The ownership-performance puzzle: agency issues in small and medium-sized family firms

Abstract

1. The purpose / topic of the research

During the last decade, the discussion about the relationship between ownership structure and performance has seen important evolutions on a theoretical as well as empirical level. Especially family ownership in public firms drew the attention of researchers. In this paper, we explore the relationship between family ownership and financial performance for small and medium-sized private family firms and extend the results by recent studies of Anderson and Reeb (2003) and Villalonga and Amit (2006). In line with Villalonga and Amit (2006), we take into account several governance characteristics including family ownership and family management. More specific, our analysis examines the moderating influence of active family control through the CEO position (Andersen and Reeb, 2003) as well as generational influences (Villalonga and Amit, 2006).

2. The research method

In this study, we use a cross sectional sample of 2,865 family firms from the 1998 National Survey of Small Business Finance (NSSBF). We measure the financial performance by the return on assets before taxes due to the fact that our database includes several organisational forms of which only C-corporations pay corporate income taxes. In order to compare the performance for all firms on an equal basis, we use in this study ROA before deduction of corporate income taxes (Ang et al., 2000). Because we concentrate on several interaction effects between the explanatory variables, we estimate several chained-interaction models (Kam and Franzese, 2005) in which we include the interaction between ownership dispersion proxies with generation and management variables. Even though interaction models are quite common in different disciplines of research, the interpretation of these models is often flawed and inferential errors are common (Kam and Franzese, 2005). By means of our study we want to contribute to the performance-governance literature avoiding any of these misinterpretations. In this study, we measure the standard linear-interactive models using OLS. The robustness of the results is checked by using the industry adjusted ROA before taxes as well as applying the model to the subsample of C-corporations (who do pay corporate taxes).

3. The theories used

In the academic literature, divergent views exist about the impact of family ownership and control on agency costs stemming from the separation of ownership and control. At one extreme, agency models (e.g. Jensen and Meckling, 1976; Ang et al. 2000; Anderson and Reeb, 2003) assume that the effects of concentrated ownership and owner-management will lead to a minimized or even zero level of agency costs. Family managers owning a large part of the firm bear nearly all of the costs and benefits of their actions. Self interest by family agents is assumed being tempered by kinship and parental altruism (Schulze et al., 2003). However, once a family has enough ownership for unchallenged control, it can begin to abuse its power by taking resources out of the business (Claessens et al., 2002). In this case, a major owner may serve as a poor de facto ‘agent’ for the minority owners: the large shareholder may use its controlling position in the firm to extract private benefits at the expense of the small shareholders (Morck and Yeung, 2003; Villalonga and Amit, 2006). Some researchers have argued that family-dominated businesses are more apt to be characterized by extraordinary dividend payouts, entrenched managers and a redistribution of wealth from employees to the family (Burkart et al., 2003). Other authors argue that restricting executive talent to a labor pool of family members can be problematic. In our paper, we also investigate the effect of generational evolution on performance. Generational evolution and ownership dispersion seem to be often entwined. From a theoretical point of view, Schulze et al. (2003) discuss the agency consequences when equity in a family firm is distributed among family shareholders instead of among outside shareholders. They argue that dispersed family ownership in the controlling owner stage and in the sibling partnership stage (Lubatkin et al., 2005) could result in free riding of family insiders. This free riding problem would decrease in the cousin consortium stage as more outside family members (not employed by the firm) become shareholder and hence, behave more as rational diversified investors.

4. The contribution of the research

In this paper, we extend the empirically debate (1) by concentrating principally on private family firms which allows us to include the underresearched zero agency-cost base case (100% owner-manager), (2) by examining the role of several management and governance characteristics as moderating variables on the ownership dispersion/performance relationship and (3) by examining the extent and nature of specific agency problems in family firms. Our results suggest that ownership dispersion has a positive influence on performance when the family firm is in the “cousin consortium” generational stage compared with the ”controlling owner” and “sibling partnership” stage. This supports the theoretical concepts of Schulze et al. (2003) concerning the agency costs and negative consequences of altruism over different generational ownership stages. Agency costs in private family firms seem to be higher than previously has been proposed by traditional agency models (e.g. Jensen and Meckling, 1976). Our results concerning the relationship between family managers and performance also suggest that the zero agency-cost base case is in fact no zero agency cost case due to ignored agency costs such as free riding by children.