Review of Economic Studies (2002) 69, 437–466 0034-6527/02/00170437$02.00 c  2002 The Review of Economic Studies Limited The Design of Optimal Education Policies GIANNI DE FRAJA University of York and CEPR First version received May 1998; final version accepted August 2001 (Eds.) This paper studies the education policy chosen by a utilitarian government. In the model, households differ in their income and in their children’s ability; income is observed by the government, but ability is private information. Households can use private education, but cannot borrow to finance it. The government can finance education with income tax, but at the cost of blunting the individuals’ incentive to exert labour market effort. The optimal education policy we derive is elitist: it increases the spread between the educational achievement of the bright and the less bright individuals, compared to private provision. It is also such that the education received by less bright individuals depends positively on their parental income. Finally, the optimal education policy is input regressive, in the sense of Arrow (1971, Quarterly Journal of Economics, 38, 175–208): households with higher income and brighter children contribute less in tuition fees towards the cost of the education system than households with lower income and less bright children. 1. INTRODUCTION This paper builds a theoretical foundation to public intervention in the provision of education and to individual households’ financial contribution to the cost of their education. In this paper, households differ in income, and therefore in their ability to pay for education, and in their children’s potential for future earnings, and therefore in their ability to benefit from education. Public intervention is justified on three grounds: first, capital market imperfections, which prevent individuals from borrowing at the market rate to finance education; second, positive external effects of education; and third, redistributive concerns. The government chooses combinations of educational achievements and financial instruments—income taxes, tuition fees, and students loans—to maximize a utilitarian welfare function. Our analysis, therefore, departs from the “political economy” approach to the theory of education provision, where individuals vote on the level of education provision in their district and on the associated local property taxation (e.g. Stiglitz, 1974; Fernandez and Rogerson, 1995). This approach captures the local finance mechanism for (primary and secondary) education provision in the U.S., but it is not necessarily the appropriate model for the rest of the world. In most other countries, education policies are set nationally, by the government of the day, which is often elected for reasons that have only little to do with its, or the opposition’s, proposals for education, and which implements the education policy which it prefers. We let the education received by a person and her financial contributions towards its cost be directly related to her parental income, which is observable, and to her ability, which, however, is private information. Therefore, unlike most of the literature (e.g. Stiglitz, 1974; Tuomala, 1986; Glomm and Ravikumar, 1992), we do not assume uniform public provision. Instead, in line with the practice in most countries, different individuals may receive different levels of publicly provided education. We also differ from most of the literature, which posit free public provision, by assuming that the government may ask individuals (and their parents) to contribute 437