Research

Deferred Acceptance (DA) mechanisms are widely implemented in centralised matching markets, where the proposing side is deliberately chosen to ensure strategy-proofness for proposers and the proposer-optimal stable matching. In practice, however, pre-match communication from the receiving side indicating strong interest — such as “ranked-to-match” messages in the NRMP — is prevalent. In a laboratory experiment with candidate–firm matching and candidate-proposing DA, we find that firms strategically target messages toward their firm-optimal stable matching partners, while candidates respond by ranking messaging firms higher. Together, these behaviours shift market outcomes away from the candidate-optimal and toward the firm-optimal stable matching, with statistically significant effects when messages are committed, meaning that firms must rank the message recipient first. Our findings identify pre-match communication as a novel form of manipulation outside the matching mechanism and show that it can undermine the objectives that market designers seek to achieve.

Selection mechanisms that rank candidates on exams, honours, or research output sort only on what they observe, yet job competence often has dimensions these signals miss. I study this trade-off in a matching tournament: high- and low-ability candidates split a fixed effort budget between observable signalling and unobserved soft-skill investment, and are hired assortatively by high- and low-quality firms on a noisy signal of that effort. A precise exam sorts more accurately but invites over-investment in signalling; a lottery removes that over-investment but sorts randomly, and a coarse exam approaches the lottery as noise increases. I show a precise exam raises welfare when the ability gap between candidates is large relative to the quality difference between firms; when the reverse holds, increasing exam noise raises welfare, and a lottery achieves the highest welfare in the limit. I test these predictions in a laboratory experiment that varies exam noise and the ability gap, and ask whether observed effort and welfare track the theoretical reversal or deviate from it, and if so, why.

In design stage

Students from low-socioeconomic (low-SES) backgrounds often face substantial information frictions in school choice. This paper examines the welfare implications of two types of information interventions — providing students with common information (CI) and group-specific information (GI) — in a many-to-one matching model with a continuum of students and a finite set of schools. We show that while CI and GI perform similarly when adopted on a small scale, GI strictly outperforms CI when information uptake is high. At scale, CI generates congestion externalities that reduce the benefit to informed low-SES students and hurt both high-SES and uninformed low-SES students. GI eliminates this congestion. These findings also hold when the district implements affirmative action policies.

This project studies how information about university financial aid shapes the educational decisions of high-school students in Victoria, Australia. We designed a tailored Financial Aid Calculator to deliver personalised information and used it to document — and address — the information gap facing rural Australian youth. The Honours thesis underpinning this project was awarded first place in the Deakin Essay Competition (2022) and received the highest score in the Economics Honours cohort.