Estimation of a Non-Parametric Principal-Agent Model with Hidden Actions
69 Pages Posted: 6 Aug 2019 Last revised: 28 Mar 2021
Date Written: August 2, 2019
The design of performance-based incentives -- commonly used in online labor platforms -- is naturally posed as a moral-hazard principal-agent problem. A key input in this setting is the dependence of agent output on effort, which is unlikely to be known in practice. In this paper, we consider the estimation of a principal-agent model, with a focus on the parameters that link agent output and effort. We first present an estimator for a non-parametric agent model and show it to be statistically consistent. To circumvent computational challenges with solving the estimation problem exactly, we approximate it as an integer program, which we solve through a column generation algorithm that uses hypothesis tests to select variables. We show that our approximation scheme and solution technique both preserve the estimator's consistency and combine to dramatically reduce the computational time required to obtain competitive parameter estimates. To demonstrate our approach, we conducted a randomized experiment on a crowdwork platform (Amazon Mechanical Turk), where incentive contracts were randomly assigned among a pool of workers completing the same task. We present numerical results illustrating how the proposed estimator combined with experimentation can shed light on the efficacy of performance-based incentives.
Keywords: principal-agent problems, performance-based incentives, estimation, integer programming, crowdwork platforms
Suggested Citation: Suggested Citation