A Bias of Screening

Abstract

This paper deals with the issue of screening. It focuses on a decision maker who, based on noisy unbiased assessments, screens elements from a general set. Our analysis shows that stricter screening not only reduces the number of accepted elements, but possibly reduces their average expected value. We provide a characterization for optimal threshold strategies for screening and also derive implications to cases where such screening strategies are suboptimal. We further provide various applications of our results to credit ratings, auctions, general trade, the Peter Principle, and affirmative action. (JEL C38, D44, F10, G24, J15)

Dynamic screening

“…This type of noises (and others) are prone to screening biases; seeLagziel and Lehrer (2019) for more details.…”

Reversals of signal-posterior monotonicity imply a bias of screening

“…1 Introduction Lagziel and Lehrer (2019) (LL) prove that for any bounded random variable of interest, there exists a noisy signal such that the expectation of the variable conditional the signal exceeding a lower cutoff is larger than conditional on the signal passing a higher cutoff. Chambers and Healy (2011) (CH) show that for any bounded support, there exists a signal such that for any random variable on that support, the distribution of the variable conditional on a lower signal realization first order stochastically dominates (FOSDs) the distribution conditional on a higher realization.…”