I've heard an economist suggest that instead of tying executive bonuses to the company's own performance, they instead compare it to the performance of other, similar firms.

Still a fuzzy measure, but it would at least give them better incentives.