The effect of school funding decisions on income inequality: an exploration of machine learning for causal analysis

The effects of public K-12 education funding disparities on student outcomes remain one of the most contentious issues in the realm of education. While many studies approach this problem through the analysis of discrepancies in short-term results such as test scores, no study has yet attempted to analyze the effect of discrepancies on long-run economic outcomes; this study attempts to fill this divide. To do so, we investigate the long-run economic health of children born from 1978-1983 and draw comparisons with school funding statistics from the 1991-1992 school year. We conduct this analysis using an instrumental variable approach combined with the deployment of machine learning regression algorithms in hopes to accurately model the causal impact of disparities in school funding. We find that machine learning models are more effective at modeling the causal relationship between school funding and income at age 35 than a standard linear regression model, using state fiscal neutrality scores as an instrument. We conclude that increases in school funding in the school district where a child grew up are causally linked to that child's outcome at age 35 and demonstrate that increasing school funding could be one potential solution to help remedy income inequality in the United States.