Bradley T. Heim
School of Public and Environmental Affairs, Indiana University

American Journal of Health Economics 1(3): 374-398, 2015

In 2006, Massachusetts passed An Act Providing Access to Affordable, Quality, Accountable Health Care, which made substantial changes to the health insurance landscape in the state with the aim of expanding coverage. Provisions of the act included the establishment of a health insurance Marketplace known as the Connector, subsidies for low income individuals to purchase of health insurance, an expansion of Medicaid, and individual and employer mandates. Since the Massachusetts reform was a model for the federal Affordable Care Act (ACA) of 2010, studying the impact of this reform not only informs on what happened in Massachusetts, but can also provide insight into the expected impacts of the ACA.

One area of recent research has been to examine the impacts of these reforms on labor market outcomes. In this paper, we focus on the phenomenon of job lock. Job lock refers to the reluctance of a worker to leave a job for another job, self-employment, or retirement due to a difference in the availability or cost of health insurance between their current position and the alternative. To the extent that job lock exists, it serves as a drag on economic efficiency, as workers are not able to switch to alternative positions for which they may be better suited or more productive. As a result, policy changes that lead to a decline in the extent of job lock, and increases in job mobility, may have beneficial impacts for the overall economy.

Because of the variety of provisions that were contained in the Massachusetts reform, the impact of the reform on job mobility is theoretically ambiguous. The Medicaid expansion and low income subsidies would be expected to increase mobility from a job to another job, self-employment, or retirement. However, the impacts of the employer mandate and changes in the price of insurance could either increase or decrease mobility, depending on the relative impacts on an individual’s current job and the attractiveness of other employment statuses.

To assess the impact of the Massachusetts reform on job mobility, we use data from a nine-year panel of tax returns that spans 2002-2010, and utilize information from W-2 forms to infer movement from a job to another employment status. We look at two different types of job separations: separation from a “primary” job, defined as the job with the largest amount of wages reported on a W-2 form, and separation from any job. To focus on job separations that are more likely to be voluntary and hence might be constrained by job lock, we further refine job separations to those with no concurrent receipt of unemployment insurance. The tax data also include information on gender, marital status, income, and age, which allow us to examine if there is heterogeneity in the response to the reform.

We estimate a difference-in-differences specification using a linear probability model, with the dependent variable denoting separating from a job (either any or a primary job) in a particular year. The treatment group consists of those residing in Massachusetts, and the control group containing those residing in a subset of other states. Pre-trends tests suggested that a control group of New York and New Jersey tended to best match the trends in Massachusetts prior to the reform, and so we use these states as the control group in our base specification. In our base specification, we treat 2002-2005 as the pre-reform period, and define the post period as being 2008-2009. We omit 2006 and 2007 from the analysis, since the law was signed during 2006, and implementation of major pieces of the reform lasted through 2007. Finally, we estimate separate impacts by gender, marital status (married or single), and type of separation (from primary job or any job)

In our main specifications for the sample as a whole, the results suggest that the Massachusetts reform had little impact on job mobility, as the vast majority of the estimated coefficients are small and statistically insignificant. The only marginally statistically significant estimate is found for married woman separations from primary jobs with no UI receipt spell, in which the Massachusetts reform is estimated to have increased separations by 0.6 percentage points (or 4% relative to the pre-reform separation rate of 15%).

In specification checks that utilized different states as control groups, we find that the results appear to be somewhat sensitive to the choice of comparison group. The estimates from our preferred specification fell in the middle of the range of effects, though the results using New York as a comparison group provide some evidence that the Massachusetts reform may have led to an increase in job mobility. However, in specification checks that varied the treatment of the transition years (2006-2007), the only positive impact of the Massachusetts reform on job separation was again found for separation from primary jobs among married women.

We then estimated separate impacts for different income and age groups. When the sample was cut by income, the Massachusetts reform is estimated have significantly increased job separations among married men and women with income below 300% FPL, though some of the magnitudes seem implausibly large. Further, when the sample is split by age, mobility appears to have increased among young married men and older married women.

Taken together, our estimates suggest the Massachusetts reform generally did not have a significant impact on job separations overall, though it appears to have increased separations from primary jobs among married women. In addition, mobility appears to have increased among low income married couples, young married men, and older married women.

The finding of increased job mobility among some subgroups (including the young and those with low income) at which major provisions of the reform (particularly the low income subsidies and Medicaid expansion) were targeted may be viewed as a positive development, particularly since much of the wage growth among individuals at early stages in their careers comes from job-to-job transitions. These results further suggest that similar increases in job mobility among such groups may result from the implementation of the ACA. Ascertaining whether such effects have materialized is clearly an important topic for future research.

^* This article is co-authored with Ithai Z. Lurie, US Department of the Treasury.