English

Interviewing Matching in Random Markets

Theoretical Economics 2023-09-07 v2

Abstract

In many centralized labor markets candidates interview with potential employers before matches are formed through a clearinghouse One prominent example is the market for medical residencies and fellowships, which in recent years has had a large increase in the number of interviews. There have been numerous efforts to reduce the cost of interviewing in these markets using a variety of signalling mechanisms, however, the theoretical properties of these mechanisms have not been systematically studied in models with rich preferences. In this paper we give theoretical guarantees for a variety of mechanisms, finding that these mechanisms must properly balance competition. We consider a random market in which agents' latent preferences are based on observed qualities, personal taste and (ex post) interview shocks and assume that following an interview mechanism a final stable match is generated with respect to preferences over interview partners. We study a novel many-to-many interview match mechanism to coordinate interviews and that with relatively few interviews, when suitably designed, the interview match yields desirable properties. We find that under the interview matching mechanism with a limit of kk interviews per candidate and per position, the fraction of positions that are unfilled vanishes quickly with kk. Moreover the ex post efficiency grows rapidly with kk, and reporting sincere pre-interview preferences to this mechanism is an ϵ\epsilon-Bayes Nash equilibrium. Finally, we compare the performance of the interview match to other signalling and coordination mechanisms from the literature.

Keywords

Cite

@article{arxiv.2305.11350,
  title  = {Interviewing Matching in Random Markets},
  author = {Maxwell Allman and Itai Ashlagi},
  journal= {arXiv preprint arXiv:2305.11350},
  year   = {2023}
}

Comments

Error in proof of Lemma 1 leaves a step unexplained, proof needs to be corrected and made more rigorous

R2 v1 2026-06-28T10:38:46.743Z