They are not alone. Commentators have lined up to denounce the supposed failure of increased competition to produce higher international test scores. The claims are usually based on the latest PISA report's fourth chapter, which analyses what makes schools successful, and for this reason has become popular among politicians looking to improve their country's performance in international tests. But is its popularity warranted?

Unfortunately, the answer is a resounding No. The chapter amounts to little more than a firework display of correlations — and, as any social scientist can point out, simple correlations tell us nothing about the underlying causal relationships. A positive association between a policy and achievement could mask a zero or even negative causal effect (and vice versa). This means that pursuing reforms based on the chapter's conclusions could in fact be harmful.

In order to understand the effect of competition from independent providers on international test scores, we must therefore look beyond the OECD's correlations and consult the economic literature. And the contrast could not be starker. The strongest study available, published in the Economic Journal by Martin West and Ludger Woessmann, finds that competition from autonomous providers improves countries' performance in PISA in the long run, benefiting pupils in state and non-traditional schools equally, while also driving down costs. The positive impact on achievement is not an isolated finding; all available economic research analysing international test score differences between countries backs it up.

So far, so good. Absolute performance levels in international tests, however, are not everything. Does not competition from new types of schools at least reduce equity, as commonly feared? Well, no. While the PISA report finds no statistically significant relationship between such competition and equity, the available economic literature — as reviewed by Eric Hanushek and Ludger Woessmann in the Handbook of the Economics of Education — shows that it actually decreases the impact of pupil background on test scores. The research methods utilised are not ideal, but they are way ahead of the OECD's correlation analyses.

Yet the same research also indicates that private funding decreases equity in international tests. This is not surprising since it is likely to reflect greater choice opportunities for the rich compared to the poor. Private funding is not necessarily always bad, but this illustrates our point: good policy design is crucial for the effects of competition.

Similarly, differential system design partly explains the more mixed within-country research analysing the impact of choice and competition on various domestic achievement measures. For example, under certain conditions, allowing academic selection may incentivise schools to compete by cream skimming rather than by raising quality. This helps explain why competition has had mixed effects in Chile, where academic selection has been permitted, but a more consistently positive impact in Sweden, where it is banned.