46 School choice 1. INTRODUCTION S chool choice mechanisms affect the educational experiences and outcomes of many students around the world. The past two decades have witnessed major reforms in this domain. In designing practical markets, institutions have relied on economic theory, computation, and controlled laboratory experiments (Roth, 2002). For example, shortly after the publication of Abdulkadirog˘lu & So¨nmez(2003), New York City high schools replaced their previous me­chanism with a variant of the student-proposing deferred-acceptance(DA) mechanism(Gale& Shapley, 1962; Abdulkadirog˘lu, Pathak,& Roth, 2005). In the school choice reforms in Boston, matching theorists directly influenced the adoption of the DA mechanism(Abdulkadirog˘lu, Pathak, Roth,& So¨nmez, 2005). In this case, experimental data helped persuade the Boston public school authorities to switch from the Boston immediate acceptance mechanism (BOS) to DA(Chen& So¨nmez, 2006). In parallel with these reforms, policy makers in Chicago and in New England independently abandoned the Boston mechanism and adopted versions of the DA(Pathak& So¨nmez, 2013). Labo­ratory experiments provide the first data for institutional redesigns when field data is not yet available. Even when field data is available, lab experiments compare the performance of different mechanisms at a level of detail that cannot be obtained from field data. In an incomplete information setting, Chen& So¨nmez(2006) present the first experimental study of three well-known school choice mechanisms, BOS, DA and the top trading cycles(TTC). They find that DA outperforms TTC in truthful preference revelation, despite the strategy-proofness of both mechanisms. Furthermore, they show that TTC does not outperform DA in efficiency, although theoretically it is efficient whereas DA is not. Among the three mechanisms, BOS performs the worst in terms of truthful preference revelation and efficiency. While a stability comparison is not presented in Chen& So¨nmez(2006), using the same experimental setting, Calsamiglia et al.(2010) find that DA is more stable than TTC, which in turn is weakly more stable than BOS. In this paper, we ask two questions. First, how do the three school choice mechanisms perform in a complete information setting? Second, how does information provision affect the performance of each mechanism? To answer these questions, we run an experiment under the complete infor­mation setting, using the same set of parameters as in the designed environment Journal of Mechanism and Institution Design (), 2016