Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
A firm's payoffs can often depend on whether it exceeds a threshold of performance. In a principal-agent setting, we demonstrate that the optimal contract induces efficient effort by basing firing decisions solely on whether the agent was successful, even though the absolute level of performance is more informative. We test the predictions of our model using thirty seasons of NFL data. We find that a coach is 6 percentage points more likely to be fired after a close loss; beyond this, the margin of victory has no significant impact on firing probabilities despite being more informative about future success. Also consistent with the predictions of our model, coaches maximize team effort (by playing their best players) when doing so has the largest impact on the probability of winning.