Medicaid Enrollment, Eligibility, and Take-up During the Great Recession

One goal of the Medicaid program is to provide a safety net for unemployed individuals and their children who would not otherwise have a source of health insurance. Over the period from 2006 to 2013, the official unemployment rate climbed from 4.7% to a peak of 9.7% and the percentage of people out of the labor force increased from 33.7% to 36.4%. Over this same time period, the percentage of non-elderly adults enrolled in Medicaid increased from 6.7% to 8.7%. However, various states also expanded Medicaid eligibility for adults over this period, and thus the extent to which increased enrollment arose from job losses or lower-paying jobs versus more generous eligibility criteria within states is uncertain.

In this paper, we assess the impact on Medicaid eligibility arising from two sources: 1) state eligibility expansions; and 2) changes in income and employment among individuals who would have been eligible irrespective of any expansion in their state. We also decompose the relative contribution to Medicaid enrollment of those two eligibility factors, as well as how changes in behavior and characteristics of the non-elderly population impacted their decision to enroll in Medicaid.

Medicaid eligibility is simulated by applying detailed, state eligibility rules to income and family composition information collected in the nationally representative Household Component of the Medical Expenditure Panel Survey. We use this simulation to measure how Medicaid eligibility varied historically and under several counterfactual scenarios, including whether individuals would have been eligible for Medicaid under the rules applicable in their state in each year from 2006 through 2013.

Using a Blinder-Oaxaca method, we decompose a linear probability model of Medicaid enrollment. We compare the relative contribution on enrollment from: 1) changes in eligibility through state expansions; 2) changes in eligibility not through expansions; and 3) changing behavior or characteristics related to the decision to enroll. We test the sensitivity of this procedure by employing other decomposition methods for limited dependent variable models including those developed by Pylypchuk and Selden (2008) and Fairlie (2003).

Our counterfactual simulations show that Medicaid eligibility increased from about 9% to 18% of non-elderly adults from 2006 to 2013, and among those who were eligible as of 2013, 44% became eligible through a state expansion since 2005. Preliminary results from our decomposition of enrollment suggest that approximately 70% of the increase in enrollment from 2006 to 2013 can be explained by our two categories of eligibility: approximately half is due to state expansions and the other from increases in eligibility that would have occurred in the absence of state expansions. Factors relating to the take-up of coverage account for the other 30% of the enrollment increase, primarily due to reductions in offers of employer-sponsored insurance.

A primary role of Medicaid program is to serve as a countercyclical support system. Our findings demonstrate that the program’s recent enrollment increases are partly attributable to this role as well as to the expansion of ACA eligibility and the erosion of employer based coverage.