Association for Public Policy Analysis & Management

Policy makers and researchers have long been interested in whether Medicaid affects the employment choices of the eligible population. This question has grown in importance as the number of non-elderly adults eligible for Medicaid has increased over time. A large literature suggests that Medicaid expansions lead to small declines in labor market activity because the program reduces returns to work and imposes an implicit tax on wages. In this paper we ask a novel question concerning Medicaid’s effect on employment. We consider whether, conditional on income eligibility limits, the set of benefits covered by Medicaid alters employment outcomes. Our theoretical model suggests that more generous benefits will increase access to medical services that reduce or alleviate functional limitations that hinder the type or amount of work conducted by beneficiaries. To study this question we consider the case of coverage for vision services (i.e. for the correction of refractive error). Medicaid vision coverage has been previously established to increase utilization of services and reduce objectively measured visual impairment. Correcting a refractive error could increase the number of jobs a beneficiary is qualified for, it could reduce barriers to transportation, or it could relieve the discomfort that often accompanies visual impairment.

Using data on low-income non-elderly adults from the Current Population Survey, we estimate the effect of adult vision benefits on employment status, hours worked, earnings, and occupation type using difference-in-differences and triple difference models. These models take advantage of state-by-year variation in vision benefits that is exogenous to individual employment decisions. We compare how this variation affects the outcomes of enrolled beneficiaries compared to similar non-enrolled adults that ought not to receive any benefit from Medicaid benefit policy. Our results stand up to a number of robustness tests including the inclusion of a large set of state-by-year controls, state-by-year linear trends, the inclusion/exclusion of sample weights, and the use of alternative comparison groups. We also demonstrate that vision benefits do not alter the observed composition of the enrolled population.

We did not find any statistically significant effects of vision benefits on employment status or earnings, but we did find significant positive effects on hours worked and full-time versus part-time employment among Medicaid participants. Beneficiaries living in states that cover vision benefits worked 6.2% more hours in a usual week and were 6.0 percentage points more likely work full-time versus part-time. We also observed a shift towards occupations that require higher levels of skill as measured by an index derived from occupation specific average wages. We found no effects in the non-enrolled population which strengthens our confidence that results do not stem from otherwise unmeasured trends.

This paper suggests that that Medicaid’s effect on employment is not homogenous, but is sensitive to the set of services that the program covers. State officials making coverage decisions for Medicaid must trade off the added cost of more generous coverage with the benefits that result from reduced work limitations.