eThesis - School of Business electronic theses

» hse_ethesis_12497.pdf   size:702 KB (717856)

This thesis tries to answer the question whether ownership structure matters for firm performance. The starting point for the analysis is the agency theory by Jensen and Meckling (1976), which predicts that higher levels of managerial ownership structure increase firm performance due to an incentive effect. Other authors have in turn suggested that large outside owners might have a role to play as monitors of the management and might thus enhance performance (Shlefer and Vishny 1986 and Zeckhauser and Pound 1990). On the other hand, the private benefits literature (Barclay and Holderness 1989 and Bebchuk 1999) suggests that high ownership concentration may lead to the extraction of the firm’s resources by the dominant owners at the expense of other shareholders. The empirical part of the thesis tests these two hypotheses, which can be combined if we allow the effect of ownership to be nonlinear.

A large amount of empirical research has been published on the subject. Demsetz and Lehn (1985) were the first to estimate the effect of ownership concentration on firm performance and they found no relationship. Important research has also been published by Himmelberg et al. (1999), who estimate a fixed effects model on firm performance and insider ownership and also fail to find a relationship. Studies using non-U.S. data have, however, found a positive relationship even when controlling for endogeneity. Controlling for the endogeneity of ownership is important both in the light of the empirical results and the theoretical discussion.

The thesis uses a panel dataset of Finnish listed companies over the years 2007-2009 to estimate a fixed effects model similar to Himmelberg et al (1999). In addition to control variables the model includes variables for insider ownership, ownership concentration and managerial compensation. The results from OLS regressions are largely in line with agency theory predictions, but statistical significance dissappears when controlling the endogeneity of ownership by two-stage least squares estimation. Upon closer inspection this turns out to be caused by the weakness of the used instruments. We then face the dilemma of choosing between OLS estimates biased due to endogeneity and 2SLS estimates biased due to weak instruments.