This paper reviews the empirical economic literature which examines the relationship between educational expenditures and outcomes. The author finds that there is a remarkable lack of consensus about the results of standard studies using the "education production function" conceptual framework, whether at the macro or at the micro level. A few examples of successful policy interventions to improve attainment and achievement have been identified using experimental techniques, for example class size reduction or remedial education by more child-friendly teachers. The econometric analysis of randomised or natural experiments can thus identify interventions that are cost-efficient, but its conclusions are bound to be context-specific and not enough studies have been conducted as yet for any regularity to emerge. Another strand of literature emphasises the incentives structure of the school systems, which affects the way in which available school resources are combined to produce outcomes. The paper concludes that the ability of economists to adequately model the functioning of schools could be further enhanced by making use of insights from other social sciences such as social psychology and sociology, relating to the behaviour of teachers and students. Although they remain quite marginal to the field, recent behavioural economics papers may provide a basis for such a renewal of the economics of education.