The literature on the impact of class size on student performance is largely inconclusive. Yet, there is still a public perception that size matters. Is the econometrician wrong? Miguel Urquiola and Eric Verhoogen argue that there is a subtle issue with class size that can seriously affect estimates.
Many studies exploit the following discontinuity. Jurisdictions mandate that class sizes cannot be larger than some number. For schools, this means when the number of students overtake a multiple of this number a new class needs to be opened. The consequence is a sudden decrease in the average number of students in classes, because the number of students is close to continuous, the number of classes is, however, a low integer. So far so good.
But the problem is with the endogenous reaction to this discontinuity. Schools are very reluctant to open a new class just because of one student, and will find ways around this, for example by increasing tuition. Also, parents select schools according to the number of students in classes. The big question is whether this matters in any significant way. Urquiola and Verhoogen build a model of schools maximizing profits and parents choosing schools and test it with data from Chile. There, private, subsidized schools are for profit and thus fit nicely with the model and indeed the above mentioned effects are important.
The schooling environment in Chile is particular, because a large portion of schools are for-profit. This may not be true elsewhere, but the fact is that schools are still reacting to incentives, and parents, too. So while the data and motives may not be as easy to model as in Chile, this needs to be taken into account for any estimate of the impact of class size.