Does one hat fit all? The case of corporate leadership structure

Source: The Corporate Library, on the Internet at www.thecorporatelibrary.com.

It is reasonable to argue that complex organizations derive significant benefits from non-duality, since such organizations are inherently more difficult for shareholders to understand and monitor. While acknowledging this possibility, the organizational complexity hypothesis postulates that, for complex firms, the incremental benefits of non-duality (over constraints imposed by labor and product markets) are dominated by the cost of lost CEO flexibility and the greater potential for distortions in transmitting information between the CEO and board chairman. A related argument is that complexity, as the number, diversity and interconnectedness of tasks and units, generates economies of specialization, thereby favoring the division of labor. However, since a non-executive chairman, by definition, does not perform executive functions, it is not clear that complexity in this sense favors separation of CEO and chairman duties.

An alternative proxy is the ratio of intangible assets to total assets, with the ratio being presumed higher for complex organizations. However, most firms have missing values for intangible assets in Compustat. Hence, it is not feasible to use this proxy.

An alternative possibility is that the CEO becomes entrenched at higher ownership levels and takes on the additional powers of board chairman. Unfortunately, it is impossible to differentiate between these two effects in the context of this paper, that is, one cannot tell whether higher ownership positively affects the probability of CEO duality because of incentive alignment or CEO entrenchment.

This point is emphasized by General Motors, ExxonMobil, Verizon Communications, and several other firms in opposing shareholder proposals recommending separation of CEO and chairman responsibilities.

These publications include Wall Street Journal, Financial Times, BusinessWeek, Fortune, Forbes, Newsweek, Time, Washington Post, New York Times, and USA Today. See Factiva for a complete listing. Press appearances range from 0 to 2,877, with mean and median values of 26.5 and 7, respectively. Thus, to reduce potential problems with outliers, I follow Milbourn (2003) and use standardize article counts (based on the empirical cumulative density function of press appearances) in my regressions.

As previously acknowledged, this result is also consistent with an entrenchment argument.

Following Morck et al. (1988), the empirical corporate finance literature typically uses breakpoints to control for insider ownership. I employ the same breakpoints as in Morck et al. (1988), i.e., ownership levels of less than 5%, between 5% and 25%, and greater than 25%.