International Evidence on Expenditure and Class Size: A Review

In the United States, evidence abounds on the effects of expenditures and class size on student achievement, but often it is controversial.¹ In other parts of the world, hard evidence is not as easy to come by, mostly because of data limitations. But over the years, testing agencies have started to collect data on student performance and family background and on school resources in countries around the world, allowing a look at the association between school resources and student achievement in different international settings. This paper reviews what can be learned from international student achievement tests such as the Third International Mathematics and Science Study (TIMSS) and Programme for International Student Assessment (PISA) exams in terms of the effects of expenditures and class size on student achievement.

Such evidence obviously is of great interest to countries that lack a national data set that allows for empirical analysis, and for many developed countries, evidence comparable to that in the U.S. literature is largely absent. That is true particularly for countries in western Europe; Psacharopoulos claims that "more research has been done on the economics of education in developing countries than in Europe."² The poor state of European research on educational production is especially disappointing because European evidence would seem to be much more directly comparable to U.S. evidence than would evidence from developing countries, given the relative levels of economic development and educational attainment of the countries involved.³

The international evidence is also of interest in discussions of the U.S. education system. Because the education systems and levels of educational expenditures in other countries differ from those in the United States, U.S. observers can learn whether resource effects would differ in a differently structured system. Observing resource effects in countries with different levels and distributions of expenditures may also help to distinguish between varying explanations for the U.S. evidence: for example, between technical explanations based on diminishing returns to resource inputs and economic explanations based on the lack of incentives to use resources to enhance student learning effectively.⁴ Also, as discussed toward the end of this paper, the controversial results from different U.S. studies may in part reflect true heterogeneity in the effect of resources, in that resources may matter in some circumstances but less so in others. Again, the international evidence may prove useful in testing for such heterogeneity, estimating resource effects on comparable data in different countries.

A final motivation to look at the international evidence stems from the fact that test scores on previous international cognitive achievement tests conducted between the mid-1960s and early 1990s have been found to be strongly associated with a country's subsequent economic growth and level of economic development.⁵ That empirical evidence supports recent theories of economic growth that stress the importance of human capital.⁶

It should be emphasized at the outset that this paper is a review of results from several previous studies on the international evidence on expenditures and class size and that it aims to be both brief and nontechnical. Therefore I present only the main results of leading studies on the topic, discussing technicalities only insofar as it is absolutely necessary to understand the results. In the trade-off between brevity and detail, the review leans strongly toward brevity. For technical details and additional robustness checks, the interested reader is referred to the original sources.

The paper first reviews aggregate international evidence on the association between educational expenditures and student performance and then international student-level evidence on the effects of class size. The aggregate evidence considers variation in expenditures and performance both across countries and over time. The micro-evidence presents results of conventional estimates and two quasi-experimental strategies to identify class-size effects—one based on natural cohort fluctuations, the other on rule-induced discontinuities—implemented by using the international data in many countries. The paper closes with a brief analysis of the association between teacher quality and class-size effects, as well as brief conclusions.

The Aggregate Picture

This section presents the aggregate picture...