Dataset for "Matching with batches: Recovering strategy-proofness in a cutoff-based mechanism"

Description

Abstract of associated article: Most centralised admissions systems for high schools and universities constrain applicants to submitting ranked preference lists that are shorter than the set of available alternatives, as complete lists would be impractical. This renders the matching mechanism non-strategy-proof: applicants lack the incentive to rank their preferences truthfully. Under these conditions, clearinghouses cannot offer simple, transparent advice. We propose batching, the sequential submission of short lists until the market clears, to restore strategy-proofness. This approach best suits systems based on entry cutoffs rather than quotas, such as the Universities Admissions Centre (UAC) in Australia. We evaluate batching in an individual decision-making experiment, comparing it to the status quo (constrained lists) and to unconstrained lists, without and with advice. Participants in batched and unconstrained treatments achieve significantly higher payoffs than those in constrained treatments, with average efficiency gains of around 60%. We find no significant efficiency gains attributable to advice.

Steps to reproduce

* Dataset.dta contains the Stata dataset of the experiment. * Analysis.do performs all analyses reported in the paper. The user-contributed outreg2 module is required to create the formatted regression output in Table 4.

Institutions

• The University of Sydney

  New South Wales, Sydney

Categories

Licence