Incentive contracts are not rigged by powerful CEOs

摘要

Morse, Nanda and Seru (2011) interpret the data to suggest that more powerful CEOs expost change their incentive contracts more. My paper points out a number of issues with their inference. First and most importantly, MNS do not control for the fact that not just the most powerful but almost all firms change their incentive contracts ex-post. This is also consistent with an optimal contracting model. Nevertheless, the MNS specification attributes all explanatory power of the average incentive realignment to the cross-coefficient, i.e., to more powerful CEOs. When the average level of ex-post contract change is also controlled for, the MNS cross-coefficient (i.e., the additional change attributed to more powerful CEOs) declines by 55% and becomes insignificant. Second, newly-hired CEOs often receive large one-time startup packages. These firms should be broken out, because the MNS theory is not about newly-hired CEOsnew CEOs could not have rigged a previously-set compensation. Third, the results are sensitive to how industry performance is adjusted for. Fourth, the results are sensitive to the level of winsorization.