Xinrong Liu

University of Texas at Austin · Economics · 2026–27 job market

Fields

Theory

Job market paper

What to Measure: Optimal Monitoring under Adverse Selection and Moral Hazard

A principal often chooses not only how much information to collect, but also what to measure. I study this choice in a contracting problem with both hidden type and hidden action, where an existing performance measure is informative about both dimensions and the principal can design an additional, scarce monitoring signal. I show that the optimal measurement need not target the dimension that is statistically least informative. Instead, it depends jointly on the incentive requirements of the contract and the information already contained in existing performance. These forces are summarized by an information-adjusted incentive requirement: the local incentive requirements premultiplied by the inverse Fisher information matrix of the existing measure. The optimal direction of scarce monitoring is the leading eigenvector of the second-moment matrix of these adjusted requirements. Existing information can therefore make it optimal for a new signal to respond to type and action with opposite signs, even when both underlying incentive requirements are positive. Under strict local and global implementation margins, the same direction governs the fully incentive-compatible problem. The results distinguish statistical informativeness from contracting value and characterize what a principal should measure first when monitoring capacity is scarce.