Language: 日本語 English

ex ante Pareto efficiency.

Atila Abdulkadiro ̆glu, Yeon-Koo Che, and Yosuke Yasuda
Sat, 2009-08-15

The Boston mechanism is among the most popular school choice pro-
cedures in use. Yet, the mechanism has been criticized for its poor incentive and
welfare performances, which led the Boston Public Schools to recently replace it
with Gale and Shapley’s deferred acceptance algorithm (henceforth, DA). The DA
elicits truthful revelation of “ordinal” preferences whereas the Boston mechanism
does not; but the latter induces participants to reveal their “cardinal” preferences
(i.e., their relative preference intensities) whereas the former does not. We show that
cardinal preferences matter more when families have similar ordinal preferences and
schools have coarse priorities, two common features of many school choice environ-
ments. Specifically, when students have the same ordinal preferences and schools
have no priorities, the Boston mechanism Pareto dominates the DA in ex ante wel-
fare. The Boston mechanism may not harm but rather benefit participants who
may not strategize well. In the presence of school priorities, the Boston mechanism
also tends to facilitate a greater access than the DA to good schools by those lack-
ing priorities at those schools. These results contrast with the standard view, and
cautions against a hasty rejection of the Boston mechanism in favor of mechanisms
such as the DA.

Atila Abdurkadiroglu, Yoen-Koo Che, and Yosuke Yasuda
Fri, 2008-11-28
Truthful revelation of preferences has emerged as a desideratum
in the design of school choice programs. Gale-Shapley’s deferred acceptance
mechanism is strategy-proof for students but limits their ability to communi-
cate their preference intensities. This results in ex-ante inefficiency when ties
at school preferences are broken randomly. We propose a variant of deferred
acceptance mechanism which allows students to influence how they are treated
in ties. It maintains truthful revelation of ordinal preferences and supports a
greater scope of efficiency.