The Instability of Matching with Overconfident Agents : Laboratory and Field Investigations

Many centralized college admissions markets allocate seats to students based on their performance on a standardized exam. A single exam's measurement error can cause the exam-based priorities to deviate from colleges' aptitude-based preferences. Previous literature proposes correcting such an error by combining pre-exam preference submission with a Boston algorithm. In a laboratory experiment, I find that pre-exam preference submission is skewed by overconfidence, which leads to even more severe and more varied welfare distortions than the exam's measurement error alone. Moreover, the mechanism exhibits a tendency to reward overconfidence and punish underconfidence, thus serving as a gender penalty for women. I also analyze field data from China and reach similar conclusions