Abstract

This paper develops and implements dominance criteria for evaluating the performance of compulsory school systems.The main criterion that we propose is shown to be the smallest transitive criterion compatible with three elementary principles for evaluating such school systems. The ?rst principle requires that improving the cognitive skills of children of a given family background is a good thing. The second principle says that the evaluation attached to a given cognitive skill of a children is all the more favorable as the children come from a family with an unfavorable background. The third principle says that, for a given distribution of children's cognitive skills and family backgrounds, reducing the correlation between family background and cognitive skill is a good thing. Our dominance criterion considers that school system A is better than school system B if, for any pair of reference family background and cognitive skill, the fraction of the children who have a better cognitive skills and come from a worst family background than the reference is weakly larger in A than in B. We then apply our criterion to the ranking of education systems of major OECD countries, taking the standardized PISA scores as the measure of cognitive skills, and considering in turn various indices of the family backgrounds. We show that, albeit incomplete, our criterion enables the comparison of quite a few educational systems. Educational systems of fast growing asian economies - and in particular Vietnam, appear rather at the top of our rankings while those of wealthy arabic countries such as Qatar or Arab Emirates appear at the bottom. We later extend our criterion by incorporating some additional value judgement about varying inequality in cognitive skills. The extension gives rise to more discriminatory dominance criteria that we also apply to the ranking of national education systems.