Does Medicaid Improve Mental Health? An Examination of Treatment Use and Financial Security

Evidence of the Medicaid program’s impact on adult health outcomes lags behind research on Medicaid coverage and health care access. Few studies have investigated the effects of adult Medicaid coverage on health outcomes, and mental health is one of the few areas in which there is some causal evidence of improvement. Most of these studies are limited to one or a handful of states, and the robustness of this relationship to other state contexts is unclear. Additionally, empirical analyses of the channels through which Medicaid coverage may improve mental health are scarce. In this paper we examine whether the introduction of Medicaid coverage for non-elderly adults in a wave of pre-ACA reforms across the nation improves mental health. Additionally, we study the degree to which changes in mental health can be attributed to changes in financial security versus changes in treatment.

We use the natural experiments created by state-level decisions to introduce Medicaid for parents and adults without dependent children between 2001-2013 to estimate the effect of acquiring Medicaid on four different measures of mental health. These measures include the SF-12 Mental Component Summary score, the Patient Health Questionnaire screening measure for depression, the Kessler index for serious psychological distress, and a global measure of self-reported mental health. We hypothesize that Medicaid may improve mental health outcomes by reducing the price of mental health treatment and/or increasing financial security. Thus, we examine the influence of Medicaid acquisition on intermediate outcomes that capture mental health service use and financial security. Data for this study consist of 12 panels of a restricted version of the nationally representative Medical Expenditure Panel Survey (MEPS) that includes state identifiers for the years 2001 – 2013 combined with a unique dataset we created that characterizes Medicaid coverage for non-elderly adults in all states and study years, the Medicaid Waiver Dataset.

We use difference-in-differences and instrumental variables methods to identify the effect of acquiring Medicaid coverage on study outcomes. We augment our models with person-level fixed effects ensuring that the identifying variation comes from within-person changes in insurance coverage and controlling for unobserved, time invariant individual factors that may be influencing outcomes. We examine the robustness of our main specifications to a number of alternative definitions of Medicaid expansion and modeling choices.

Study estimates increase our understanding of the Medicaid program’s effectiveness in achieving a key program objective, improving the health of its beneficiaries.