There is a debate on whether executive pay reflects rent extraction due to "managerial power" or is the result of arms-length bargaining in a principal-agent framework. In this paper we offer a test of the managerial power hypothesis by empirically examining the CEO compensation of U.S. public companies that were ever in financial distress between 1992 and 2005. Using a bias-corrected matching estimator that estimates the causal effects of financial distress, we find that, for the distressed firms, CEO turnover rates increase markedly and their CEOs, both incumbents and successors, experience significant reductions in total compensation. The bulk of the reduction in total compensation derives from the decline in value of stock option grants, which we argue is due to a change in the opportunistic timing of option grants. We define "lucky" grants as those with grant prices below or at the lowest stock price of the grant month, and we find that the proportion of lucky grants for financially distressed firms is higher before insolvency and lower upon and after insolvency, while the proportion for similar but solvent firms remains stable throughout the period. We interpret this evidence as consistent with a decrease in managerial power induced by a tightening in the "outrage" constraint due to the episode of financial distress.