Does Competition Among Public Schools Benefit Students and Taxpayers? Comment

School choice policies promise to align the incentives of school administrators with the demands of parents, and may therefore lead to more efficient educational production (Milton Friedman 1962; Geoffrey Brennan and James Buchanan 1980; John Chubb and Terry M. Moe 1990). Absent a large-scale school voucher program in the United States, however, this prediction has been difficult to test. Several authors (e.g., Melvin V. Borland and Roy M. Howsen 1992; Clive R. Belfield and Henry M. Levin 2002) have suggested studying the effects of "Tiebout choice," the use of the residential location decision to select among local monopoly education providers. The idea here is that fragmented governance induces competition among school districts analogous to that which would occur among schools with nonresidential choice. In an influential paper, Caroline M. Hoxby (2000) points out that current governance structures are potentially endogenous to school productivity, and proposes that variation in topography, which may have influenced optimal jurisdiction size before modern transportation technologies, provides a source of exogenous variation. She estimates instrumental variables regressions of individual test scores and school spending on a metropolitan-level Tiebout choice index, defined as one minus a Herfindahl concentration index with districts' enrollments as their "market shares," using as excluded instruments the number of larger and smaller streams in the area. She reports substantial positive effects of district fragmentation on student test scores and negative effects on spending. This Comment presents a reanalysis of Hoxby's test score results, which form the core of her empirical analysis. These results turn out to be quite sensitive to plausible alterations to Hoxby's specification. In particular, the large, significant effect of choice on achievement obtains