International cognitive tests, such as PISA or TIMSS, are increasingly used for benchmarking of education systems and designing educational policies. This study applies a quasi-experimental approach for estimating the effect of educations systems on the performance of students in international cognitive tests by exploiting variation in time in school induced by school entry rules. This method produces estimates of gains in achievement in the year before the test which enable a comparison of the performance of education systems based on gain scores instead of level scores. We find that one year of time in school on average matters for student performance in international cognitive tests and the effect of time in school decreases with age. We also find large differences between countries.
Remarkably, we find no association between the level of test scores and the estimated gains in achievement. At all levels of test scores we observe education systems that produce high gains and low gains in achievement. As such, a countries’ ranking in international cognitive tests might misguide educational policy. Assessments of the performance of education systems can be improved by using estimates of gains in student achievement.