K-12 School Finance Reform

Hendren and Sprung-Keyser (2020) arrive at a cost estimate by first showing that the estimated results in Jackson et al. (2016) imply that every $1 of spending leads to a present discounted value increase in earnings of $4.49. To see this, note that Jackson et al. (2016) estimate that a 1% increase in spending leads to a 0.77% increase in wages in later life. Average family income at age 30 in Jackson et al. (2016)’s sample is $49,308, and to get to an estimate of individual income Hendren and Sprung-Keyser assume there are 1.302 earners per family, implying an individual income of $37,871. This implies that a 1% increase in school funding would increase yearly income by $293 on average. As schools in Jackson et al. (2016)’s sample spend on average $4,800 per year per student, a $1 increase in spending implies a 0.0017% increase in total educational spending over the ages 5-17. Combining these numbers, Hendren and Sprung-Keyser (2020) estimate that $1 of additional spending increases incomes by $0.51 in a given year between the ages of 20 and 45. The authors then project this yearly increase up to age 65 using standard forecasting methods, estimating a discounted sum of earnings increases of $10.81 for every $1 of upfront spending (at a 3% discount rate). Given the $10.81, the authors then apply a CBO tax and transfer rate of 18.8%. This suggests an increase in government revenues of $2.03 for every additional dollar of spending and a net cost of the policy of $-1.03 (95% CI of [-2.02, -0.06]).

Combining cost and WTP estimates yields an infinite MVPF. The estimates are statistically quite precise: taking the sampling uncertainty from the underlying parameter estimates suggests a 95% CI of [\infty, \infty]. This implies that the primary uncertainty about the MVPF estimates is about the validity of the lifetime income projection and any potential concerns about the empirical design used to estimate the causal effect.