Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
A performance measure may or may not reflect the relative importance of different tasks for the production of benefit: it can be aligned or unaligned. Here, I examine when using an aligned measure generates a larger surplus in a principal-agent relationship than using an unaligned but otherwise identical measure. I find that (i) the agent's effort costs matter for the optimal way of measuring performance, and (ii) the optimal measure is not aligned but tilted toward tasks that the agent finds easy. Failing to recognize these insights may lead to false predictions about the use of incentives. (c) 2008 by The University of Chicago.