A possible trade-off between greater efficiency and equal opportunity is central to the debate on school vouchers, for instance. Whereas economists since Milton Friedman have argued that school choice would align provider incentives with the interests of consumers and thus lead to increases in educational quality, a more recent literature points to the possibility that school choice might exacerbate stratification by socioeconomic status across schools. In "School Choice, Stratification, and Information on School Performance: Lessons from Chile," Patrick J. McEwan, Miguel Urquiola, and Emiliana Vegas contribute to this literature in two ways. First, using highly disaggregated, district-level data for Chile and a regression-discontinuity design, they find evidence that the entry of private schools into the education market is not associated with significant increases in test scores but is associated with increases in social stratification. This conclusion is consistent with previous results from the literature, suggesting that greater school choice leads to increased sorting, with no commensurate improvements in average achievement. Second, McEwan, Urquiola, and Vegas consider the implications of sampling variance (and even of "population variance" over time) for policy schemes that reward or punish schools on the basis of changes in average test scores. Policymakers have focused on these so-called value added (that is, first-differenced) measures of test scores, since test score levels are highly correlated with family socioeconomic status, as a result of the stratification previously discussed. Changes in scores also reflect sampling variation, however, including variation among the specific groups of students starting school in any given year, as well as genuine changes in the quality of the services provided by the schools. Although this issue is unlikely to be a serious problem for large municipalities, it can generate substantial rerankings for individual schools and even for smaller districts, as the authors confirm through a number of statistical tests. These findings suggest that the Chilean debates on school choice, on the ideal design for voucher schemes, and on the precise manner in which data on student achievement can be used to reward or punish schools have contributed to advances in the educational agenda in Latin America but have not yet reached a final conclusion.
