Class size, teacher hours and educational attainment

We employ a regression-discontinuity design to identify effects on educational attainment (years of education) of class size and the number of pupils per weekly teacher hour using administrative rules as instruments. We use a Danish administrative panel data set based on a 10% random sample of eight cohorts of pupils (and their parents). Restricting the sample to observations close to enrollment discontinuity points, instrumental variables estimates are consistently negative. Although the estimates are not very precisely determined, the point estimates indicate rather large positive effects of reducing class size, and especially of reducing the number of pupils per teacher hour.