Working Paper: NBER ID: w23265
Authors: Atila Abdulkadiroglu; Yeonkoo Che; Parag A. Pathak; Alvin E. Roth; Olivier Tercieux
Abstract: In 2012, New Orleans Recovery School District (RSD) became the first U.S. district to unify charter and traditional public school admissions in a single-offer assignment mechanism known as OneApp. The RSD also became the first district to use a mechanism based on Top Trading Cycles (TTC) in a real-life allocation problem. Since TTC was originally devised for settings in which agents have endowments, there is no formal rationale for TTC in school choice. In particular, TTC is a Pareto efficient and strategy-proof mechanism, but so are other mechanisms. We show that TTC is constrained-optimal in the following sense: TTC minimizes justified envy among all Pareto efficient and strategy-proof mechanisms when each school has one seat. When schools have more than one seat, there are multiple possible implementations of TTC. Data from New Orleans and Boston indicate that there is little difference across these versions of TTC, but significantly less justified envy compared to a serial dictatorship.
Keywords: School Choice; Top Trading Cycles; Justified Envy; Pareto Efficiency
JEL Codes: D47; I20
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Top Trading Cycles (TTC) mechanism minimizes justified envy (D51) | TTC mechanism is Pareto efficient (D61) |
| TTC mechanism allows for justified envy (D46) | DA eliminates justified envy (Y60) |
| TTC mechanism results in significantly less justified envy than serial dictatorship (D79) | TTC maintains comparable level of efficiency (L91) |
| Different implementations of TTC yield similar outcomes in terms of justified envy compared to serial dictatorship (D79) | TTC is constrained-optimal in minimizing justified envy (D61) |
| Lack of rigorous understanding regarding roles of priorities and preferences in TTC (D81) | limited use of TTC in practice (R48) |