Does Medicaid Enhance Educational Achievement? Evidence from a Natural Experiment

Background: Medicaid is the largest government insurance program in the United States specifically designed for low-income households. Previous studies show increased health service utilization and improved health outcomes in the short run due to Medicaid expansions. However, there is a scarcity of research linking the children health insurance program with its long-term impacts. We specifically focus on the outcome of educational achievements due to the existence of two conflicting views. On one hand, empirical evidence shows positive impacts of prenatal and pregnancy care on early childhood health and cognitive development; on the other hand, Murray (1984) proposed that the short-term benefits provided by welfare programs harm recipients by crowding out their work ethic and eroding their human capital in the long-term.

Objective: This study examines the long-term effect of Medicaid on educational attainments, i.e. high school graduation rates by exploiting a policy discontinuity of Medicaid expansion in 1990.

Methods: We take advantage of a natural experiment, the Omnibus Budget Reconciliation Act of 1990 (OBRA 1990). The OBRA 1990 regulates that children born after Oct, 1983 from families living below the federal poverty line are eligible for Medicaid, while the cohorts born before October, 1983 under the same conditions experience considerably lower rates of Medicaid eligibility. Therefore, individuals born right before Oct 1983 are ideal counterfactuals of individuals born immediately after Oct 1983 since they are non-eligible, but otherwise face the same environment. Hence, this source of random variation enables us to examine the treatment effect of Medicaid by utilizing a fuzzy regression discontinuity (RD) design.

To perform the estimation, we employ both linear models and non-parametric local linear regressions following Lee & Lemieux (2010). We include different model specifications in linear models incorporating socioeconomic variables, fixed effects and macroeconomic variables. We pre-define a window of 60 birth months. Therefore, cohorts born 5 years before and after Oct. 1983 are included in the interval [-60, 59].

For the first stage estimation, we use regression adjusted difference-in-differences analysis to investigate whether OBRA 1990 increases Medicaid enrollment.

Moreover, to show robustness, we conduct a balance check for covariates and a falsification test.

Data: The main data used in this study are from the National Health Interview Survey. The first stage analysis uses the National Longitudinal Survey of Young 1979 and NLSY79 Child & Young Adults.

Results: The first stage analysis shows Medicaid enrollment increases 6.4 percentage points by OBRA 1990, and this positive effect is significant among black children but not among Hispanic children. These findings provide support for the main findings shown below. Using the RD design, we find that Medicaid expansion by OBRA 1990 is associated with increased high school graduation rates for the overall sample. More specifically, Medicaid increases high school graduation rates by 1.4 percentage points. Further, we find a strong positive effect of Medicaid on high school graduation rates among black children (2.6 percentage points) but not among Hispanic children.