981) who rst foresaw, with unprecedented prescience and insight, that managers of widely held

companies, being the managers of other peoples money, cannot well be expected to watch over it with the same anxious vigilance as, say, the managers of a partnership. Smiths sentiments were developed in the pioneering work of Berle and Means (1932) and formalized by Jensen and Meckling (1976), creating the catalyst for a fruitful area of nancial research: agency theory. Jensen and Meckling argued that traditional conceptions of rms as control structures are awed. Firms, as such, are merely legal ctions and are a nexus between a series of complex contractual relationships between various stakeholders. An important aspect of the contract is the relation between the owner and the manager. This relationship takes on the special character of being an agency relationship, which involves delegation of decisions. If both parties are rational utility maximizers, it follows that the contract may leave scope for the manager to make decisions that are not in the owners interests. The costs of this divergence of interests, in combination with the costs of monitoring and bonding the manager to limit this divergence, are termed agency costs. A great deal of the literature has been devoted to applying agency theory to diffusely held corporations; that is, those rms where professional managers pursue the control function on behalf of a variety of diversied, atomized shareholders. However, the effect of agency costs in other ownership structures, in particular, the family-founded, -owned, and -controlled rm is covered by the literature with relative paucity. While important work has been directed towards developing an understanding of this unique ownership structure, when it is considered that family rms account for around 20 percent of the listed companies in Australia (Mroczkowski and

Our 2SLS procedure is analogous to the methodology used by Agrawal and Knoeber (1996).

BARTHOLOMEUSZ AND TANEWSKI

247

Tanewski 2006; Harijono, Ariff, and Tanewski 2004), approximately one-third of the S&P 500 in the United States (Anderson, Mansi, and Reeb 2003), and more than half of the 250 largest rms on the Paris and Frankfurt bourses are family-controlled (Blondel, Rowell, and Van der Heyden, 2002; Klein and Blondel 2002). It would seem that the economic signicance of family rms has been underrepresented by the academic literature. Research on family rms provides an avenue for developing agency theory in a new and fundamentally different context. This is largely the result of entirely different incentives that motivate family shareholders in contrast to diversied shareholders. Consistent with Demsetz and Lehn (1985), Anderson and Reeb (2003) acknowledge two reasons why family rms have several incentives to reduce agency costs. First, as family rms have concentrated shareholdings, they have an increased incentive to reduce agency costs because the more concentrated is ownership, the greater the degree to which the benets and costs are borne by the same owner (Demsetz and Lehn 1985, p. 1156). However, family ownership extends beyond concentration. As the well-being of the family is tied directly to the welfare of the company, families have further incentives to reduce agency costs that might impede performance. Additionally, a familys special technical knowledge concerning a rms operations may put it in a better position to monitor the rm more effectively. Essentially, families in family rms have an additional incentive to counteract the free rider problem that prevents atomized shareholders from bearing the costs of monitoring, ultimately reducing agency costs.

Second, it is maintained that family rms make decisions on much longer time horizons than nonfamily rms. It is suggested that family owners view the rm as an asset to be passed on to subsequent generations (Chami 1999), leading to strict adherence to maximizing the value of the rm ( James 1999). Contemplate the mindset of a new corporate executive ofcer (CEO) who knows the average CEO tenure in Australia to be 4.4 years (Booz Allen Hamilton and Business Council of Australia 2003), accepts that it is unlikely that he or she will outperform the average, and therefore decides to manage the company over a short time frame with strict adherence to his or her own utility maximization. Contrast that outlook against the mindset of a founding family CEO who intends to maximize the long-term wealth of a family company that bears the familys name, and with it carries the familys reputation. Anderson, Mansi, and Reeb (2003) elaborate on this point. Following Jensen and Meckling (1976), they argue that atomized shareholders have an incentive to take on risky projects with a view to expropriating the wealth of bondholders. Family members (because of their undiversied shareholding, long-term interest, and concern for reputation) have a fundamentally different risk prole to typical equity holders. As a consequence, they are more likely to maximize the overall value of the company (as opposed to the value of equity) reducing the agency cost of debt.3 Indeed, Schulze et al. (2001), while echoing the benets of personal involvement and special relationships, also acknowledge the altruistic benets unique to family ownership. They cite Bergstrom (1995) and Becker (1981) who consider altruism to be an economic

This is empirically corroborated by Anderson, Mansi, and Reebs (2003) results that suggest that the average cost of debt nancing is 32 basis points lower for family rms than nonfamily rms.

248

JOURNAL OF SMALL BUSINESS MANAGEMENT

phenomenon that occurs when an individual pursues their self-interest by promoting that of others. Eschel et al. (1998), Simon (1993), and Jensen (1979) develop this notion in the context of the familyit is in each family members interests to pursue their own interests by promoting the interests of the rest of the family. This in turn develops loyalty, facilitates communication, and increases time horizons for decision-making, ultimately reducing agency costs. In contrast, it has also been posited that family ownership leads to an increase in agency costs (for example, Anderson and Reeb 2003; PerezGonzalez 2002). Morck, Stangeland, and Yeung (2000) identify the possibility that family rms might use their concentrated blockholding to expropriate wealth from other shareholders through excessive compensation, related party transactions, and special dividends. Anderson et al. (2003) add the possibility of risk avoidance, that is, because of their undiversied exposure; family rms may use their control of a rm to avoid risks acceptable to other more diversied shareholders. Morck and Yeung (2003) note the potential for a family business group to organize itself into a pyramidal control structure that facilitates the expropriation of wealth from nonfamily shareholders in family subsidiaries to family holding companies. It is asserted that agency costs in family business groups stem from either management not acting for the shareholders, or rather, acting only in the interests of family shareholders. Moreover, in this context, altruism has the potential to create agency costs if family members pursue the interest of other members at the expense of outsiders. Buchanan (1975) notes that parents have an incentive to be excessively generous, allowing their children to free ride. Schulze et al. (2001) elaborate on the potential for nonfamily shareholders to be held to ransom at the mercy of family

owners. While Jensen and Meckling (1976) acknowledge the potential for conict to arise between owners interests, these would be resolved efciently by one of the owners selling their shares. However, this assertion is predicated on the assumption that capital markets are efcient, allowing the conceding owner to sell their stake in the rm at its full value. Schulze et al. (2001) point to inefciencies in the capital market that may put pressure on the conceding owner to acquiesce. While the focus of their research is directed at proprietary companies (with illiquid capital markets), similar problems could affect small public companies. In the spirit of Coase (1937), who acknowledged that activities will be included within the bounds of the rm when it is less expensive to do so than to go through external markets, Demsetz and Lehn (1985) were the rst to directly acknowledge that, while concentrated ownership generally serves the interests of diffuse shareholders well, if the concentrated owner can achieve their consumption goals more effectively through the rm than privately, the welfare of the diffuse owners may be threatened.

Corporate Governance and Economic Signicance

While new research on family rms would further develop a edgling body of literature that provides a new perspective on an established academic paradigm, including corporate governance within the scope of discussion, gives this paper an economic signicance that would be valuable to a range of the markets participants. Research addressing the relationship between family rms and corporate governance would be of nontrivial value to professionals, policymakers, regulators, and anyone else concerned with either sphere of interest. While the concepts encompassed by corporate governance are as old as Adam

BARTHOLOMEUSZ AND TANEWSKI

249

Smith, the term itself has only really gained mainstream prominence since the corporate collapses of the 1980s. The Cadbury Commission report (1992, p. 12) . . . based on compliance with a voluntary code coupled with disclosure . . . was the progenitor of corporate governance discussion within the United Kingdom; the response to a number of high prole collapses. Recommendations of the report related to nonexecutive directors, audit committees, disclosure of directors compensation, disclosure of accounting statements, internal control systems, and the role of the auditor. The Commissions sentiments were echoed in Australia by the Australian Investment Managers Associations Blue Book (1999).4 More recently, corporate governance has become increasingly topical because of the adoption of the SarbanesOxley Act of 2002 in the United States, a legislative response to the excesses encompassed by the spectacular collapse of the dot-com and telco speculative bubbles of the late 1990s. Provoked by the collapse of Enron (which used complex off-balance sheet vehicles to hide extraordinary losses leading to a write-down in shareholder funds of US$1.2 billion) and Worldcom (which used accounting improprieties to inate reported earnings and cash ows by US$3.9 billion) (Bartholomeusz 2002), the Securities and Exchange Commission (SEC) response, while similar in theme to earlier reforms, differs fundamentally in that the requirements are mandatory. The implications for other regulators are

unambiguous, if their companies are to compete in international markets, their standards of corporate governance must be no less effective than those imposed in the largest capital market in the world. Notwithstanding the SECs mandatory requirements, the attitudes of other key regulators have maintained their voluntary tone.5 In any respect, policy on corporate governance continues to be generated, yet, it is unclear exactly how much of this policy is grounded upon rm empirical evidence rather than intuition, anecdotal accounts, or perhaps, partisan politics. For example, the response in the United States, the United Kingdom, and Australia has largely focused on the role of independent directors. However, extant research has shown no compelling relationship between the proportion of outsider directors on the board and performance (Hermalin and Weisbach 2003; Klein 1998). It would seem that normative policy is being advocated without the grounding of a rm positive understanding of how the misalignment of interests between various participants is actually brought into equilibrium. Convincing research on the role of corporate governance in family rms would (1) allow policymakers to develop a regulatory framework that recognized the contingencies inherent in different ownership structures based on sound theory corroborated by empirical evidence rather than rhetoric; and (2) improve market efciency by providing information that

The Blue Book is a corporate governance compendium for Australian investors, institutional investors, and the ASX. 5 For example, in the United Kingdom see Higgs (2003) report and the Smith Report (2003). This papers sample is drawn from Australia, which follows a similar regime to that in the United Kingdom. For those that are interested, contrast the mandatory elements of the ninth redevelopment of the Audit Reform and Corporate Disclosure (Commonwealth of Australia 2003) and the Corporate Disclosure (Commonwealth of Australia 2002) with the voluntary aspects of the Principles of Good Corporate Governance (ASX 2003).

250

JOURNAL OF SMALL BUSINESS MANAGEMENT

allowed market participants to correctly value different ownership structures after taking into account the differences in value attributed to different corporate governance structures. Ultimately, research relating corporate governance and family rms would be of practical value to a myriad of interested parties, providing a positive basis for a consistent set of normative assertions.

Corporate Governance and Family Firms

While there are compelling reasons for pursuing research on family rms and corporate governance independently, there are further incentives to develop the relationship between the two. One incentive, in particular, is motivated by the emerging body of nancial literature that focuses on rights as a determinant of corporate value. On a macro level, recent research by La Porta et al. (1999, 1998) has identied the relationship between variation in property rights and corporate value between different property rights regimes. While the focus of La Porta and his colleagues paper is directed across different rms in different nations, the essence of the argument operates at the rm level: the rights attached to a security affect the ability of the security holder to procure the securitys value and hence, the value of the security itself. While, in a macro sense, the greatest source of variation in the rights of a security will relate to differences in legal regimes, within the same legal regime, the greatest source of variation in shareholder rights will relate to rm-specic corporate governance (that is, the Corporations Act of 2002 operates fairly uniformly across all Australian rms). As corporate governance mechanisms are largely set by the internal participants within the rm, they should be consistent with wealth maximization. That is, according to the optimal governance hypothesis (Anderson et al. 2003), rm participants will voluntarily adopt

corporate governance mechanisms to the point where the marginal benet of adopting further mechanisms (the reduction in agency costs) is just offset by the extra cost of implementation. Thus, a purpose of this paper is to determine whether the optimal governance hypothesis holds true in the presence of family control. One stream of extant literature suggests that family control is, potentially, an agency costreducing mechanism in itself. Family rms are concentrated blockholders with a unique incentive to overcome the free rider problem that prevents atomized shareholders and, indeed, some other blockholders from effectively monitoring management (Anderson and Reeb 2003; Tufano 1996). Furthermore, as the wealth of the family is directly tied to the future of the companyand decision-making in family rms is predicated on much longer time horizons than in nonfamily rmsfamily rms more strictly adhere to wealth maximization than their counterparts (Chami 1999; James 1999). These reasons, in combination with the ancillary benets of altruism (Schulze et al. 2001), suggest that family control is an agency cost-reducing mechanism. When it is considered that family control is one of several potentially substitute or reinforcing mechanisms, it follows that the combination of corporate governance structures adopted by family rms will be different to the combination of structures used by nonfamily rms but no less consistent with wealth maximization. However, there is also a line of argument within the agency theoretic that suggests family control creates (rather than negates) agency costs. It has been suggested that family control provides family members with a unique opportunity (not available to other shareholders) to use their concentrated blockholding to expropriate the wealth of outside shareholders through excessive compensation, related-party transactions, special

BARTHOLOMEUSZ AND TANEWSKI

251

dividends, and risk avoidance (Anderson and Reeb 2003; Morck and Yeung 2003; Anderson et al. 2002). If family members use their control of the rm to expropriate the wealth of outside shareholders, it would be expected that corporate governance structures would be different between family and nonfamily rms and inconsistent with wealth maximization. The logic behind this idea is straightforward. Assuming the effects of leverage to be constant, shareholders only benet when management attempts to maximize the value of the company. If families in family rms are able to derive benets through means that are not shared with other nonfamily shareholders, their actions may not be consistent with maximizing the value of the company (that is, wealth maximization). One of the objectives of this paper is to examine this fundamental conict empirically. Any differences in corporate governance variables between family rms and nonfamily rms are, ex-ante, predictable: if the wealth of outside shareholders is being expropriated, managerial compensation should be higher in family than in nonfamily rms, the sensitivity of compensation to performance should be lower, board characteristics should portray inefciencies, and the CEOs personal characteristics should be consistent with entrenchment. The effects of inside ownership are more complex when it is considered that lower levels of inside ownership facilitate the reduction of agency costs whereas higher levels of inside ownership may create further entrenchment effects. The following hypotheses, stated in the alternative, naturally follow: H1: Family rms adopt different corporate governance structures to nonfamily rms. H1a: Managerial compensation is higher/lower in family rms than in nonfamily rms.

H1b: Managerial compensation is more/less sensitive to performance in family rms than in nonfamily rms. H1c: Board characteristics (that is, proportion of insiders, proportion of outsiders, and likelihood that the chairperson and CEO roles are merged) are higher/lower between family and nonfamily rms. H1d: Ownership structure holdings (e.g., CEOs holdings, board holdings) are higher/lower between family and nonfamily rms. H2: Family rms subadopt optimal corporate governance structures.

Methods
Sample Selection
The sample includes a cross-section of 100 rms, trading on the Australian Stock Exchange (ASX) in 2002. Each rm has a common CEO (that is, the same CEO over the sample period) over 20012002 and a common reporting period ( June 30). The constraint of a common CEO diminishes the effects of confounding noise on the compensation variables. It also increases the likelihood that randomly chosen nonfamily rms will have similar ownership characteristics to family rms (as family control, by denition, includes a CEO with a large shareholding). However, it does have the potential to introduce survivorship bias (that is, more successful companies are less likely to have CEO turnover). Sampling proceeded on the basis that all family rms that met the above constraint were included in the sample. After eliminating outliers, 50 family rms were present in the sample. A further 50 matched nonfamily-controlled rms were chosen on the basis of size (total assets) and ASX industry classication. Unfortunately, thin coverage over the ASX industry codes meant that companion rms

252

JOURNAL OF SMALL BUSINESS MANAGEMENT

could not be identied for about 10 companies. As a consequence, a further 10 companies were chosen at random to provide symmetry to the sample. Firm-level data on corporate governance and accounting variables were collected for each rm as reported in corporate reports (available from Datastream). Share-price information and some further accounting information were collected from Datastream. Data were collected solely for 2002 for all variables except those required in the compensation analysis (where data covering two years were collected). Corporate governance variables are reasonably static over time such that the absence of a time-series dimension to the analysis remains a limitation of the paper. The adopted denition of family control corresponds with that used by Mroczkowski and Tanewski (2006), who dene a family-controlled rm as an entity controlled by a private individual, directly or indirectly, in conjunction with close family members (p. 10). Inclusion is based upon the following criteria: The existence of a founding member involved in management with greater than 20 percent of voting shares. The shareholder is CEO or a key member of the board (that is, chairperson). At least one other related party is a member of the board. The original shareholder and the related parties hold greater than 40 percent of the voting shares of the company. That is, 40 percent of the equity of each family rm in the sample is held by family members. The family-controlled rm dummy variable (FF) takes on the value 1 for rms that satisfy the above criteria or 0 otherwise (see Appendix A for variable abbreviations and their respective denitions).

Compensation
The measurement of compensation variables requires an analogous technique to that used by Anderson et al. (1998). Fixed emoluments (e.g., salary, cash bonus, superannuation, allowances motor-vehicle expense) are aggregated to form the xed portion of total income (C_FIXINC02 and C_FIXINC02). A second measure calculates the value of the CEOs option portfolio (C_OPTINC02 and C_OPTINC01). Details on each executive option package were manually collected from annual reports (exercise price, time to maturity, and the number of options granted). Share price information on DataStream was used to calculate six-month historical volatilities. The ex ante dividend yield was approximated by the ex-post two-year average. Separate rm-level volatilities and dividend yields were calculated for each of the two-sample years. The Australian 10-year bond yield (as of the two reporting dates) was used to proxy for the shadow risk-free rate. The value of the portfolio was calculated using the Black and Scholes (1973) model adjusted for continuously paid dividends.

Board of Directors
Board composition is determined using Weisbachs (1988) trichotomous classication scheme. A director who is a full-time employee of the company is classied as an inside director. A director who is neither an employee nor has extensive dealings with the company is referred to as an outside director. All other directors, who are not full-time employees but have relationships with the company (for example, family relationships, consultants) are designated as gray directors or afliates. Director classication is determined by reading biographies in annual reports, analyzing related party transactions, and by inference from the denition of family rm.

BARTHOLOMEUSZ AND TANEWSKI

253

Performance
Performance is measured using Tobins Q ratios (Q). The adopted measure of Q is calculated by dividing the sum of the market value of assets by their book value. This paper uses Q as a measure of performance to maintain the close analogy between the method used by Agrawal and Knoeber (1996) and that used in this paper in the simultaneous equations setup. Moreover, Morck, Shleifer, and Vishny (1998) advocate the use of Q ratios in cross-sectional studies involving ownership structure and corporate performance. The real value of using Q here, as opposed to accounting measures of performance, is gained through the capacity to measure deviations from wealth maximization. In this context, wealth maximization means maximizing the value of the companys assetsthat is, maximizing Q before deating by book values.

Compensationij = b0 j + b1j FamilyFirmi + b2j MVEi + b3 j MVEi FFi + b4 j EBITDAi + b5j EBITDAi FFi + e ij , where Compensation is the rst difference in any of the ( j ) compensation measures between 2002 and 2001. The compensation measures include the change in the level of xed income and option income, and the change in the aggregate of the two. The interaction terms test whether compensation is more (or less) sensitive to previous performance in family rms as opposed to nonfamily rms. Previous performance is measured by the change in the market value of equity (MVE) (a stock performance measure) and the change in EBITDA (an accounting measure). If family rms adopt compensation policies that promote entrenchment, it is predicted that the direction of the interaction terms will be negative. The relation between executive compensation and rm performance has been intensely debated within the literature (see Bebchuk and Fried 2004, 2003 for an extensive discussion of the problemthe thin link between pay and performancerecent empirical evidence and theoretical developments). It would be particularly useful to establish empirical evidence that condently places an understanding of family control within the context of the wider debate.

Empirical Specication
The models used to test H1 are similar to those used in Anderson et al. (1998) in the context of diversication and corporate governance, whereas H2 involves an adaptation of the methodology used by Agrawal and Knoeber (1996) in their analysis of alternative control mechanisms. The rst set of regressions is as follows: Varij = b0 j + b1j FamilyFirmi + b2j Sizei + b3 j Industryi + e ij , where the subscript i denotes the rmlevel observation for each variable in 2002 and Varj represents each of the possible corporate governance variables that may be used as dependent variables (see Appendix A). It is proposed that each of the ownership, board composition, and compensation variables be substituted as the dependent variable. The second set of regressions, directed at testing the sensitivity of compensation to performance, take the following form:

2SLS Regression Analysis

Tobinss Q is adopted as a measure of performance. However, extant literature highlights potential endogeneity problems surrounding regression analyses of corporate governance mechanisms and performance. For example, consider the following equations:

254

JOURNAL OF SMALL BUSINESS MANAGEMENT

InsideOwnership = b0 + b1Qi + e i (1) Qi = d 0 + d 1InsideOwnershipi + d 2Sizei (2) + d 3 Industryi + mi . Inside ownership is a function of performance (that is, when rms perform well, insiders increase their holdings), however, performance is not exogenous. Performance is a function of inside ownership (that is, inside ownership leads to a decrease in agency costs and an increase in performance) as well as other control variables. If e and m are correlated, Q (a function of m) will be correlated with e and b1 in equation 1 will be biased. The simultaneous bias will occur as long as e and m are correlated and Equation (a) is over-identied (this is so even if inside ownership is not a determinant of Q in equation 2 but Q is not exogenous). Agrawal and Knoeber (1996) propose the use of 2SLS regressions in the context of endogenously determined corporate governance mechanisms. The method involves, rst, estimating ordinary least squares (OLS) predictions for each endogenous regressor. Second, each of the predictions is regressed on Q together to determine consistent estimates for each endogenous regressor. This method allows for the interdependence and alternative use of all of the governance mechanisms. To test H2, an analogous method to that of Agrawal and Knoeber is applied to a restricted subsample that includes only the 50 family rms. A 2SLS regression is estimated by regressing six endogenous corporate governance variables on Q. Predictions for each of the endogenous independent variables is estimated from the following equations: P _ CEOi = b0 +

P _ODi = b0 +

j P _ OD

b jVarij

+ b7STDEVi + b8 B _ SIZEi + b9SIZEi + b10 INDi + e ij P _ BLOCK i = b0 +

j P_ BLOCK

(4)

b jVarij

+ b7 INDEX i + b8SIZEi + b9 INDi + e ij B _ POi = b0 +

j B_ PO

(5)

b jVarij

+ b7SIZEi + b8 INDi + e ij (6) C _PEQ 02i = C _OPTINC 02i C _OPTINC 02i + C _FIXINC 02i
j PEQ02

= b0 +

b jVarij

+ b7STK _ RETi + b8SIZEi + b9 INDi + e ij DEBTi = b0 + (7) b jVarij + b7CFRi (8)

j DEBT

+ b8SIZEi + b9 INDi + e ij ,

where Varij (rm i, variable j) is the observed value of each of the seven endogenous variables for each rm observation in the restricted sample (Q is also endogenous). Control variables have been adopted in line with Agrawal and Knoeber. By estimating each of the above regressions, predictions for each endogenous regressor may be purged of simultaneous bias. Hence, the coefcients of the independent variables in the following regression should be consistent: Qi = b0 + b1P _ CEOi + b2 P _ CEOi2 + b3 P _ODi + b4 P _ BLOCK i + b5 B _ POi + b6C _ PEQ 02i + b7 DEBTi + b8SIZEi

j P _ CEO

b jVarij

+ b7STDEVi + b8SIZEi + b9 INDi + e ij (3)

+ b9 INDi + e i ,

(9)

BARTHOLOMEUSZ AND TANEWSKI

255

where the rst seven independent variables (excluding the constant term) are the predicted values from regressions 1 through 6. If the coefcients in equation 9 are signicant, the null in H2 will be rejected: there is evidence to suggest that family rms adopt suboptimal corporate governance structures. In other words, any signicance in the models independent variables that persists into the second stage is inconsistent with wealth maximization. That is, signicant positive coefcients suggest that increasing the use of the control mechanism would improve performance, whereas negative coefcients suggest that reducing the use of the control mechanism would lead to performance improvements. If the mechanism is used optimally, it should not be signicantly related to performance in the second stage (its coefcient should not be signicantly different from zero).

Results
Table 1 provides descriptive statistics on the ownership variables. Each variable seems to be consistent with both intuition and prior research. It should be noted at the outset that, because the incidence of family rms is greater in the sample (that is, 50 : 50 or family : nonfamily rms) than it probably is in the population (Mroczkowski and Tanewski 2006 suggest a split of 20 : 80), it is possible that many of the summary statistics are a function of the overweighting of family rms in the sample. Table 2 reports the results of 15 regressions that test various aspects of H1. In each regression, a family-control dummy variable and a rm-size control variable (the natural logarithm of total assets) are regressed on each of the ownership variables. The results of regressions demonstrate that the CEO is likely to hold 13 percent ( p < .001) more of the outstanding capital in family rms than in nonfamily rms (Panel A, equation 1). The CEOs direct shareholding in family

rms is not signicantly (b = 0.0059, p = .346) different from that of CEOs in nonfamily rms (Panel A, equation 5). Rather, their incentives are likely to be derived from indirect shareholdings: typically nominee companies where the benecial interest is held jointly with other family members. Similarly, outside directors are likely to hold 12 percent ( p < .003) less of the outstanding capital (Panel B, equation 1). Consistent with the notion of reduced external monitoring, outside blockholders are likely to hold 17 percent less of the issued capital in family rms than in nonfamily rms (Panel B, equation 7). Given the parsimony of the models, the R2 values suggest that family control is an economically signicant determinant of ownership composition. The regressions provide clear evidence that family ownership displaces nonfamily owners. Large blockholders are less likely to be prevalent. Moreover, outside directors are likely to hold substantially fewer shares (and, consequently, have less of an incentive to monitor). In both instances, the scope for external discipline of the CEOs actions is reduced. Table 2 (Panel C) supports the differences in board composition and while the proportion of insider representation is not signicantly different between family and nonfamily rms, family control is associated with 19 percent less outsider representation (Panel C, equation 1), compensated by 15 percent more representation by afliates (gray) directors (Panel C, equation 5). This result is robust to the inclusion of a rm-size dummy (which is signicant in two of the three models). The R2 terms for the B_PO and B_PG (21 and 20 percent, respectively) variables further emphasize the economic signicance of family control as a determinant of board composition. Regressions 7 and 8 nd no evidence of any variation in board size between family and nonfamily rms.

256

JOURNAL OF SMALL BUSINESS MANAGEMENT

Table 1 Descriptive Statistics

Panel A: Ownership P_CEO_I 0.1150 0.0042 0.8788 0.000 0.1926 1.8983 6.0693 99.314 (0.000) 100 Panel B: Board Composition B_PI 0.3568 0.3333 1.0000 0.125 0.1792 1.4110 5.8824 67.798 (0.000) 100 0.5531 0.6000 0.8571 0.000 0.2221 0.9571 3.4392 16.072 (0.000) 100 B_PO B_PG 0.0901 0.0000 0.7500 0.000 0.1625 2.0670 7.0192 138.518 (0.000) 100 B_SIZE 5.5800 5.5000 12.0000 2.000 1.9857 0.7563 3.8270 12.382 (0.000) 100 0.0309 0.0032 0.4877 0.000 0.0699 3.8998 21.8315 1731.081 (0.000) 100 0.1156 0.0264 0.9877 0.000 0.2016 2.5786 9.4532 284.338 (0.000) 100 0.0393 0.0029 0.8036 0.000 0.1039 5.0810 33.7907 4,380.565 (0.000) 100 P_CEO_D P_OD_A P_OD_D P_OD_I 0.0764 0.0047 0.9760 0.000 0.1795 3.4095 14.6701 761.206 (0.000) 100 P_BLOCK 0.2145 0.1534 0.9914 0.000 0.2092 1.2240 4.5153 34.539 (0.000) 100

P_CEO_A

Mean Median Maximum Minimum S.D.a Skewness Kurtosis Jarque-Bera Probability Observations

0.1459 0.0495 0.8788 0.000 0.2001 1.5216 4.6324 49.693 (0.000) 100

BARTHOLOMEUSZ AND TANEWSKI

B_DUAL

Mean Median Maximum Minimum S.D. Skewness Kurtosis Jarque-Bera Probability Observations

0.1600 0.0000 1.0000 0.000 0.3685 1.8549 4.4405 65.987 (0.000) 100

257

258

Table 1 Continued
Panel C: Compensation C_OPTINC01 C_OPTINC02 C_TOTAL01 C_TOTAL02 C_PEQ01 C_PEQ02 0.159 0.002 0.953 0.000 0.235 1.546 4.601 50.012 0.000 99 0.166 0.015 0.922 0.000 0.238 1.447 4.148 39.976 0.000 99

C_FIXINC01

C_FIXINC02

JOURNAL OF SMALL BUSINESS MANAGEMENT

345,403.20 394,197.60 880,082.90 989,432.70 518.71 3,707.24 382,393.00 440,000.00 8,022,212.00 8,249,577.00 10,122,212.00 11,716,618.00 0.00 0.00 41,675.00 41,675.00 1,183,477.00 1,339,635.00 1,531,104.00 2,029,678.00 4.907 4.943 3.708 4.349 27.807 27.574 18.456 21.960 2,935.902 2,894.061 1212.204 1,794.919 0.000 0.000 0.000 0.000 99 99 99 99

Mean 534,679.80 595,235.10 Median 312,213.00 358,362.00 Maximum 4,274,806.00 9,037,873.00 Minimum 41,675.00 41,675.00 S.D. 706,430.90 1,037,610.00 Skewness 3.278 6.041 Kurtosis 14.805 46.527 Jarque-Bera 752.158 8,417.426 Probability 0.000 0.000 Observations 99 99

S.D.: standard deviation.

Table 2 Regression Analysis of Ownership and Board Composition

Panel Aa P_CEO_A (1) Intercept FF LN(TA) R2 F statistic 0.2658 (0.166) 0.1252 (0.001) 0.0102 (0.334) 0.0901 5.9041 (0.004) (2) 0.2658 (0.166) 0.1252 (0.001) 0.0102 (0.334) 0.0901 5.9041 (0.004) (3) 0.2599 (0.132) 0.1119 (0.003) 0.0112 (0.273) 0.0976 5.2443 (0.007) Panel B P_OD_A (1) Intercept FF LN(TA) R2 F statistic 0.0961 (0.622) 0.1179 (0.003) 0.0044 (0.683) 0.0695 4.6994 (0.011) (2) 0.0997 (0.624) 0.0729 (0.000) 0.0023 (0.974) (3) 0.0891 (0.618) 0.0628 (0.082) 0.0010 (0.915) 0.0311 1.5580 (0.216) P_OD_I (4) 0.0905 (0.013) 0.0281 (0.000) 0.003 (0.133) (5) 0.0070 (0.945) 0.0551 (0.008) 0.0033 (0.550) 0.0749 3.9252 (0.023) P_OD_D (6) 0.0243 (0.137) 0.0118 (0.000) 0.0004 (0.648) P_BLOCK (7) 0.0969 (0.608) 0.167 (0.000) 0.022 (0.037) 0.1842 12.1733 (0.000) (8) 1.1662 (0.007) 0.1657 (0.000) 0.4978 (0.001) P_CEO_I (4) 0.1775 (0.113) 0.0675 (0.003) 0.0079 (0.200) (5) 0.0133 (0.933) 0.0059 (0.346) 0.0010 (0.792) 0.0098 0.4798 (0.620) P_CEO_D (6) 0.0214 (0.140) 0.0024 (0.405) 0.0008 (0.999)

(0.000)

Panel Cb B_PO (1) Intercept FFd LN(TA)e R2 F statistic 0.1958 (0.324) 0.1857 (0.000) 0.0250 (0.023) 0.2064 13.8715 (0.000) (2) 0.7424 (0.115) 0.1747 (0.000) 0.4885 (0.003) (3) 0.6529 (0.000) 0.0394 (0.268) 0.0176 (0.073) 0.0255 2.2949 (0.106) B_PI (4) 1.1392 (0.004) 0.0361 (0.233) 0.2828 (0.057) (5) 0.1514 (0.300) 0.1463 (0.000) 0.0075 (0.351) 0.1970 13.1401 (0.000) B_PG (6) 0.0000 (0.685) 0.1429 (0.000) 0.0000 (0.693) LN(B_SIZE) (7) 0.8244 (0.002) 0.0389 (0.458) 0.1369 (0.000) 0.4752 45.8283 (0.000) (8) 5.3310 (0.000) 0.0459 (0.385) 2.4850 (0.000) B_DUAL (9) 3.9680 (0.180) 1.3396 (0.035) 0.3659 (0.032) 0.1145 10.0646 (0.007)

BARTHOLOMEUSZ AND TANEWSKI

259

Table 2 Continued
Panel D: Compensation Level Variablesc LN(C_FIXINC02) (1) C FFa LN(TA) 6.106 (0.000) 0.1535 (0.185) 0.3756 (0.000) (2) 4.9429 (0.004) 0.1121 (0.340) 6.165 (0.000) LN(C_OPTINC02+1) (3) 4.2756 (0.492) 1.8705 (0.119) 0.637 (0.068) (4) 24.4971 (0.151) 1.8703 (0.119) 10.9829 (0.064) LN(C_TOTALINC02) (5) 5.1835 (0.000) 0.3077 (0.029) 0.4452 (0.000) (6) 7.7581 (0.000) 0.2971 (0.037) 7.254 (0.000)

The odd-numbered regressions are ordinary least squares regressions. Where Levenes test for homogeneity of variance is rejected, test statistics have been corrected using Whites (1980) heteroskedastic-consistent variancecovariance matrix. Regression coefcients and p-values (in parentheses) from the even-numbered regressions are derived through M-estimation using Hubers (1981) inuence function. b Regressions 1, 3, 5, and 7 are OLS regressions. Where Levenes test of homogeneity of variance is rejected, test statistics have been calculated using Whites (1980) heteroskedastic-consistent standard errors. Regression 8 is a logistic regression (z statistics in curved brackets, p-value in square brackets). R2 is the adjusted R2 except in regression 9, in which case McFaddens R2 has been reported. Regressions 2, 4, and 6 are derived using Hubers (1981) M-estimator. c Regressions 1, 3, and 5 are OLS regressions. Where Levenes test for homogeneity of variance is rejected, test statistics have been corrected using Whites (1980) heteroskedastic-consistent variance covariance matrix. Regression coefcients and p-values (in parentheses) in equations 2, 4, and 6 are derived through M-estimation using Hubers (1981) inuence function. d FF: family-controlled rm dummy variable. e LN(TA): natural logarithm of total assets.

Equation 9 (Table 2Panel C) is a logistic regression relating the probability of the chairperson and CEO roles being occupied by the same person to family control. Taking e to the power of 1.34, equation 9 suggests that singularity of the chairperson and CEO positions is 3.82 ( p < .05) times more likely for family rms than it is for nonfamily rms. The regressions add further validity to the notion that family control is typically characterized by large shareholdings concentrated in the hands of a few family members who occupy most of

the seats on the board. Nonfamily shareholders are offered little chance for either direct representation on the board, or alternatively, indirect protection through the presence of external monitoring (through either outsiders occupying board seats or the presence of institutional activists or other blockholders). Table 2 (Panel D) reports the results of six regressions examining the relation between family control and the level of compensation in 2002.6 Firm size is signicant in each of the three models.

2001 data were not examined, as data were not collected for a number of control variables used to test for robustness of the results.

260

JOURNAL OF SMALL BUSINESS MANAGEMENT

Interestingly, while the family-rm dummy is not signicant in either of the rst two models (where family control is regressed on the xed component and the equity component of income separately), the nal model that aggregates the two dependent variables of the previous models together, is signicant at the 5 percent level for both specications. At rst glance, the notion that family rm CEOs receive less remuneration than their nonfamily counterparts seems to counteract the other results which are all consistent with entrenchment. However, it is plausible that familyrm CEOs are more likely than nonfamily CEOs to receive perquisites and other benets that are unobservable, which more than compensate them for the decit in disclosed remuneration. It would be useful if further research elaborated on this point with the luxury of a larger sample and a clever research design. Table 3 reports the results of the 2SLS regression on the subsample of 50 family rms. Each of the independent variables in the rst column (with the exception of LN(TA)) is endogenous. To purge the model of simultaneous bias, each endogenous regressor is formed as the predicted value from a separate regression (hence, the sufx F). Agrawal and Knoeber (1996) suggest that, once the optimal determinants of each endogenous regressor are derived in the rst stage, any signicance that persists into the second stage is inconsistent with the notion of optimal use. That is, positive coefcients on independent variables suggest that increasing the use of the control mechanism would improve performance, whereas negative coefcients suggest that reducing the use of the control mechanism would lead to performance improvements. If the mechanism is used optimally, the marginal benets from increasing its use should be offset by the marginal costs, that is, it should not be signicantly related to per-

formance in the second stage (its coefcient should not be signicantly different from zero). Tests of the H1 rejected the null of no differences between the corporate governance structures of family and nonfamily rms. The analysis has now been redirected to determine whether these differences are consistent with optimal use after adjusting for the interaction and substitution effects between different control mechanisms. The coefcient on P_CEO_AF suggests that, at lower levels, family rms could benet from performance improvements by increasing the level of the CEOs shareholdings. At higher levels (captured by the quadratic term), inside ownership is consistent with optimal use. Together, these results suggest that family ownership is only efcient when the CEO has a signicant personal stake in the equity of the company. More interestingly, the negative coefcient on P_OD_AF suggests that family rms would derive benets by reducing the proportion of shares held by nonfamily board members. It seems that the presence of nonfamily shareholders on the board creates a tension that ultimately results in suboptimal performance. Perhaps the long-term interests of family members compete with the interests of nonfamily directors. In any event, the evidence suggests that family rms would benet by redistributing the shares of nonfamily board members to the CEO. The P_BLOCKF coefcient is positive but marginally signicant. It (weakly) suggests that performance improvements would be experienced by increasing the amount of external monitoring. Taken together with the negative coefcient on the P_OD_AF variable, there appears to be some evidence that performance improvements could be experienced by substituting internal (boardroom) oversight with external monitoring. It should be noted that the two results are

BARTHOLOMEUSZ AND TANEWSKI

261

Table 3 Two-Stage Least Squares Results: Corporate Governance and Firm Performance in Family Firmsa
Qb (1) C P_CEO_AF P_CEO_AF^2 P_OD_AF P_BLOCKF B_POF C_PEQ02F DLF LN(TA) F p-value
a

6.2132 (0.003) 8.1719 (0.030) 1.4051 (0.739) 14.7447 (0.001) 5.5882 (0.094) 1.3909 (0.312) 1.9998 (0.138) 2.0690 (0.001) 0.3447 (0.010) 29.159 (0.000)

In the rst stage, predictions of six endogenous control mechanisms are modeled from the equations specied in the empirical specication section. The six variables are P_CEO_A, P_OD_A, P_BLOCK, B_POF, C_PEQ02 and DL. In the second stage, the predictions are regressed on Q along with Qs specic control variables (LN(TA) and industry dummies). For the sake of brevity, estimates relating to each of the 12 industry dummies (reecting ASX industry codes) have been omitted from the table. Six of the twelve dummies were signicant in the nal model (at the 10 percent level) despite thin coverage of the sample over ASX industry classications. The tests reported in this table were conducted exclusively on family rms (n = 50). The F sufx in the rst column denotes that the variable is the prediction from the rst-stage model specied in the empirical specication section. Q: Tobins ratios.

not necessarily contradictorytaken together, the two results merely suggest that one form of monitoring is more desirable than another. What is not clear is whether more or less monitoring is desirable in aggregate, that is, regardless of the form.7 The insignicance of the coefcient on C_PEQ02F suggests that the fraction of equity-based compensation is set at

optimal levels. This is despite previous evidence suggesting that the proportion of the CEOs equity-based compensation is likely to be lower in family rms than in nonfamily rms. Interestingly, the coefcient on DLF suggests that family rms are suboptimally leveraged: family rms could benet by increasing the amount of debt in their capital structure. Previous research has suggested that

Decreasing independent board representation decreases internal monitoring. Increasing equity held by disinterested blockholders increases external monitoring. The regression results advocate pursuing both concurrently. It is not clear whether overall monitoring would increase or decrease as a consequence.

262

JOURNAL OF SMALL BUSINESS MANAGEMENT

family rms experience a lower cost of debt capital (Anderson, Mansi, and Reeb 2003) than nonfamily rms. Moreover, Australian evidence suggests that family rms are likely to be more highly leveraged than nonfamily rms (Harijono et al. 2004). Despite this, the evidence suggests that family rms do not fully utilize this competitive advantage over nonfamily rms. On the balance, the presence of statistical signicance in the 2SLS regression suggests that family rms do not utilize optimal corporate governance structures. This is somewhat odd given that all the mechanisms examined should be determined endogenously by the rms participants. These results prompt an obvious question: why do nonfamily shareholders hold capital in family rms if family decisions are not strictly value-maximizing? One possible interpretation is that inefciencies in capital markets prevent nonfamily shareholders from selling out. Schulze et al. (2001) interpret their ndings in this manner. An alternative, more plausible, explanation is that family rms are simply more efcient than their nonfamily counterparts. Some of those agency gains are distributed to family members, while the rest accrue to nonfamily shareholders. Suboptimal corporate governance structures will be observed whenever family members share unequally in those gains (that is, more than what nonfamily shareholders holding the same proportion of equity would be entitled to). However, it is still possible that nonfamily shareholders share sufciently in the agency benets of family control such that they are prepared to forgo some of the benets that the family diverts to itself. Under this scenario, all shareholders benet from family control despite the observed deviations from wealth maximization evidence of suboptimality merely corroborates the notion that the agency benets of family control are shared

between family and nonfamily members. If family rms adopted optimal corporate governance structures, they would share any agency efciencies equally with the rest of the shareholders. However, family rms, by virtue of their control, are able to divert a greater share of the gains toward themselves. Shareholders acquiesce, as they are at least better off than they would be in the absence of family control such that the observed deviation from wealth maximization holds in equilibrium. This interpretation is consistent with prior research that suggests that family rms are inherently more efcient than their nonfamily counterparts (for example, Anderson and Reeb 2003). A further possibility is that the samples short-time horizon fails to capture the market in equilibrium.

Discussion and Implications

This paper posits that family control is one of several alternative substitute or ancillary internal control mechanisms, with the potential to either ameliorate or exacerbate the agency problem. It produces evidence that family control interacts with other control mechanisms. Family control displaces other owners: large blockholders are less likely to hold the capital of family rms and board members are less likely to hold shares. As a consequence, family rms are less likely to be subject to the discipline of disinterested or independent monitoring (disinterested or independent in the sense that the monitor is not related to the family). This paper also produces evidence that family rms are likely to have a lower proportion of disinterested or independent directors on their boards than nonfamily rms. Moreover, just as the proportion of outsiders on family boards decreases, so does the proportion of gray directors increase, suggesting that family rms substitute

BARTHOLOMEUSZ AND TANEWSKI

263

outsiders monitoring with interested bystanders. Furthermore, family rms are considerably more likely than nonfamily rms to allow the CEO and the chairperson roles to be occupied by the same person. Together, these ndings suggest that families maintain a close locus of control with little opportunity for external discipline. Further evidence was produced to suggest that the compensation of familyrm CEOs is less sensitive to prior performance than nonfamily-rm CEOs. In contrast, there was some weak evidence to suggest that family-rm compensation might be lower than that of nonfamily rms. These seemingly paradoxical ndings are reconcilable if it is accepted that family CEOs derive income (or, at least, utility) from a broader set of sources than nonfamily CEOs. It would seem that there is substantial evidence to suggest that family rms adopt distinctly different corporate governance structures to nonfamily rms. On the balance, the evidence suggests that family rms adopt corporate governance structures that are inconsistent with maximizing the value of the company. Tests of H2 sought to attack this issue directly by introducing a performance metric (Q) to measure the degree to which family-rm corporate governance structures deviate from optimality in a simultaneous equations framework. The regression results highlighted several suboptima: at lower levels, family-rms CEOs could benet from holding more shares, independent directors hold too many shares, and outside blockholders hold too few. Several plausible alternatives are put forth to explain why nonfamily shareholders remain in family companies despite the suboptimum: (1) capital market inefciencies prevent nonfamily shareholders from selling; (2) family rms are inherently more protable than nonfamily rmsfamilies divert some,

but not all, of these efciencies to themselves such that corporate governance structures are not completely consistent with wealth maximization but everyone is better off so nobody sells; and (3) the sample fails to capture the market in equilibrium. In summary, this paper nds that family rms adopt substantially different corporate governance structures to nonfamily rms. There is some evidence to suggest that these differentials ultimately impact upon rm performance. The practical implications of these ndings indicate that in order to improve rm performance and maximize rm value, family owners need to adopt more transparent corporate governance structures and be subject to greater discipline of independent monitoring. This also implies that family rms operating on the capital market should be sensitive to both shareholders and investors need for greater disclosure and transparency from rms. In the current operating climate in developed countries, shareholders and investors want to be assured that companies are professionally run and have sufcient independence from management. More importantly, both shareholders and investors believe that executive compensation is a key board-performance indicator. Thus, how a family rm remunerates its executives is a litmus test for whether the board as a whole is working in the shareowners interests or not. If a board is not sufciently in charge of the executive-pay process, it may not be sufciently in charge of broader processes, it may be too subservient to management, and it may not have a strong enough grasp of the issues. It is difcult for shareholders and owners to say what is an appropriate amount or method to pay the CEO. Hence, the family rm should rely on its board of directors, and particularly the compensation committee, to do its job in terms of negotiating any packages with executives.

264

JOURNAL OF SMALL BUSINESS MANAGEMENT

References
Agrawal, A., and C. R. Knoeber (1996). Firm Performance and Mechanisms to Control Agency Problems Between Managers and Shareholders, Journal of Financial and Quantitative Analysis 31, 377397. Anderson, R. C., and D. M. Reeb (2003). Founding-Family Ownership and Firm Performance: Evidence from the S&P 500, Journal of Finance 58, 13011328. Anderson, R. C., T. W. Bates, J. M. Bizjak, and M. L. Lemmon (1998). Corporate Governance and Firm Diversication, Working Paper, Kogod School of Business, American University, Washington, DC. Anderson, R. C., S. A. Mansi, and D. M. Reeb (2003). Founding Family Ownership and the Agency Cost of Debt, Journal of Financial Economics 68, 263285. Australian Stock Exchange Limited (ASX). (2003). Principles of Good Corporate Governance and Best Practice Recommendations. Sydney: ASX. Bartholomeusz, S. M. (2002).After Enron: The New Reform Debate, University of New South Wales Law Journal 25, 580593. Bebchuk, L., and J. Fried, (2003). Executive Compensation as an Agency Problem, Journal of Economic Perspectives 17, 7192. Bebchuk, L., and J. Fried (2004). Pay Without Performance: The Unfullled Promise of CEO Compensation. Cambridge, MA: Harvard University Press. Becker, G. S. (1981). A Treatise on the Family. Cambridge, MA: Harvard University Press. Bergstrom, T. C. (1995). On the Evolution of Altruistic Rules for Siblings, American Economic Review 85(5), 5881. Berle, A., and G. Means (1932). The Modern Corporation and Private Property. New York: Macmillan.

Black, F., and M. Scholes (1973). The Pricing of Options and Corporate Liabilities, Journal of Political Economy 81, 399418. Blondel, C., N. Rowell, and L. van der Heyden (2002). Prevalence of Patrimonial Firms on Paris Stock Exchange: Analysis of the Top 250 Companies in 1993 and 1998, INSEAD Working Paper Series, INSEAD, Fontainbleau, France, July. Booz Allen Hamilton and Business Council of Australia (2003). CEO Turnover in 2002: Trends, Causes and Lessons Learned, Research report, Melbourne, Australia. Buchanan, J. M. (1975). The Samaritans Dilemma. Altruism, Morality and Economic Theory, ed. E. S. Phelps. New York: Russell Sage Foundation. Cadbury, A. (1992), The Financial Aspects of Corporate Governance. London: The Committee on Financial Aspects of Corporate Governance. Chami, R. (1999). Whats different about family business? Working Paper Series 6769, University of Notre Dame and the International Monetary Fund, Notre Dame, IN. Coase, R. H. (1937). The Nature of the Firm, Economica 4, 331351. Commonwealth of Australia (2002). Corporate Disclosure: Strengthening the Financial Reporting Framework. Canberra: Australian Government Publishing Service. Commonwealth of Australia (2003). Audit Reform and Corporate Disclosure: Corporate Law Economic Reform Program (CLERPG). Canberra: Australian Government Publishing Service. Demsetz, H., and K. Lehn (1985). The Structure of Corporate Ownership: Causes and Consequences, Journal of Political Economy 93, 11551177. Eschel, I., L. Samuelson, and A. Shaked (1998). Altruists, Egoists and Hooligans in a Local Interaction Model, American Economic Review 88(1), 157179.

BARTHOLOMEUSZ AND TANEWSKI

265

Harijono, M. Ariff, and G. A. Tanewski (2004). The Impact of Family Control of Firms on Leverage: Australian Evidence, Unpublished EMF Classication Code 140, 131. Hermalin, B. E., and M. S.Weisbach (2003). Boards of Directors As an Endogenously Determined Institution: A Survey of the Economic Literature, Economic Policy Review, Federal Reserve Bank of New York 9, 726. Higgs, Derek (2003). Review of the Role and Effectiveness of Non-executive Directors. London: The Department of Trade and Industry. Investment and Financial Services Association (1999). A Guide for Investment Managers and Corporations (The Blue Book). Sydney, Australia: Australian Investment Managers Association. James, H. (1999). Owner As Manager: Extended Horizons and the Family Firm, International Journal of the Economics of Business 6, 4156. Jensen, M. C. (1979). Toward a Theory of the Press: Economics and Social Institutions, Edited by Karl Brunner. Leiden, the Netherlands: Martinus Nijhoff Publishing Company. Jensen, M. C., and W. H. Meckling (1976). Theory of the Firm: Managerial Behaviour, Agency Costs and Ownership Structure, Journal of Financial Economics 3, 305360. Klein, A. (1998). Firm Performance and Board Commitee Structure, Journal of Law and Economics April, 275303. Klein, S. B., and C. Blondel (2002). Ownership Structure of the 250 Largest Listed Companies in Germany, Working Paper No. 2002/123 IIFE, INSEAD, Fontainebleau, France. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. W. Vishny (1998). Law and Finance, Journal of Political Economy 106, 11131155. (1999). Corporate Ownership around the World, Journal of Finance 54, 471518.

Morck, R., and B. Yeung (2003). Agency Problems in Large Family Business Groups, Entrepreneurship Theory and Practice Summer, 367382. Morck, R., A. Shleifer, and R. Vishny (1988). Management Ownership and Market Valuation: An Empirical Analysis, Journal of Financial Economics 20, 293316. Morck, R., D. A. Stangeland, and B. Yeung (2000). Inherited Wealth, Corporate Control and Economic Growth: The Canadian Disease? in Concentrated Corporate Ownership. Ed. R. Morck. Chicago, IL: The University of Chicago Press, pp. 319369. Mroczkowski, N. A., and G. A. Tanewski (2006). Delineating Publicly Listed Family and Non-family Controlled Firms: An Approach for Capital Market Research in Australia, Journal of Small Business Management (in press). Perez-Gonzalez, F. (2002). Inherited Control and Firm Performance, SSRN Working Paper, July, http://ssrn.com/ abstract-320888. Schulze, W. S., M. H. Lubatkin, R. N. Dino, and A. K. Bucholtz (2001). Agency Relationships in Family Firms: Theory and Evidence, Organization Science 12, 99116. Simon, H. A. (1993). Altruism and Economics, American Economic Review 83, 156161. Smith, A. (1981). An Inquiry in to the Nature and Causes of the Wealth of Nations. Indianapolis, IN: Liberty Press. Tufano, P. (1996). Who Manages Risk? An Empirical Examination of Risk Management Practices in the Gold Mining Industry, Journal of Finance 51, 10971137. Weisbach, M. S. (1988). Outside Directors and CEO Turnover, Journal of Financial Economics 20, 431460. White, H. (1980). HeteroskedasticityConsistent Covariance Matrix and a Direct Test for Heteroskedasticity, Econometrica 48, 817838.

266

JOURNAL OF SMALL BUSINESS MANAGEMENT

Appendix A Variable Labels and Denitions

Ownership Variables P_CEO_A P_CEO_D P_CEO_I P_OD_A P_OD_D P_OD_I P_BLOCK Proportion of ordinary shares directly or indirectly owned by the CEO Proportion of ordinary shares directly owned by the CEO Proportion of ordinary shares indirectly owned by the CEO Proportion of ordinary shares directly or indirectly owned by outside directors Proportion of ordinary shares directly owned by outside directors Proportion of ordinary shares indirectly owned by outside directors Proportion of ordinary shares directly or indirectly owned by nonafliated shareholders with more than 5 percent share.

Board Composition Variables B_DUAL B_PI B_PO B_PG B_SIZE A dummy variable equal to 1 when the chairperson is the CEO; 0 otherwise The proportion of insiders on the board The proportion of outsiders on the board The proportion of grey members (that is, afliated nonexecutives) on the board The number of members on the board

Compensation Variables C_FIXINC01 C_FIXINC02 C_OPTINC01 The CEOs total xed income, the aggregate of salary, superannuation and bonus in 2001 The CEOs total xed income, the aggregate of salary, superannuation and bonus in 2002 The value of the CEOs option portfolio at the start of the year, plus the value of any options awarded throughout the year, plus the value of any prots associated with exercise in 2001 The value of the CEOs option portfolio at the start of the year, plus the value of any options awarded throughout the year, plus the value of any prots associated with exercise in 2002 The CEOs total xed income plus the value of the option portfolio in 2001 The CEOs total xed income plus the value of the option portfolio in 2002 The proportion of compensation that is equity based (C_OPTINC / [C_FIXINC + C_OPTINC]) in 2001 The proportion of compensation that is equity based (C_OPTINC / [C_FIXINC + C_OPTINC]) in 2002

C_OPTINC02

C_TOTAL01 C_TOTAL02 C_PEQ01 C_PEQ02

BARTHOLOMEUSZ AND TANEWSKI

267