Does CEO Duality Matter: An Integrative Approach

Some firms allow their CEO to hold the position of Chair of the Board of Directors while other firms choose to split those two positions between two different individuals. This dissertation first examines whether agency control mechanisms, agency problems, and other firm characteristics are related to the observed choice of one or two individuals in the two positions. The empirical research is based on the hypothesis that having the positions split between two individuals is a means of controlling agency problems when used in concert with (either as a substitute for or as a complement to) other mechanisms for controlling agency problems. Firms are thus viewed as evaluating the costs and benefits of split positions as well as of other agency control mechanisms. They choose the most cost effective means of addressing the problems they face. If split positions are more cost effective, then the firm should choose to split positions other things equal. The very high predictive power of the estimated logistic model confirms the hypothesis that the probability of choosing split positions is related to control mechanisms and agency problems as well as to size and other factors. Some agency control mechanisms perform as complementary agency control mechanisms and some as substitutes for split positions. The results suggest that firms with higher agency costs of debt and equity are more likely to have chosen to split positions. The results are thus consistent with the view of the choice of split positions as a means of managing agency problems in concert with other mechanisms in an integrative decision framework.

The second part of the dissertation examines the linkage between the CEO-Chair choice and the performance of the company (as measured by returns or operating efficiency). Shareholder activists argue that poor performance results when the CEO serves simultaneously as the Chair of the Board of Directors, so called "CEO duality." This dissertation examines whether firms that split these positions experience higher accounting performance than firms that do not split the positions. Several hypotheses are tested regarding the effects on firms with split positions. The empirical model indicates that firms that have split positions exhibit, on average, no lower or higher performance than other firms after integrating into the model industry effects, the role of other agency control mechanisms, the size of agency problems, and other firm characteristics. This is consistent with an irrelevance hypothesis as well as with the possibility that firms choose their policies optimally once other factors are accounted for. However, the firms with split positions do exhibit a different relationship between information asymmetry and performance as well as between other agency control mechanisms and performance. The use of agency control mechanisms, for example as measured by the proportion of the firm held by institutional investors, have a greater effect on performance for non-split than for split firms.

Overall the results support the notion that firm and management characteristics (such as the level of agency problems, information asymmetry, ownership structure, and the existence of other agency control mechanisms) influence the choice to split positions and influence the role and effectiveness of split positions. The vast majority of firms' choice can be predicted by using such characteristics in an integrated model of the decision. The results imply that the benefits of split positions may be firm specific, that split positions are only appropriate for some firms, and that a net benefit will not be captured by all firms that simply enact a policy of split positions independent of their fundamental characteristics.