Thoughts on the New Brookings School Choice Report

A new Brookings Institution report suggests ways for the federal government to promote school choice. On the eve of its release, I voiced some practical and constitutional objections to the idea. Now that the report is out, contributing author Jay Greene asks if I’m still apprehensive. The short answer is yes.

Brookings assembled an impressive group of scholars to write the report, and their education policy recommendations deserve serious consideration. Their goal of ensuring more and better access to more and better educational choices is one that I share, and I hope the following comments will help advance that goal.

Good policy, like good science, is grounded in concrete evidence. Only where evidence is lacking is it wise to fall back on theory. The Brookings report relies on theory in a couple of important areas where extensive evidence is available to show us the way. In particular, the authors acknowledge that U.S. experience with alternative school systems is minimal, but (with a single exception — see below) they do not discuss the vast wealth of evidence from other nations that have more extensive and longer-running experience with school choice systems.                   

Some have argued that we can learn little about school governance from other nations because cultural and economic factors affect educational outcomes too greatly. This criticism does not apply to within-country comparisons of alternative school systems — and virtually all of the literature comparing alternative school systems is within-country. A comparison of government-run, government-funded private, and parent-funded private schools within India, for instance, is not muddied by cultural or economic differences between India and the United States.

What’s more, if we see the same pattern of inter-sectoral results manifested within many different countries, we can be even more confident that the observed differences are truly systemic than if we had inter-sectoral results for only a single nation. A thorough review of the worldwide within-country comparative school governance research is thus essential to optimal education policy design.

It’s also worth noting that openness to non-U.S. research is the norm in engineering and in other scientific fields. If an Indian computer programmer develops an improved sorting algorithm, his American counterparts don’t discount it for cultural reasons. Nor do U.S. civil engineers disregard what can be learned from French bridge projects. Nor do physicians ignore the results of high quality genetic or drug research performed in Iceland or England. As long as the methodologies employed control for cultural and other mitigating factors that might affect the results, American scientists and engineers are not, as a rule, parochial. Education seems the sole exception.

The one bit of foreign school choice evidence discussed in the new Brookings report is a paper by Hsieh and Urquiola reporting results from Chile’s highly regulated voucher-like system. More than a dozen studies of Chile’s voucher system have been conducted, the majority finding that private voucher schools significantly outperform public schools after suitable controls, or that the competition engendered by the program has improved overall academic achievement, attainment, or ultimate earnings of graduates.

Seeming to deviate from this pattern, Hsieh and Urquiola found that the regions in which private school enrollment grew most quickly had overall achievement (of public and private students) that was no better or was even worse than in regions where it grew more slowly. Hsieh and Urquiola concluded from this that increased competition did not improve achievement. But their evidence also supports a quite different conclusion: that regions with bad and worsening public schools drove families more quickly into the private sector.

Hsieh and Urquiola looked at the first 16 years of the program, during which time public schools still enrolled the majority of students (though that share was falling continuously and has since dropped below 50 percent). Overall academic performance was thus still chiefly a function of public school performance, and public schools are protected from private sector competition by receiving extra municipal funding that is not tied to enrollment and to which the private voucher-receiving schools do not have access. So it was never reasonable to expect, based on the design of the Chilean system, that public schools would show significant gains from competitive forces, because those forces never really touched them.

Given the evidence just presented and the fact that the bulk of studies of the Chilean program contradict Hsieh and Urquiola’s conclusion, it seems likely that they did indeed misinterpret their findings.

More importantly, Hsieh and Urquiola’s conclusion is not only an aberration within the Chilean research, it is also an aberration within the worldwide literature on the relative merits of market and monopoly provision of education. When I reviewed this research for the Journal of School Choice last year, I found 65 studies reporting 156 separate statistical comparisons of public and private school achievement. When truly market-like programs are compared to monopolistic ones such as U.S. public schooling, the statistically significant results favoring markets outweigh the results favoring monopolies by a margin of 15 to 1 (and they also handily outnumber the insignificant findings).

By offering only the uncharacteristic Hsieh and Urquiola study out of the vast literature just described, the Brookings paper is apt to give readers a mistakenly negative perception of the worldwide evidence. This is particularly true because the authors are avowed supporters of school choice.

In the absence of the international evidence, the Brookings authors are forced to resort to theory on several crucial policy issues, such as the need for and merits of state regulation of the marketplace (e.g., with respect to virtual schools) and state information services to supply a perceived gap in information available to parents. Consulting that evidence leads to the conclusion that government regulatory efforts to improve the quality of educational services are ineffective at best and counterproductive at worst (see the paper linked above for the evidence supporting that pattern, and see James Tooley’s The Beautiful Tree for an explanation of its cause). It shows, furthermore, that even illiterate parents in the poorest slums of the third world are not only capable of making wise educational choices for their children in the absence of government advice, but that they are already doing so in massive numbers.

In addition to the above concern, the practical and constitutional issues I raised in advance of the Brookings report’s publication still apply now that I have read it in full.

That said, this is one of the most benign sorts of disagreements scholars can have, rooted as it is on our having developed our policy recommendations from different data sets. I look forward to hearing how the authors of the Brookings report think the worldwide evidence bears on school choice policy.