Does class size matter? How, and at what cost?

Using Greek high quality administrative data we show that class size has a hump shaped effect on achievement whereas most of the literature assumes linearity. We then embed our estimates for this relationship in dynamic structural model with costs hiring and firing. find sizes around 19 maximize attainment, existing result an average 22–23 being chosen. Firing are about same as consistent presence unions. Reducing firing to zero raises by 4–6 students reduces 0.8 1.2 GPA points. 3–5 increases 0.3 Raising wages 50%, 2 falls points suggesting unions (which raise salaries costs) do size, but have small achievement. caps costly, more so schools, even when they set at levels well above size.