That educational inputs should be important determinants of educational outcomes is a proposition that appeals to common sense, but is nevertheless controversial in the literature both for developed and lessdeveloped countries. Surveys by Hanushek (1986), for developed countries, and (1996), for developing countries, argue that school facilities have at best tenuous effects on outcomes, particularly on test scores. Kremer (1996), emphasizes that such a negative overall assessment of the evidence rests on Hanushek?s interpretation of statistically insignificant findings as evidence against an effect of school quality, but notes that there is a singular absence of evidence from developing countries that the pupil-teacher ratio is an important determinant of outcomes.