I think this is the most interesting excerpt from the study:
"The level of incentive compensation is significantly negatively related to the forward ROA, while the level of cash compensation is positively related to the level of ROA.
Overall, we conclude that our results seem most consistent with the hypothesis that overconfident CEOs accept large amounts of incentive pay and consequently engage in value destroying activities that translate into future reductions in returns and firm performance. "
I wonder about that "overconfident" label, and it might be possible to tease various effects out of the data.
Consider two possibilities:
1) CEOs with large incentive pay are irrationally confident in the success of risky policies
2) CEOs with large incentive pay are behaving like economically rational agents... betting someone else's money.
That is, incentive pay is more like a free pass to a casino where your losses are made good but you keep some fraction of your winnings. Under those circumstances taking large risks may be the most rational thing to do, IF the cash component of the CEO's compensation is sufficient to keep them in caviar and summer homes while they gamble the firm's money away.
Your second possibility covers not only executives who partake in high-risk gambles but also executives who deliberately bankrupt their company for personal profit. A canny executive doesn't rob the firm by dipping his fingers in the till but by, for instance, loading the firm with gold-plated rubbish and extracting a personal profit in the form of compensation, stock-options etc, before the sheen wears off.
Small problem: the incentive compensation is mostly options, and the way that package is measured is based on the value of the options at that moment in time. Options with high valuations are effectively pay for past performance that is consequently now worth something.
"The level of incentive compensation is significantly negatively related to the forward ROA, while the level of cash compensation is positively related to the level of ROA.
Overall, we conclude that our results seem most consistent with the hypothesis that overconfident CEOs accept large amounts of incentive pay and consequently engage in value destroying activities that translate into future reductions in returns and firm performance. "