Association for Public Policy Analysis & Management

In New York State, school districts that experience declining enrollment do not lose state aid funding, leading these districts to have artificially inflated levels of revenues and expenditures per pupil. This results from a controversial provision called “Save Harmless”, implemented in 1974, which stipulates that districts will not lose any state funding regardless of declining student need. Through a robust set of descriptive analyses and falsification tests we demonstrate that, conditional on current year enrollment and district and year fixed effects, enactment of the Save Harmless provision appears essentially random (it is not correlated with any observable district characteristics or trends). This allows us to employ a measure of percentage enrollment change from the prior year as a conditionally-exogenous instrument for per-pupil expenditures.

We find that per-pupil expenditures have a large and statistically significant impact on student achievement, in greater magnitude than estimates produced in direct two-way fixed effect regressions of achievement on per pupil expenditures. Furthermore, we conclude that some estimates of the effects of per-pupil expenditures may underestimate the longer-term general equilibrium effects. Increases in school spending leads to higher immediate rates of teacher hire, which has a countervailing negative impact on student performance from the introduction of inexperienced teachers to district classrooms. We show that controlling for changes in the experience level of the teaching labor force meaningfully increases effect sizes of per-pupil spending, with larger gains experienced by economically disadvantaged students than in the general population. The findings in this study demonstrate that even in the state with the highest per-pupil spending in the nation, school resources can positively influence student outcomes – and these effects may grow over time as newly hired teachers gain experience.