Inequality in Health Insurance Coverage Before and After the Affordable Care Act

Although there has been substantial progress in improving health outcomes across the board, substantial disparities in health care access and a variety of health indicators persist. In the health economics and epidemiology literatures, the concentration index (CI) emerged as the workhorse tool for measuring socioeconomic related disparities in health and a variety of techniques for decomposing these income related inequalities (IRI) have emerged. In this study we construct the CI for health coverage using data from the 2010 and 2014 American Community Survey. We use total family income as our measure of socioeconomic status. The CI is bounded between -1 and 1 with positive values indicating a disproportionate concentration of individuals with health coverage in the upper tail of the income distribution. We apply the Erreygers correction to account for the discrete nature of the health insurance variable. We find the concentration index for health insurance coverage is indeed skewed in favor of individuals from higher income households, but the degree of inequality declined over the period analyzed. However, the decline in IRI appears to have only occurred for public health insurance. In particular, we find that the CI for having any type of health insurance coverage decreased from 0.15 to 0.11 from 2010 to 2014, suggesting the ACA has contributed to decreased disparity in health insurance coverage. Moreover, the CI for employer sponsored health insurance (ESHI) is markedly higher at 0.42 and did not change much between these years suggesting that most of these gains worked through the expansion of the Medicaid program. Our analysis proceeds in two stages. First, we compute the CI for the entire sample and we use the Wagstaff decomposition to determine how much of the IRI in health insurance coverage is due to the IRI between racial/ethnic groups. to test for the robustness of the CI we account for the economy of scale within the family (for example the cost of raising a third child may be different than the cost of raising the first child).

Next, we compute the CI by racial/ethnic group and we observe dramatic difference across race/ethnic groups. For example while blacks and Hispanics have a CI of 0.57 and 0.50 respectively; whites have a much lower CI (0.40). We then proceed to decompose these differences using the Heckley, Gerdtham, and Kjellsson (HGK) approach which builds upon existing approaches by employing recentered influence function (RIF) regression. The HGK approach is a generalization of the Oaxaca-Binder decomposition and it is used to determine how much of the difference in two CIs is due to observable factors such as family income, education, marital status, employment status and geographical location.