Ownership structure, control and firm performance: the effects of vote-differentiated shares

Abstract

This article contributes to the literature on ownership, control and performance by exploring these relationships for Swedish listed companies (1997–2002). We find that firms, on average, are making inferior investment decisions and that the use of dual-class shares have a negative effect on performance. Marginal q is used as a measure of economic performance. It was presented in an article by Mueller and Reardon in 1993 and has recently been used in empirical studies of ownership and performance by, among others, Gugler and Yurtoglu (2003). Frequently Tobin's q is used in studies of this type, but Tobin's q has a number of disadvantages which can be circumvented by employing a marginal q. This study adds to earlier studies by investigating how the separation of vote and capital shares’ creates a wedge between the incentives and the ability to pursue value-maximization. The relationships between the performance and different ownership characteristics like ownership concentration and foreign ownership are also investigated.

Acknowledgments

Financial support from sparbankernas Forskningsstiftelse of Daniel Wiberg's and Johan Eklund's dissertation work is gratefully acknowledged. A research grant from the Ratio Institute and the Marcus and Amalia Wallenberg Memorial Fund Foundation is also gratefully acknowledged. Furthermore, we are grateful towards the participants at the ISNIE conference 2004 and the International Workshop on Corporate Governance and Investment 2005 for valuable comments. A special thanks also to Åke E. Anderson, Dennis C. Mueller, Evgeni Peev and Niclas Berggren.

Notes

¹ A similar interpretation of the behaviour of investors according to Hirschman's classification can be found in Hedlund (1984).

² Sundin, A. and Sundqvist S.-I. (1998–2002).

³ Note that when differentiating with respect to investments, I_t , the deprecation rate, δ, disappear, and hence have no relevance for the interpretation of q_m .

⁴ The number is attained by plugging in the regression coefficient estimates from the fixed effects model with respect to the capital share regression (A).

⁵ Separate regressions were estimated when the sample firms were divided into two subsets, with or without vote-differential. These regressions supported the findings in Tables 3 and 4. As the results were generally the same, these regressions are excluded but are available upon request; suffice it to say that the results in Tables 3 and 4 in this way are confirmed.