Learning Approximately Optimal Contracts

Alon Cohen, Argyrios Deligkas, Moran Koren

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In principal-agent models, a principal offers a contract to an agent to preform a certain task. The agent exerts a level of effort that maximizes her utility. The principal is oblivious to the agent’s chosen level of effort, and conditions her wage only on possible outcomes. In this work, we consider a model in which the principal is unaware of the agent’s utility and action space: she sequentially offers contracts to identical agents, and observes the resulting outcomes. We present an algorithm for learning the optimal contract under mild assumptions. We bound the number of samples needed for the principal obtain a contract that is within ϵ of her optimal net profit for every ϵ> 0. Our results are robust even when considering risk averse agents. Furthermore, we show that when there only two possible outcomes, or the agent is risk neutral, the algorithm’s outcome approximates the optimal contract described in the classical theory.

Original languageEnglish
Title of host publicationAlgorithmic Game Theory - 15th International Symposium, SAGT 2022, Proceedings
EditorsPanagiotis Kanellopoulos, Maria Kyropoulou, Alexandros Voudouris
PublisherSpringer Science and Business Media Deutschland GmbH
Pages331-346
Number of pages16
ISBN (Print)9783031157134
DOIs
StatePublished - 1 Jan 2022
Externally publishedYes
Event15th International Symposium on Algorithmic Game Theory, SAGT 2022 - Colchester, United Kingdom
Duration: 12 Sep 202215 Sep 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13584 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th International Symposium on Algorithmic Game Theory, SAGT 2022
Country/TerritoryUnited Kingdom
CityColchester
Period12/09/2215/09/22

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science (all)

Fingerprint

Dive into the research topics of 'Learning Approximately Optimal Contracts'. Together they form a unique fingerprint.

Cite this