Association for Public Policy Analysis & Management

Panel: Employment Lock: New Evidence from Changes in Health Insurance Provision and Healthcare Reform

(Health Policy)

The strong link between health insurance and employment is a defining feature of the US healthcare system. Before age 65, most individuals are insured through employer-sponsored group insurance plans. This can lead to “job lock” in which individuals delay job transitions, forgo self-employment opportunities, or retire at later ages in order to ensure coverage. Our panel considers three policy changes that introduce quasi-experimental variation in health insurance to study this phenomenon. Each paper makes novel conclusions about the extent of job-lock in the United States and the degree to which policies can and cannot loosen such frictions.

The first paper (Chattopadhyay) contributes to the literature on whether the presence of employer provided health insurance acts as a barrier to self-employment by exploiting variations in health insurance coverage that arose as a result of the Massachusetts Health Care Reform of 2006. Previous papers which examined the effect of this reform on self-employment had used difference in difference (DID) models to answer this research question and had provided conflicting evidence. However, since most of the post-reform period was affected by the Great Recession, the parallel trends assumption of DID is likely violated which makes these estimates suspect. This paper uses triple difference models to separate the effects of the reform from that of the recession and examines whether the health insurance coverage increases of the reform was able to weaken “job lock” and lead to a rise in self-employment.

The second paper (Feeney) explores whether the Medicaid expansion and tax credits/subsidies on the state and federal exchanges encourage early-retirement, focusing on individuals eligible for early Social Security benefits. The paper exploits variation within states across income thresholds, which allows us to analyze “retirement lock” across a range of incomes. Prior work focusing on Medicaid expansion only has been unable to do this in a traditional differences-in-differences framework, and our paper finds significant increases in retirement attributable to ACA policies not reported elsewhere.

Given the incentives embedded in the ACA to provide greater access to retirement health insurance (RHI) to those who previously did not have it, the final paper (Lee) evaluates the role of ACA on individuals’ retirement decisions while taking into account their heterogenous health insurance literacy rates. In comparing the retirement decisions of individuals who had RHIs prior to ACA with those who had not, the joint impact of ACA and insurance literacy is estimated using a triple difference (DDD) design. The insurance literacy is measured based on varying degrees of access to ACA assister programs, levels of education, cognitive abilities, and knowledge of insurance system and related terminologies.

All three papers take advantage of quasi-experimental variation to underline how the relationship between employment and health insurance poses a significant barrier for labor market transitions and how recent policies have been successful in weakening this link. Efforts that aim to erode these provisions may have impacts on future labor supply decisions for those nearing retirement and considering self-employment.