The effects of class size on student achievement: New evidence from population variation

I identify the effects of class size on student achievement using longitudinal variation in the population associated with each grade in 649 elementary schools. I use variation in class size driven by idiosyncratic variation in the population. I also use discrete jumps in class size that occur when a small change in enrollment triggers a maximum or minimum class size rule. The estimates indicate that class size does not have a statistically significant effect on student achievement. I rule out even modest effects (2 to 4 percent of a standard deviation in scores for a 10 percent reduction in class size).