Working Paper: NBER ID: w11965
Authors: Atila Abdulkadiroglu; Parag A. Pathak; Alvin E. Roth; Tayfun Sonmez
Abstract: In July 2005 the Boston School Committee voted to replace the existing Boston school choice mechanism with a deferred acceptance mechanism that simplifies the strategic choices facing parents. This paper presents the empirical case against the previous Boston mechanism, a priority matching mechanism, and the case in favor of the change to a strategy-proof mechanism. Using detailed records on student choices and assignments, we present evidence both of sophisticated strategic behavior among some parents, and of unsophisticated strategic behavior by others. We find evidence that some parents pay close attention to the capacity constraints of different schools, while others appear not to. In particular, we show that many unassigned students could have been assigned to one of their stated choices with a different strategy under the current mechanism. This interaction between sophisticated and unsophisticated players identifies a new rationale for strategy-proof mechanisms based on fairness, and was a critical argument in Boston's decision to change the mechanism. We then discuss the considerations that led to the adoption of a deferred acceptance mechanism as opposed to the (also strategy-proof) top trading cycles mechanism.
Keywords: Boston school choice; deferred acceptance mechanism; strategic behavior; school assignment
JEL Codes: C7; I2
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Boston school choice mechanism (C78) | parental strategic behavior (D15) |
| parental strategic behavior (D15) | inefficiencies in assignment process (D61) |
| existing Boston mechanism (L64) | misrepresentation of true preferences (D91) |
| misrepresentation of true preferences (D91) | unassigned students (Y40) |
| transition to strategyproof mechanism (D47) | alleviate strategic burden on families (H56) |
| transition to strategyproof mechanism (D47) | improve overall assignment efficiency (D61) |