Panel Paper: Using Machine Learning to Examine Heterogeneity of the Effects of Changes in the Earned Income Tax Credit on Child Health and Development

Thursday, November 2, 2017
McCormick (Hyatt Regency Chicago)

*Names in bold indicate Presenter

David Rehkopf, Stanford University

The examination of the effects of social policies on health has understandably focused on the overall population average treatment effects. However, within this population average, there may be substantial differences in the effects of the policy, with the potential for a policy increasing or decreasing health inequalities depending on the population groups that do and do not benefit. Traditionally, examination of the heterogeneity of treatment effects has proceeded by priors from the literature, and due to power issues generally has examined only a few potential factors leading to heterogeneous effects. At the same time, there have been considerable advances in machine learning algorithms that scan over a large number of covariates to establish models of covariates that best explain a specified outcome, penalizing for greater degrees of freedom that come from multiple comparisons. My analysis uses this approach to examine potential heterogeneity of treatment effects of the largest anti-poverty policy in the United States, the Earned Income Tax Credit. I examine the spatial and temporal changes in the generosity of the policy over time (1986 to 2012) as an exogenous exposure with effects on child development outcomes using data from the 1979 National Longitudinal Survey of Youth. Rather than examining heterogeneity of treatment effects only by basic demographic factors, I using an ensemble machine learning approach (using multiple machine learning algorithms including random forest, Elastic-Net, Least Angle Regression, Support Vector Machine, Bayesian GLM) to examine whether treatment effects differ by several dozen potential demographic, socioeconomic, environmental and behavioral factors. I find substantial heterogeneous effects of the policy by race, age, year of the policy, parental test scores and paternal level of education. If confirmed in another dataset, these results will offer guidance toward modifications of the policy as well as supplemental programs that ensure that the health benefits of the Earned Income Tax Credit are beneficial across the population, and do not act to exacerbate health inequalities with respect to obesity and other measures of child development.