Education Production Functions Using Instructional Time as an Input

This paper contributes to the literature on the effectiveness of schools. The analysis is unique in five respects. First, the data employed include the minutes of instruction per five day week in each of four subjects for all the public schools in Illinois. Few education production function papers have any information on the amount of instruction students receive in a given subject. Second, the theory section of the paper argues that class size should interact with student, teacher, and school characteristics as well as with the instructional times. The empirical analysis tests this hypothesis. Third, the interactions show that the small marginal effects of class size or of teacher qualifications may result because the harm of larger class sizes is undone by better trained personnel, and the benefits of better trained teachers is undone by large class sizes. Fourth, the data span three years enabling the use of panel data techniques. Individual school specific effects are estimated and found to be strongly significant. Moreover, in the presence of these effects, class size always has the theoretically predicted sign and is highly significant. Fifth, the time series aspect of the data allows for tests of the comparability of test scores over time. If the test scores are to be compared over time as a guide to school performance and improvement, then those scores should not differ significantly over time. The results reveal that year to year differences can be substantial and statistically significant