Approximate Strategyproofness in Large, Two-Sided Matching Markets

TL;DR

This paper investigates how large two-sided matching markets approximate strategyproofness, showing that as markets grow, agents have fewer beneficial deviations, especially with shorter preference lists and higher preference correlation.

Contribution

It provides empirical evidence that large markets naturally approximate strategyproofness, supporting the continued use of the deferred acceptance algorithm despite its lack of true strategyproofness.

Findings

Deviations decrease as market size increases

Shorter preference lists improve approximation

Higher preference correlation enhances strategyproofness

Abstract

An approximation of strategyproofness in large, two-sided matching markets is highly evident. Through simulations, one can observe that the percentage of agents with useful deviations decreases as the market size grows. Furthermore, there seems to be a strong connection between the length of preference order lists, the correlation of agent preferences, and the approximation of strategyproofness. Interestingly, approximate strategyproofness is reached easier with a shorter length of preference orders and higher preference correlation. These findings justify the use of the deferred acceptance algorithm in large two-sided matching markets despite it not being strategy-proof.