Atila AbdulkadIroglu, Joshua D. Angrist, Yusuke Narita, Parag A. Pathak, and Roman A. Zarate
Many school and college admission systems use centralized mechanisms to allocate seats based on applicant preferences and school priorities. When tie-breaking uses non-randomly assigned criteria like distance or a test score, applicants with the same preferences and priorities are not directly comparable. The non-lottery setting does generate a kind of local random assignment that opens the door to regression discontinuity designs. This paper introduces a hybrid RD/propensity score empirical strategy that exploits quasi-experiments embedded in serial dictatorship, a mechanism widely used for college and selective K-12 school admissions. Using this approach, the research estimates achievement effects of Chicago’s exam schools.
Subscribe for Updates