A Game Theoretic Approach to Peer Review of Grant Proposals

Abstract

This paper studies the grant peer review process employed by the Turkish regional development agencies, which is adapted from a review procedure of the Education, Audiovisual and Culture Executive Agency of the European Union. To model this process, we consider a Bayesian strategic-form game played by three reviewers who observe both a common and a private score signal about an evaluated project and assign their scores to minimize the sum of their disutilities from the false acceptance and false rejection of the project. We numerically compute the Bayesian Nash equilibria of this game and conduct several comparative statics exercises, after calibrating the model parameters accordingly. We also introduce two simpler review processes and compare their performances to that of the calibrated process in terms of outcome statistics, involving pass and fail rates of the evaluated projects, and manipulation statistics, involving the reviewers' manipulation rate and size of scores.