The Affordable Care Act and the labor market

Chicago Fed Letter

In 2010, Congress passed the Affordable Care Act (ACA), the largest expansion of health insurance since the advent of Medicaid and Medicare roughly a half century ago. Because the law is being phased in slowly and many provisions are still years away from being launched, the law's impact on employment, wages, job mobility, retirement, self-employment, economic efficiency, and overall well-being remains contentious.

To examine what is currently known, on March 21, 2014, the Federal Reserve Bank of Chicago and the Institute of Government and Public Affairs of the University of Illinois brought together scholars who have been studying labor market implications of the ACA.

How might the ACA impact the labor market?

The ACA has many elements that could conceivably affect the employment and work decisions of firms and employees.1 We begin this Chicago Fed Letter by highlighting three key provisions-expansion of Medicaid coverage, health insurance exchange subsidies, and the employer health insurance mandate.

The ACA expands insurance coverage among low-income households in two ways. First, households with income up to 138% of the federal poverty level (FPL) are now eligible for Medicaid coverage. Notably, insurance is available irrespective of family circumstances, allowing childless adults to qualify for Medicaid for the first time in many states. Second, the ACA sets up an "exchange" to purchase health insurance and offers subsidies for households with annual income between 100% and 400% of the FPL. The amount of the subsidy drops as income rises. For example, the annual insurance premium for a household at the FPL is capped at 2% of income (or currently $23,550 x 0.02 = $471 for a family of four) but 9.5% of household income for those that are 400% above the FPL. Households that earn above 400% of the FPL or who are offered qualified insurance through their employer do not qualify for the exchange subsidy.

Both the Medicaid expansion and the health insurance exchange subsidies disentangle employment from access to health insurance. This makes it easier for workers to change jobs, become self-employed, or stop working, as they no longer have to worry about retaining their employer-provided health insurance (EPHI).

Like many other means-tested social safety programs, including those that offer food, housing, or wage assistance, the Medicaid expansion and the exchange subsidies may create a disincentive to work for some. The disincentive arises from two factors. First, some benefits become less valuable as income rises, discouraging more hours of work per week or more weeks of work per year (the substitution effect). An important exception is the increase in Medicaid eligibility to 138% of FPL, which potentially increases work incentives for those living in states that previously capped eligibility at a lower FPL. Second, by providing health insurance these programs effectively make eligible households wealthier, which may also discourage work (the income effect). These work disincentives may be particularly important around program phaseouts-in particular, at 138% of FPL for Medicaid and 400% of FPL for the exchange subsidies.

A series of tax and eligibility criteria in the ACA may further alter incentives to work. These include the payroll tax for Medicare's hospital insurance program, an excise tax on certain high-cost ("Cadillac") health insurance plans, and the penalty that an individual must pay for being uninsured.

It is worth emphasizing, however, that the majority of households are not affected by the expansion of health insurance coverage under the ACA. Moreover, to date, at least 19 states are considering opting out of the federally financed Medicaid expansion program.2 As a result, residents in those states may not face the work disincentives associated with Medicaid. Moreover, in those states, residents with income below 100% of the FPL (but not 100% to 138% of the FPL) will be ineligible for exchange subsidies as well. For these households, work incentives are positive, since increasing household income up to the FPL qualifies them for exchange subsidies. Finally, even if there is agreement on the direction of the work incentive effects of various pieces of the ACA, the magnitudes of the income and substitution effects are currently unknown.

A third key feature of the ACA is firms with at least 50 full-time (30 or more hours per week) employees must offer affordable health insurance or face a penalty.3 If an employer cannot offset the penalty via lower wages, say because employees are already paid the state's minimum wage, the demand for workers will fall. The penalty may also encourage employers to reduce workers' schedules to fewer than 30 hours to avoid being subject to the provisions. However, since 96% of firms with more than 50 employees already offered health insurance prior to the launch of the ACA,4 the employer mandate will affect few firms and mainly those whose insurance does not pass the affordability guidelines.

The research

The Affordable Care Act was signed into law in March 2010, but its provisions are still being phased in. The health insurance exchanges opened in late 2013, the Medicaid expansion begins in 2014, and the employer mandate starts in 2016. Thus, it will be some time before we can directly estimate the impact of the law on labor markets. One way to infer its impact is by studying earlier state-level policy changes in Tennessee, Wisconsin, and Massachusetts, as four of our conference papers did. Two others simulated stylized models of an economy with and without the ACA to infer the impact of health insurance reform.

Craig Garthwaite (Northwestern University), Tal Gross (Columbia University), and Matthew Notowidigdo (University of Chicago) studied the employment response to a large contraction in the availability of public health insurance for low-income people in Tennessee.5 In 1994, Tennessee moved its existing Medicaid population into managed-care plans and simultaneously provided free or subsidized health insurance to uninsured and uninsurable low-income people, in particular able-bodied childless adults, who did not otherwise qualify for Medicaid. In 2005 the state eliminated this expanded program, thereby removing insurance coverage for approximately 4% of the nonelderly adult Tennessee population.

Garthwaite and colleagues used this policy change to assess the degree to which Tennesseans who lost insurance coverage turned to the labor market for work and EPHI relative to similar people in southern states that did not change eligibility rules. They found the fraction of people in Tennessee who reported receiving "publicly provided" insurance fell by about 4.6 to 7.3 percentage points, from a base of about 20%. The fraction of people employed increased by about 2.5 to 4.6 percentage points, from a base rate of 70%. These estimates imply that the desire to work is highly responsive to the availability of alternative forms of health insurance. For every 100 people who lost coverage, about 54 to 63 people entered the labor market. These estimates imply that perhaps a half million to a million people will choose not to work for pay in response to the ACA's new Medicaid eligibility threshold, reducing the national employment rate by about 0.3 to 0.6 percentage points.

Laura Dague (Texas A&M University), ThomasDeLeire (Georgetown University), and Lindsey Leininger (University of Illinois at Chicago) reached a similar conclusion about the impact of expanded health insurance availability in Wisconsin.6 In January 2009Wisconsin introduced the BadgerCare Plus Core Plan, a new public insurance program for childless adults with incomes below 200% of the FPL. Demand for the new program was intense and unsustainable, so on October 5, 2009, then-Governor Jim Doyle announced that October 9 was the last day to enroll. After that, applications would be placed on a waitlist. Dague and colleagues matched administrative records of individuals who enrolled or were placed on the waitlist with employment records from the state's unemployment insurance system. First, they compared the employment outcomes of people who just made the enrollment deadline with those who just missed it (and were on the waitlist). Second, they used data on all applicants, regardless of how close they were to the deadline, and modeled the difference in employment trends between people on the program and those on the waitlist. They found that employment rates were 0.9 to 9.0 percentage points higher (depending on the data and specification) among people on the waitlist. The average employment rate in this population is around 43%, so their estimates imply a decline of 2% to 21%.

Jonathan Kolstad (University of Pennsylvania) and Amanda Kowalski (Yale University) studied the precursor to the ACA, the 2006 Massachusetts health insurance reform, to understand the welfare consequences of the individual and employer health insurance mandates.7 Economic theory argues that levying taxes and using the proceeds to provide health insurance to uninsured individuals leads to a loss of economic efficiency, since taxes discourage work. Lawrence Summers's (1989) seminal work argued that mandates can be a more efficient mechanism to increase insurance coverage.8 Kolstad and Kowalski extended Summers's theoretical analysis and developed a method to empirically estimate the welfare consequences of the Massachusetts employer and individual mandates. They estimate that workers in Massachusetts who gained coverage as a result of the 2006 reform experienced earnings declines of about $5,350 per year. This is close to firms' average cost of health insurance, which they peg at about $6,105 per employee per year. That workers are willing to forgo wages to have EPHI suggests it is unlikely that many firms will opt to drop coverage and pay the associated penalty. Their analysis also indicates that the loss of efficiency from the dual mandates was on the order of only 2% of what the loss would have been if the insurance expansion had been financed by a tax levy.

In another Massachusetts-based study, Bradley Heim (Indiana University) and Ithai Lurie (U.S. Department of the Treasury) tested whether the increased availability of insurance affects job-to-job mobility, the decision to retire, and the decision to become self-employed. Economists have argued that tying health insurance to employment may prevent otherwise valuable job switches. Using a panel of tax returns from 2002 to 2010, Heim and Lurie compared mobility rates in Massachusetts to those in neighboring states and estimated the reform reduced overall job separations by about 1.5 to 3.8 percentage points, or 5% to 16% of the initial level. This decline in separations is driven by people with income above 300% of the FPL and concentrated among people moving from one job to another, not from a decline in people moving from a job to selfemployment. Some subgroups experienced an increase in job mobility, such as those with incomes less than 300% of the FPL. These individuals were eligible for insurance subsidies and perhaps fit the expected pattern that increased availability of insurance allowed them to find more suitable employment.

Finally, two papers at the conference analyzed the labor market impact of the ACA. using calibrations of economic models. Rong Hai (University of Chicago) estimated the effects of the ACA's Medicaid expansion and employer mandate on the difference in overall health insurance coverage, wages, and employment rates between high school and college graduates.9 Over the past several decades, the earnings gap and the gap in EPHI between high school and college graduates have increased. Hai's model indicates that the ACA will significantly decrease the gap in health insurance coverage between high school and college graduates, but all of this reduction comes from an increase in Medicaid participation, not from an increase in EPHI. Her results imply that the Medicaid expansion will actually induce many lower-skilled people to work, as they can retain Medicaid coverage with higher earnings levels, thereby decreasing the gap in employment between high school and college graduates.

The ACA creates a number of important tax "wedges" that we described earlier. For example, the employer mandate creates an incentive for employers to retain fewer than 50 full-time workers or employ workers for fewer than 30 hours per week. The exchange subsidies create new implicit marginal taxes on earnings due to the income phaseouts. Moreover, the subsidies are contingent on EPHI not being available, creating an incentive for an employee to work part-time if an employer offers coverage only to full-time workers. In an analysis of the impact of these and other implicit tax wedges, Casey Mulligan (University of Chicago) found that the combined effects of the ACA have the same effect as a tax equal to about two hours' worth of wages per week for full-time workers and increase the average marginal tax rate on earnings by about 1.4 percentage points.10 About half of the working-age population, his results imply, will experience a reduced incentive to work and, thus, total employment will fall by about 3%.

The panel discussion

David Card, Class of 1950 Professor of Economics (University of California, Berkeley), provided a summary and critique, concentrating particularly on the Tennessee, Wisconsin, and Massachusetts papers. Card cautioned against drawing strong conclusions about negative employment effects or even reduced job lock due to the ACA, given the limited available evidence. He noted that a recent study in Oregon finds no impact of increased access to health insurance on employment or earnings.11 That study is particularly noteworthy because, unlike the "natural experiments" in Tennessee, Wisconsin, and Massachusetts, the Oregon evidence is based on a lottery in which households were randomly placed into either a treatment group that received Medicaid or a control group that did not. More generally, there is striking variance in the magnitude of the employment and job transition effects, both across the various studies and within each study across demographic groups, which remains unexplained. Card also pointed to the experience under the Medicare program. Despite the dramatic increase in access to public health insurance at age 65, there is very little change in employment between ages 64 and 65.

Jon Gruber, Ford Professor of Economics (Massachusetts Institute of Technology) and one of the key architects of the 2006 Massachusetts health reform and the ACA, emphasized that while there are volumes of credible research on the effect of health insurance on employment and job mobility, our understanding of the consequences for individual wellbeing and economic efficiency is quite limited. Economic welfare is reduced when individuals choose not to work in the job they most desire, or at which they are most productive, because of the lack of health insurance at that job. For example, the firm a person would want to move to may not offer insurance at all or may offer insurance but have a waiting period for new employees. Similarly, we do not know a priori what the welfare implications would be if the ACA were to induce some people to choose not to work because they can get affordable health insurance through new, nonemployment means. Gaining a better understanding of the welfare and efficiency implications of these effects will be crucial for determining the overall impact of the ACA. Gruber also emphasized that the taxes, subsidies, and income thresholds created by the new law, as well as the variation across states in whether Medicaid is expanded, will provide ample opportunities to learn about the effects of the ACA directly and establish whether the past experiences of Tennessee, Wisconsin, Massachusetts, and Oregon are indeed generalizable to related but different reforms.

Conclusion

Our goal in hosting this conference was to bring together many of the leading scholars studying the labor market effects of the ACA. Because of the complicated nature and incremental implementation of the law, it is no easy feat to predict how the ACA might ultimately impact labor markets. Indeed, one of the key lessons of the conference was the extent to which we are still very much in the dark about what ultimately might transpire. The conference highlighted the excellent research efforts already begun on this important agenda.

Copies of the papers presented and a video of the conference can be found at www.chicagofed. org/webpages/events/2014/ affordable_care_act.cfm.

Because of the complicated nature and incremental implementation of the law, it is no easy feat to predict how the ACA might ultimately impact labor markets.

1 An excellent summary is in appendix C of Congressional Budget Office, 2014, The Budget and Economic Outlook: 2014 to 2024, report, Washington, DC, February.

2 See http://kff.org/health-reform/stateindicator/ state-activity-around-expanding-medicaidunder-the-affordable-care-act/.

3 The premium must cover at least 60% of the actuarial value of the cost of benefits. Affordable is defined as not exceeding 9.5% of an employee's household income. The annual penalty is the lesser of $3,000 for each employee who receives a premium subsidy on the exchange or $2,000 per full-time employee, excluding the first 30.

4 See http://meps.ahrq.gov/mepsweb/ data_stats/summ_tables/insr/state/ series_2/2012/tiia2.htm.

5 See http://faculty.chicagobooth.edu/ matthew.notowidigdo/research/ggn_ tenncare_dec2013.pdf.

6 See www.chicagofed.org/digital_assets/ others/ events/ 2014/affordable_care_ act/chicago_fed_affordable_care_dague_ deleire_leininger.pdf.

7 See www.econ.yale.edu/~ak669/malabor. latest.draft.

8 Lawrence H. Summers, 1989, "Some simple economics of mandated benefits," American Economic Review, Vol. 79, No. 2, May, pp. 177-183.

9 See www.chicagofed.org/digital_assets/ others/ events/ 2014/affordable_care_act/ ronghai_affordable_care.pdf.

10See www.chicagofed.org/digital_assets/ others/ events/ 2014/affordable_care_act/ mulligan_economic_consequences_ health_care_reform.pdf.

"Katherine Baicker, Amy Finkelstein,Jae Song, and Sarah Taubman, 2013, "The impact of Medicaid on labor force activity and program participation: Evidence from the Oregon Health Insurance Experiment," National Bureau of Economic Research, working paper, No. 19547, October.

Daniel Aaronson, vice president and director of microeconomic research, and Darren Lubotsky, associate professor, Institute of Government and Public Affairs and Department of Economics, University of Illinois at Chicago