Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2016 Sep 3;73(4):713–722. doi: 10.1093/geronb/gbw115

Health Reform and Retirement

Helen Levy 1,✉, Thomas C Buchmueller 2, Sayeh Nikpay 3
PMCID: PMC6019028  PMID: 27591731

Abstract

Objective

To analyze whether there was an increase in retirement or in part-time work among older workers after January 2014, when new health insurance coverage options became available because of the Affordable Care Act (ACA).

Method

We analyze trends in retirement and part-time work for individuals aged 50–64 years in the basic monthly Current Population Survey from January 2008 through June 2016. We test for a break in trend in January 2014. We also test for differences in trends, both before and after 2014, in states that expanded their Medicaid programs in January 2014 under the ACA compared with those that did not.

Results

We find that there was no change in the probability of retirement or part-time work among older workers in 2014 and later, either overall or in Medicaid expansion states relative to nonexpansion states.

Discussion

Although many observers had predicted that an unintended consequence of health reform would be reduced labor supply, we find no evidence of this for older workers in the first 2.5 years after the law’s major coverage provisions took effect.

Keywords: Health care policy, Insurance, Labor force dynamics, Retirement

Most Americans obtain health insurance as a fringe benefit of employment (Smith & Medalia, 2014). Prior to the Affordable Care Act (ACA), few alternatives to employer-sponsored coverage were available for early retirees. This may have discouraged retirement before age 65, the age of near-universal eligibility for Medicare. Beginning in 2014, the ACA made alternatives to employer-sponsored health insurance available through two channels. First, the ACA established a health insurance marketplace for nongroup coverage in every state. These marketplaces pool risk, encourage price competition between insurers, impose minimum standards on the benefits provided, and administer substantial subsidies for individuals with family income below 400% of poverty. Combined with new rules that limit the allowable variation in premiums with regard to age and prohibit insurers from using health information to set premiums or deny coverage, marketplaces should substantially lower the cost of nongroup coverage for most early retirees. Second, about half of all states are taking advantage of an ACA provision that allows them to expand Medicaid coverage to low-income adults (those with family incomes below 138% of the poverty level, or about $22,000 for a couple in 2014). Taken together, these provisions imply a dramatic increase in the availability of affordable alternatives to employer-sponsored coverage for workers nearing retirement. To the extent that older workers had been experiencing “job lock”—that is, remaining in jobs only because those jobs provided health insurance—these new alternatives might be expected to reduce labor supply either by increasing retirement or by inducing a shift from full-time to part-time work.

The possibility of such reductions in labor supply has been one of the most politically contentious aspects of health reform. The nonpartisan Congressional Budget Office (CBO) projects that ACA will reduce hours worked by 1.5%–2.0% (the equivalent of 2.0–2.5 million workers) during the period 2017–2024 (Congressional Budget Office, 2014). CBO attributes this effect mainly to reductions in labor supply, as opposed to labor demand, but does not make detailed projections about how these reductions might be split between reductions in hours (e.g., shifts from full-time to part-time work) versus exits from the labor force (e.g., retirement), nor do they make projections for workers in specific age ranges. Nonetheless, given the relatively greater value of health insurance for older workers, it seems reasonable that the largest labor supply effects might be observed in this group, and there is some evidence in the existing literature, discussed in greater detail in the section “Background on Health Insurance and Retirement”, that this is the case (Dague, Deleire, & Leininger, 2014; Garthwaite, Gross, & Notowidigdo, 2013; Guy, Atherly, & Adams, 2012).

In this paper, we present evidence from the monthly Current Population Survey on trends in retirement through June 2016, 30 months after the ACA’s major coverage provisions took effect. Our main questions of interest are (a) whether labor supply of older Americans as a group declined after January 2014 in response to ACA provisions and (b) whether any decrease in labor supply after January 2014 was larger in states that expanded their Medicaid programs than in states that did not. Using linear regression models that allow for the possibility of a shift in January 2014 and later, we find no evidence of increases in retirement or part-time work among individuals aged 50–64 years in January 2014 through June 2016 compared with earlier years. We also find that there is no differential trend in either outcome in states that expanded Medicaid in January 2014 under the ACA compared with those that did not.

Background on Health Insurance and Retirement

A large literature analyzes the effect of health insurance on the retirement decision (Blau & Gilleskie, 2001, 2006, 2008; Fitzpatrick, 2014; French & Jones, 2011; Gruber & Madrian, 1995, 1996; Gustman & Steinmeier, 1994; Johnson, Davidoff, & Perese, 2003; Karoly & Rogowski, 1994; Leiserson, 2013; Lumsdaine, Stock, & Wise, 1996; Madrian & Beaulieu, 1998; Madrian, Burtless, & Gruber, 1994; Nyce, Schieber, Shoven, Slavov, & Wise, 2013; Robinson & Clark, 2010; Rogowski & Karoly, 2000; Rust & Phelan, 1997; Scholz & Seshadri, 2013; Shoven & Slavov, 2014; Strumpf, 2010). Nearly all of these papers find that the availability of insurance that is not contingent upon one’s own continued work—from Medicare, as a dependent on a spouse’s policy, from coverage intended for early retirees, or from COBRA—significantly increases the probability of retirement.

A related literature uses exogenous changes in eligibility for public insurance coverage to estimate the impact of insurance on labor supply, not necessarily restricting attention to older workers and the retirement decision. Studies that do estimate effects separately for older workers generally find significant labor supply responses to changes in public insurance coverage. One study found substantial increases in labor supply in response to cuts in Medicaid eligibility for childless adults in Tennessee in 2005, with the largest increases among individuals aged 40–64 years (Garthwaite et al., 2013). Another study, analyzing the expansion of Medicaid to childless low-income adults in Wisconsin in 2009, found significant reductions in labor supply, with the largest effects for individuals older than 55 years (Dague et al., 2014). An analysis of Medicaid expansions in 11 different states between 2001 and 2008 finds significant reductions in labor supply, with the largest effects for workers aged 55–64 years (Guy et al., 2012). There is some evidence that Massachusetts’ health insurance reform in 2007 increased retirement among full-time workers in Massachusetts (Heim & Lin, 2016).

In contrast, the Oregon Health Insurance Experiment suggests that there was no labor supply response to the expansion of Medicaid benefits for childless adults in Oregon in 2008, although this analysis is for adults of all ages, not only those close to retirement age (Baicker, Finkelstein, Song, & Taubman, 2014). Two recent studies of employment patterns before and after 2014, using a framework similar to the analysis we present here, also find no change in the probability of work beginning in 2014 (Gooptu, Moriya, Simon, & Sommers, 2016; Kaestner, Garrett, Gangopadhyaya, & Fleming, 2015); these papers, like the Oregon experiment, do not focus specifically on the decisions of older workers. To summarize the existing literature: there is mixed evidence on whether the availability of public health insurance significantly reduces labor supply; studies that stratify their analyses by worker age have found larger reductions for older workers than younger ones.

Background on the Major Coverage Provisions of the ACA

The ACA includes provisions to expand both private coverage and Medicaid that are intended to reach some of the 50 million individuals who were uninsured 2010 when the law was enacted (DeNavas-Walt, Proctor, & Smith, 2011). The Medicaid expansions target very low-income adults without dependents, whom we refer to henceforth as “childless adults,” although they may in fact have children who are not currently their dependents. Prior to the ACA, states covered low-income children and their families through Medicaid and the State Children’s Health Insurance Program (CHIP). Most states did not provide coverage for nonelderly, childless adults although a small number had done so (Kaiser Family Foundation, 2013). For example, Massachusetts implemented comprehensive reforms in 2007 that made subsidized insurance coverage available to all nonelderly adults with incomes below 300% of the poverty level (Long, 2008). The ACA allocated new federal funding for all states to expand Medicaid to all nonelderly adults under 138% of the federal poverty level. Six states took advantage of an option in the law to begin expanding their Medicaid programs prior to 2014, although two of these limited their early expansions to shifting individuals already covered by a state-funded program into Medicaid (Sommers, Arntson, Kenney, & Epstein, 2013; Sommers, Kenney, & Epstein, 2014). Although the ACA as enacted would have required the remaining states to expand their Medicaid programs in January 2014, a June 2012 Supreme Court ruling made the expansion optional. As a result, there is variation across states in when and whether they expanded Medicaid coverage for childless adults. Ten states and the District of Columbia had significantly expanded coverage for childless adults prior to 2014; this includes states that had raised the Medicaid eligibility threshold for childless adults to 100% of poverty or higher by 2011 (AZ, DC, DE, HI, NY, and VT), states that had adopted other comprehensive reforms affecting childless adults (MA and WI), and states that increased coverage substantially through the ACA’s Medicaid expansion prior to 2014 (CA, CT, and MN). In some cases, these early expansions were phased in gradually. In the remaining states, 15 expanded coverage sharply in January 2014, 6 expanded between January 2014 and June 2016, and 19 had not implemented expansion as of June 2016, when our analysis ends. Table 1 summarizes state decisions about Medicaid expansion to date.

Table 1.

Timing of State Medicaid Expansion for Childless Adults

(I) (II) (III) (IV)
“Early expanders” “Expansion states” “Late expanders” “Nonexpansion states”
Expanded before January 2014 Expanded January 2014 Expanded between January 2014 and June 2016 No expansion as of June 2016
Arizona Arkansas Michigan (April 1, 2014) Alabama
California Colorado New Hampshire (August 15, 2014) Pennsylvania (January 1, 2015) Florida
Connecticut Illinois Indiana (February 1, 2015) Georgia
Delaware Iowa Alaska (September 1, 2015) Idaho
Hawaii Kentucky Montana (January 1, 2016) Kansas
Massachusetts Maryland Louisianab
Minnesota Nevada Maine
New York New Jerseya Mississippi
Vermont New Mexico Missouri
Washington, DC North Dakota Nebraska
Wisconsin Ohio North Carolina
Oregon Oklahoma
Rhode Island South Carolina
Washingtona South Dakota
West Virginia Tennessee
Texas
Utah
Virginia
Wyoming
n = 11 states n = 15 states n = 6 states n = 19 states
Fraction of older individuals living in each group of states (%)
27.90 24.30 11.00 36.70
Open in a new tab

Notes: aAlthough New Jersey and Washington States also adopted early Medicaid expansion under the Affordable Care Act, their early expansions were limited and involved primarily or exclusively shifting individuals who had previously been enrolled in state-financed programs onto Medicaid (Sommers et al., 2014). Full expansion of Medicaid eligibility to all individuals below 138% of poverty did not occur in these states until 2014. Therefore, we code them as having expanded Medicaid in January 2014.

bExpansion in Louisiana went into effect in July 2016, so Louisiana is coded in our analysis as a nonexpansion state.

The law also implements private insurance market reforms beginning in January 2014, such as prohibiting plans from denying coverage or increasing premiums based on an applicant’s pre-existing condition. It establishes new health insurance marketplaces, sometimes called “exchanges,” which are intended to facilitate individuals’ plan choices by providing a website where enrollees can easily compare their plan options. Importantly, the law provides premium tax credits for families with income between 100% and 400% of poverty to purchase coverage through the marketplaces, provided that they do not already have access to Medicaid or coverage through an employer. The family’s share of the premium is determined on a sliding scale and is capped at between 2% and 9.5% of family income. Premiums for marketplace plans cannot vary based on health status, and the law limits allowable variation based on age, so that older enrollees cannot be required to pay more than three times what a younger enrollee would be charged for the same plan.

Collectively, these reforms should mean that beginning in 2014, older individuals face a much lower effective price for health insurance coverage options that do not depend on employment than they did before 2014. In addition, the means-tested health insurance subsidies in the ACA, like any means-tested subsidies, reduce the incentive to supply labor (Mulligan, 2015). Both of these effects should, in theory, lead older workers to work less. Thus, we might expect to see a reduction in labor supply among older workers in all states as a result of the availability of marketplace coverage and premium tax credits, with an even larger reduction in states that also expanded their Medicaid programs. Whether this is in fact what has happened is an empirical question, so we now turn to our analysis of data.

Method

Empirical Approach

Our main questions of interest are (a) whether labor supply of older Americans as a group declined after January 2014 in response to ACA provisions and (b) whether any decrease in labor supply after January 2014 was larger in states that expanded their Medicaid programs than in states that did not. To avoid the possibility that coverage expansions occurring prior to or shortly after January 2014 may obscure the effect of expansions that occurred in January 2014, our main analyses exclude the “early expander” and “late expander” states, and focus instead on the 15 states that expanded coverage in January 2014 (“expansion states”) and the 19 states that had not implemented expansion as of June 2016 (“nonexpansion states”).

We analyze trends in labor supply by estimating regression models of the following form:

Yist= b0+ b1·(year/month) + b2·(year≥2014)      + b3·(year/month)·(year≥2014)       + b4·Xi+ b5·(unemployment_ratest)     +b5·(calendar month dummies)

The subscript i indexes individuals, s indexes states, and t indexes time (year and month). This specification estimates a linear trend in the outcome, with the possibility of a jump in the intercept and a change in the slope occurring after January 2014. This is, in effect, an interrupted time series model. The coefficient on year/month is a linear trend that estimates the average monthly change in the outcome variable for years prior to 2014. The variable year/month is defined so that November 2013 is −1, December 2013 is 0, January 2014 is 1, and so on; this means that the coefficient on year ≥ 2014 measures whether there is a jump in the average level of the outcome variable in January 2014 and later. The coefficient on the interaction term (year/month)·(year ≥ 2014) allows the slope of the linear trend to differ in 2014 and later from the trend prior to 2014. The vector Xi includes additional characteristics: gender, race, education, marital status, and age. We also include controls for the monthly unemployment rate at the state level and calendar month dummies to controls for possible seasonality.

Our analysis begins in 2008. We choose this starting point because of evidence showing that the Great Recession, which began in December 2007, fundamentally shifted retirement planning and labor supply decisions among older workers (McFall, 2011; Munnell & Rutledge, 2013; Szinovacz, Martin, & Davey, 2013). We estimate separate models for two outcome variables: retirement and part-time work. In the retirement model, which is estimated using the full sample of individuals aged 50–64 years, the dependent variable Yist is 1 if individual i living in state s at time t is retired and 0 if s/he is not. The part-time model is estimated only for individuals aged 50–64 years who are employed; ideally, we would correct this model for selection into employment, but because we have no instruments for employment, we simply estimate the model for workers only.

We estimate separate models for expansion and nonexpansion states and pooled models that use data from all 34 states. The pooled models indicate whether there are changes in retirement and part-time work in 2014 for older individuals as a group. The expansion and nonexpansion models allow us to see whether there are different responses across the two groups of states, which may be attributable to Medicaid expansion. The expansion and nonexpansion models are estimated simultaneously to allow correlation in the error terms across the two types of states; we then test the equivalence of the key coefficients across the two models. In drawing inference from the comparison of the two models, it is important to establish that trends in the outcome were similar in the two groups of states before 2014. If they were not similar before 2014, this would suggest that labor markets in the two groups of states are fundamentally different from each other and it would be incorrect to infer that any changes were the result of Medicaid expansion. As we discuss in the section “Retirement Trends in Expansion Versus Non-Expansion States,” trends in the outcomes we study were very similar across the two groups of states prior to 2014.

All models are estimated using linear regression in order to facilitate the interpretation of the interaction terms, in light of the well-documented complexity of interpreting interaction terms in nonlinear models (Ai & Norton, 2003). In all models, standard errors are clustered at the state level, so they are robust to serial correlation (Bertrand, Duflo, & Mullanaithan, 2004). All analyses are carried out using Stata 14. Estimates are weighted using sampling weights provided with the data.

Data: Current Population Survey

The main data for our analysis come from the basic Current Population Survey (CPS), a monthly survey of households conducted by the Census Bureau for the Bureau of Labor Statistics. We use data from the period January 2008 through June 2016, the most recent data available. Each month, the sample includes about 150,000 individuals. We focus on individuals aged 50–64 years in order to capture the age range in which retirement is likely. At age 50 years, the age at which the Current Population Survey begins to query respondents only 2.5% of individuals in our sample are retired. At age 65, when half of all individuals report that they are retired, Medicare eligibility becomes near-universal, so that the availability of other sources of health insurance becomes much less important for the retirement decision. Moreover, the new coverage options under the ACA are, by design, not available to individuals aged 65 and older, so the availability of these new options in 2014 did not change marginal retirement incentives for individuals aged 65 and older. Therefore, we restrict our analysis to individuals aged 50–64 years, which gives us a sample of 16,000–18,000 individuals in each month in the 34 states in our analysis.

We measure retirement using the monthly labor force recode variable provided on the CPS public use data set (“PEMLR”). This variable takes on seven values: (1) employed—at work; (2) employed—absent from work; (3) unemployed—on layoff; (4) unemployed—looking for work; (5) not in labor force—retired; (6) not in labor force—disabled; and (7) not in labor force—other. For the variable “retired,” people who were coded in the CPS as “not in the labor force—retired” were coded as 1 and all others were coded as 0. For the variable “part-time,” employed respondents who worked less than 30 hours per week were coded as 1, employed respondents who worked 30 hours per week or more were coded as zero, and people who were unemployed or out of the labor force were excluded from the analyses of this dependent variable. The 30-hour threshold was chosen because it is the cutoff used to define full-time work in the ACA. Additional detail on the construction of these variables is provided in the Supplementary Material, in the section titled “Variable construction in the CPS.”

Table 2 shows unweighted sample counts and selected characteristics, by year, for our sample of all individuals who are aged 50–64 years as well as the subsamples living in expansion versus nonexpansion states. Men make up 48% of the sample throughout this period in both groups of states. About 40% of the sample has no education beyond high school; this fraction is slightly higher in nonexpansion states than expansion states. Our regression analyses will control for gender and education in addition to race, marital status, and age to address the possibility that differences in these characteristics across the two types of states may help explain for differences in labor supply.

Table 2.

Sample Characteristics of Older Individuals Aged 50–64 Years Overall and by State Medicaid Expansion Status Basic Monthly Current Population Survey

Male No education beyond high school Unweighted sample size
Medicaid expansion status as of January 2014 Nonexpansion Expansion Nonexpansion Expansion Nonexpansion Expansion
Year
 2008 0.481 0.486 0.444 0.415 104,130 84,361
 2009 0.480 0.485 0.434 0.416 107,306 85,596
 2010 0.481 0.487 0.438 0.418 109,179 86,214
 2011 0.481 0.487 0.436 0.417 109,537 86,439
 2012 0.483 0.484 0.431 0.410 109,423 86,440
 2013 0.481 0.486 0.433 0.410 109,531 87,292
 2014 0.480 0.483 0.429 0.411 113,622 86,468
 2015 0.482 0.481 0.425 0.400 117,853 83,001
 2016 0.481 0.482 0.418 0.400 59,490 40,213
Open in a new tab

Note: Estimates of characteristics are weighted using the Census-provided variable pwcmpwgt. Data for 2016 are for January through June only. All other years have 12 months of data. The sample excludes individuals living in states that expanded Medicaid before January 2014 or between February 2014 and June 2016 (see Table 1).

Results

Retirement Trends in Expansion Versus Nonexpansion States

Figure 1 shows the fraction of individuals aged 50–64 years who are retired in each month from January 2008 through June 2016 in expansion states versus nonexpansion states, defined using the categories shown in Table 1. These states contain 61% of all individuals aged 50–64 years. Figure 1 also includes error bars indicating 95% confidence intervals for a subset of the data points in each line (for visual clarity, we do not include all of the error bars). Leading up to 2014, the fraction of older individuals who were retired fluctuates but with no clear trend up or down. In 2014, there is no obvious increase in retirement either in absolute terms or in expansion states relative to nonexpansion states. Indeed, if there is any appearance of change after 2014 in Figure 1, it would be an increase in the fraction retired in nonexpansion states, contrary to what we would have expected. The 95% confidence interval in the figure spans a range of about ±1% point, so we can effectively rule out the possibility that the fraction of individuals in this age range who are retired increased by more than a percentage point.

Figure 1.

Figure 1.

Open in a new tab

Fraction of 50- to 64-year-old individuals who are retired. Source: Basic monthly CPS, January 2008 through June 2016.

In order to test more formally whether there is any break in trend in the probability that an individual is retired, we estimate the regression models described earlier. Table 3 reports the coefficients on year/month, year ≥ 2014, and the interaction between these two variables from models with the dependent variable equal to 1 if the individual is retired. Results are reported for the pooled sample of 34 states (Column 1), expansion states (Column 2), and nonexpansion states (Column 3). As described earlier, these regressions also include controls for calendar month, gender, education, marital status, race, age, and state-level unemployment rate; full regression results for all covariates are available in Supplementary Table A1.

Table 3.

Multivariable Regression Models: Selected Coefficients Dependent Variable = 1 If Retired; Sample Includes all Individuals Aged 50–64 Years

All states Medicaid expansion status Test of H0:
βexpansion = βnonexpansion
Expansion Nonexpansion
(1) (2) (3)
Year/month (linear) −0.000171 −0.000165 −0.000172 χ2 = 0.01
(0.000032) (0.000042) (0.000044) p = .9059
p = .000 p = .000 p = .000
Year ≥ 2014 0.004885 0.005048 0.004850 χ2 = 0.00
(0.003242) (0.003773) (0.004946) p = .9746
p = .141 p = .181 p = .327
Year/month*Year ≥ 2014 0.000209 −0.000008 0.000352 χ2 = 1.89
(0.000135) (0.000203) (0.000166) p = .1697
p = .131 p = .969 p = .034
Unweighted n 1,666,095 726,024 940,071
Open in a new tab

Note: Additional explanatory variables are gender, education, marital status, race, state-level unemployment rate, and calendar month. Columns 1–3 report regression coefficients, their standard errors (in parentheses), and p values. The final column reports tests of the null hypothesis that the coefficients in Columns 2 and 3 are equal.

The linear trend reported in Table 3 prior to 2014 (i.e., the coefficient on year/month) suggests that the probability of being retired was declining significantly but slowly; the coefficient in the pooled model, −0.000171 implies an annual decline in the probability of being retired of one fifth of a percentage point (0.000171 * 12 = 0.002). In all three models, the coefficient on the year ≥ 2014 dummy is positive but small and statistically insignificant, suggesting that there is no break in trend in January 2014 either overall or in either subgroup of states. The interaction term year/month*year ≥ 2014 is positive in the pooled model, suggesting that there may be a gradual increase in the probability of being retired after 2014. This coefficient is not significant at conventional levels, however. Moreover, the results in Columns 2 and 3 suggest that this result is driven by the nonexpansion states, which is not consistent with the idea that this change is due to coverage expansion. Finally, the tests reported in Column 4 of Table 3 indicate that we cannot reject that the coefficients for the expansion and nonexpansion states are the same, suggesting that Medicaid expansion had no significant effect on retirement.

Trends in Part-Time Work

As already noted, older workers may also respond to new health insurance options by switching to part-time work rather than outright retiring. To investigate this possibility, we present trends in the fraction of older workers who work less than 30 hours per week in expansion versus nonexpansion states in Figure 2. Each trend is accompanied by error bars indicating a 95% confidence interval for a subset of observations. In the figure, the fraction who are part-time increases gradually from 2008 through 2012, then gradually declines, in both expansion and nonexpansion states. The part-time rate is about 1.5 percentage points higher in expansion than nonexpansion states, and this gap does not change significantly in January 2014. These patterns are not consistent with the idea that coverage expansion—either at the national level or in expansion states relative to nonexpansion ones—led to an increase in part-time work among older workers. Regression coefficients reported in Table 4, with full model results available in Supplementary Table A2, confirm these patterns. In all three models, the linear trend is positive but small and insignificant prior to 2014. In January 2014, the probability of part-time work drops by a small and insignificant amount in the pooled sample and also in both subsets of states, as indicated by the coefficients on year ≥ 2014. The linear trend in the probability of part-time work flattens by a small and insignificant amount (the coefficient of negative 0.000137 on year/month·year ≥ 2014 in Column 1); looking at this coefficient separately in expansion versus nonexpansion states shows a slightly larger flattening, marginally significant with p = 0.060, in expansion states, while the trend in nonexpansion states flattens out by a slightly smaller and insignificant amount. This is, of course, the opposite of what we would have expected if Medicaid expansion had caused older workers to shift from full-time to part-time work. Finally, the tests reported in Column 4 of Table 4 indicate that we cannot reject that the coefficients for the expansion and nonexpansion states are the same, suggesting that Medicaid expansion had no significant effect on the probability of part-time work among older workers.

Figure 2.

Figure 2.

Open in a new tab

Fraction of workers aged 50–64 years who are part-time. Source: Basic monthly CPS, January 2008 through June 2016.

Table 4.

Multivariable Regression Models: Selected Coefficients Dependent Variable = 1 If Part-Time; Sample Includes Employed Individuals Aged 50–64 Years

All states Medicaid expansion status Test of H0:
βexpansion = βnonexpansion
Expansion Nonexpansion
(1) (2) (3)
Year/month (linear) 0.000049 0.000031 0.000067 χ2 = 0.24
(0.000035) (0.000049) (0.000054) p = .6236
p = 0.170 p = .520 p = .213
Year ≥ 2014 −0.002127 −0.002159 −0.003269 χ2 = 0.06
(0.001823) (0.003278) (0.003034) p = .8037
p = .252 p = .510 p = .281
Year/month*Year ≥ 2014 −0.000137 −0.000210 −0.000130 χ2 = 0.11
(0.000141) (0.000111) (0.000207) p = 0.7355
p = .337 p = .060 p = .529
Unweighted n 1,110,828 487,305 623,523
Open in a new tab

Note: Additional explanatory variables are gender, education, marital status, race, state-level unemployment rate, and calendar month. Columns 1–3 report regression coefficients, their standard errors (in parentheses), and p values. The final column reports tests of the null hypothesis that the coefficients in Columns 2 and 3 are equal.

Specification and Robustness Checks

We carried out a number of specification tests and robustness checks, which we discuss briefly here, with full results reported in the Supplementary Material.

First, as mentioned earlier, we tested to see whether trends in outcomes were similar in expansion and nonexpansion states prior to 2014. Between 2008 and December 2013, the linear trend in retirement was nearly identical in expansion versus nonexpansion states (as reported in Supplementary Table A3): −0.000166 in expansion states and −0.000174 in nonexpansion states. Trends in part-time work prior to 2014 were also nearly identical to each other (full results are available in Supplementary Table A4). The similarity of these trends prior to 2014 suggests that it may be reasonable to use the nonexpansion states as a “counterfactual” for the expansion states.

Second, we tried two alternative definitions of part-time work: less than 20 hours and less than 35 hours per week. This yielded results that were qualitatively very similar to those we found using a 30-hour cutoff. Full results are reported in Supplementary Tables A5 and A6.

Third, we re-estimated the models reported in Tables 3 and 4 using the full sample of all 50 states plus the District of Columbia. We included “early expanders” in the expansion group and “late expanders” in the nonexpansion group. Full results from these models, which are qualitatively very similar to the main results reported in Tables 3 and 4, are reported in Supplementary Tables A7 and A8.

Finally, we re-estimated the models restricting the sample to individuals with no education beyond high school, who make up just more than 40% of the sample. The rationale for this restriction is that these individuals are more likely to be income-eligible for Medicaid or tax credits, or at least close enough that additional reductions in labor supply would make them eligible. Based on this rationale, other papers analyzing labor supply responses to the ACA coverage provisions have restricted their attention either to very low-income individuals or those with a high school education or less (Gooptu et al., 2016; Kaestner et al., 2015). Although this restriction changes some of the coefficients, the overall pattern of results (reported in Supplementary Tables A9 and A10) is very similar to what was reported in Tables 3 and 4. Therefore, we conclude that our main finding of little effect on labor supply persists even in a sample of lower-skilled individuals for whom labor supply disincentives may be most salient.

Discussion

We find no evidence of an increase in retirement or a shift to part-time work among older workers during the first 2.5 years in which the ACA’s new alternatives to employer-sponsored coverage were widely available, either overall or in states that expanded their Medicaid programs on January 1, 2014 compared with those that did not. It would be premature to conclude, however, that the labor supply disincentives embodied in the ACA are not relevant for older workers. We have three reasons for sounding a note of caution in interpreting our results. The first is that, at the aggregate level, we simply do not know what would have happened had the ACA’s coverage provisions not been introduced in January 2014; there is no “counterfactual” that indicates what would have happened had the full package of health insurance marketplaces, premium tax credits, and Medicaid expansion not been implemented. We are on somewhat firmer ground, in terms of drawing causal inference, when we compare the experiences of states that did and did not expand their Medicaid programs; our analysis on this score suggests that Medicaid expansion, at the margin, has so far had little effect on retirement.

A second reason for caution is that our data measure the prevalence, rather than the incidence, of retirement. That is, we observe the fraction of individuals aged 50–64 years who are retired in repeated monthly cross-sections, rather than the probability of a transition from work into retirement. Even if the probability of work-to-retirement transitions increased following 2014 (or in expansion states compared with nonexpansion states), it might take some time for this to affect appreciably the fraction of older individuals who are retired.

A final reason for caution is the extraordinary uncertainty that has characterized the early years of Obamacare, which may have led prospective retirees to exercise caution in relying on ACA coverage so far. Well-publicized obstacles to enrollment in health insurance marketplaces in the first open enrollment period in late 2013 and early 2014 might have led some older workers to adopt a wait-and-see attitude before abandoning employer-provided coverage for the relative uncertainty of the marketplaces—particularly because many workers in this age group may fear losing access to their current health care providers if they change health plans. Prospective retirees may also have been prudently waiting to see whether the marketplace tax credits survived significant legal challenges that were not resolved until a US Supreme Court ruling (King v. Burwell) in June 2015, well after the open enrollment period in late 2014 for marketplace coverage starting in 2015. Medicaid coverage, too, has been something of a moving target in some states. In Michigan, for example, the continuation of the state’s Medicaid expansion was contingent on the approval of a federal waiver which came very shortly before the deadline that would have required the expansion to be dismantled (Udow-Phillips, Fangmeier, Corneail, & Hirth, 2016). In Kentucky, the election of a new governor in 2015 has brought into question the future of one of the ACA’s most successful Medicaid expansions (Herman, 2015). The upcoming presidential election—which will occur during the 2016 open enrollment for coverage in 2017—now adds an additional layer of uncertainty to the retirement calculus for older workers.

Workers close to retirement may simply not feel that it is worth the risk of trying these new and relatively untested options. Depending on the resolution of current uncertainty, the ACA’s reforms may become more firmly established and more familiar. In that case, the availability of subsidized coverage that is not tied to employment may still lead to increases in early retirement or shifts to part-time work among older workers, and future research using a longer time series of data will be necessary to monitor this possibility.

Supplementary Material

Supplementary data are available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online.

Funding

This work was supported by grants from the Social Security Administration through the Michigan Retirement Research Center (Grant RRC08098401) and the National Institute on Aging (NIA K01AG034232 to H. Levy). The findings and conclusions expressed are solely those of the authors and do not represent the views of the Social Security Administration, any agency of the federal government, or the Michigan Retirement Research Center.

Author contributions

All three authors planned the study. S. Nikpay and H. Levy carried out the data analysis. All three authors reviewed and discussed the results. S. Nikpay and H. Levy drafted the paper. T. Buchmueller contributed to revising the paper.

Conflict of Interest

The authors have no conflicts of interest to declare.

Supplementary Material

Supplementary Materials
Click here for additional data file. (102KB, docx)

References

  1. Ai C., & Norton E. C (2003). Interaction terms in logit and probit models. Economics Letters, 80, 123–129. doi:10.1016/S0165-1765(03)00032-6 [Google Scholar]
  2. Baicker K. Finkelstein A. Song J., & Taubman S (2014). The impact of medicaid on labor market activity and program participation: Evidence from the Oregon Health Insurance Experiment. American Economic Review, 104, 322–328. doi:10.1257/aer.104.5.322 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bertrand M., Duflo E., & Mullainathan S (2004). How much should we trust differences-in-differences estimates?. Quarterly Journal of Economics, 119(1), 249–275. [Google Scholar]
  4. Blau D. M., & Gilleskie D. B (2001). Retiree health insurance and the labor force behavior of older men in the 1990s. Review of Economics and Statistics, 83, 64–80. doi:10.1162/003465301750160045 [Google Scholar]
  5. Blau D. M., & Gilleskie D. B (2006). Health insurance and retirement of married couples. Journal of Applied Econometrics, 21, 935–953. doi:10.1002/jae.921 [Google Scholar]
  6. Blau D. M., & Gilleskie D. B (2008). The role of retiree health insurance in the employment behavior of older men. International Economic Review, 49, 475–514. doi:10.1111/j.1468-2354.2008.00487.x [Google Scholar]
  7. Congressional Budget Office (2014). The Budget and Economic Outlook: 2014 to 2024. Washington, DC: Congressional Budget Office. [Google Scholar]
  8. Dague L. DeLeire T., & Leininger L (2014). The effect of public insurance coverage for childless adults on labor supply. Cambridge, MA: National Bureau of Economic Research. [Google Scholar]
  9. DeNavas-Walt C. Proctor B., & Smith J (2011). Income, poverty, and health insurance coverage in the United States: 2010. Current Population Reports, Series P60, No. 239. Washington DC: US Government Printing Office. [Google Scholar]
  10. Fitzpatrick M. D. (2014). Retiree health insurance for public school employees: Does it affect retirement?Journal of Health Economics, 38, 88–98. doi:10.1016/j.jhealeco.2014.03.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. French E., & Jones J. B (2011). The effects of health insurance and self-insurance on retirement behavior. Econometrica, 79, 693–732. doi:10.3982/ECTA7560 [Google Scholar]
  12. Garthwaite C. Gross T., & Notowidigdo M. J (2013). Public health insurance, labor supply, and employment lock. Cambridge, MA: National Bureau of Economic Research. [Google Scholar]
  13. Gooptu A. Moriya A. S. Simon K. I., & Sommers B. D (2016). Medicaid expansion did not result in significant employment changes or job reductions in 2014. Health Affairs, 35, 111–118. doi:10.1377/hlthaff.2015.0747 [DOI] [PubMed] [Google Scholar]
  14. Gruber J., & Madrian B. C (1995). Health-insurance availability and the retirement decision. American Economic Review, 85, 938–948. doi:10.3386/w4594 [Google Scholar]
  15. Gruber J., & Madrian B. C (1996). Health insurance and early retirement: Evidence from the availability of continuation coverage. Advances in the economics of aging (pp. 115–146). Chicago, IL: University of Chicago Press. [Google Scholar]
  16. Gustman A. L., & Steinmeier T. L (1994). Employer-provided health insurance and retirement behavior. Industrial & Labor Relations Review, 48, 124–140. doi:10.1177/001979399404800109 [Google Scholar]
  17. Guy G. P. Atherly A., & Adams E. K (2012). Public health insurance eligibility and labor force participation of low-income childless adults. Medical Care Research and Review, 69, 645–662. doi:10.1177/1077558712457050 [DOI] [PubMed] [Google Scholar]
  18. Heim, B. T., & Lin, L. (2016). Does health reform lead to an increase in early retirement? Evidence from Massachusetts. ILR Review, 0019793916653965. [Google Scholar]
  19. Herman B. (December 30, 2015). New Ky. governor to overhaul Medicaid expansion. Modern healthcare Retrieved January 30, 2016 from http://www.modernhealthcare.com/article/20151230/NEWS/151239995
  20. Johnson R. W. Davidoff A. J., & Perese K (2003). Health insurance costs and early retirement decisions. Industrial & Labor Relations Review, 56, 716–729. doi:10.1177/001979390305600410 [Google Scholar]
  21. Kaestner R. Garrett B. Gangopadhyaya A., & Fleming C (2015). Effects of ACA Medicaid expansions on health insurance coverage and labor supply (No. w21836). National Bureau of Economic Research. [DOI] [PubMed] [Google Scholar]
  22. Kaiser Family Foundation (Producer). (2013, September 25, 2015). Medicaid: A Primer Retrieved from www.kff.org
  23. Karoly L. A., & Rogowski J. A (1994). The effect of access to post-retirement health insurance on the decision to retire early. Industrial & Labor Relations Review, 48, 103–123. doi:10.1177/001979399404800108 [Google Scholar]
  24. Leiserson G. (2013). Essays on the economics of public sector retirement programs (PhD thesis, Massachusetts Institute of Technology, Cambridge, MA: ). [Google Scholar]
  25. Long S. K. (2008). On the road to universal coverage: Impacts of reform in Massachusetts at one year. Health Affairs, 27, w270–w284. doi:10.1377/hlthaff.27.4.w270 [DOI] [PubMed] [Google Scholar]
  26. Lumsdaine R. L. Stock J. H., & Wise D. A (1996). Why are retirement rates so high at age 65?Advances in the economics of aging (pp. 61–82). Chicago, IL: University of Chicago Press. [Google Scholar]
  27. Madrian B. C., & Beaulieu N (1998). Does Medicare eligibility affect retirement? In Wise D. A. (Ed.), Inquiries in the economics of aging (pp. 109–131). University of Chicago Press. [Google Scholar]
  28. Madrian B. C. Burtless G., & Gruber J (1994). The effect of health insurance on retirement. Brookings Papers on Economic Activity, 1994, 181–252. doi:10.2307/2534632 [Google Scholar]
  29. McFall B. H. (2011). Crash and wait? The impact of the Great Recession on the retirement plans of older Americans. American Economic Review, 101, 40–44. doi:10.1257/aer.101.3.40 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Mulligan C. B. (2015). Side effects and complications: The economic consequences of health-care reform. University of Chicago Press. [Google Scholar]
  31. Munnell A. H., & Rutledge M. S (2013). The effects of the Great Recession on the retirement security of older workers. The Annals of the American Academy of Political and Social Science, 650, 124–142. doi:10.1177/0002716213499535 [Google Scholar]
  32. Nyce S., Schieber S. J., Shoven J. B., Slavov S. N., Wise D. A. (2013). Does retiree health insurance encourage early retirement?Journal of Public Economics, 104, 40–51. doi:10.1016/j.jpubeco.2013.04.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Robinson C., & Clark R (2010). Retiree health insurance and disengagement from a career job. Journal of Labor Research, 31, 247–262. doi:10.1007/s12122-010-9091-4 [Google Scholar]
  34. Rogowski J., Karoly L. (2000). Health insurance and retirement behavior: Evidence from the health and retirement survey. Journal of Health Economics, 19, 529–539. doi:10.1016/S0167-6296(00)00038-2 [DOI] [PubMed] [Google Scholar]
  35. Rust J., & Phelan C (1997). How social security and medicare affect retirement behavior in a world of incomplete markets. Econometrica, 65, 781–831. doi:10.2307/2171940 [Google Scholar]
  36. Scholz J. K., & Seshadri A (2013). Health insurance and retirement decisions. Michigan Retirement Research Center Research Paper (2013-292), Ann Arbor MI. [Google Scholar]
  37. Shoven J. B., Slavov S. N. (2014). The role of retiree health insurance in the early retirement of public sector employees. Journal of Health Economics, 38, 99–108. doi:10.1016/j.jhealeco.2014.03.013 [DOI] [PubMed] [Google Scholar]
  38. Smith J. C., & Medalia C (2014). Health insurance coverage in the United States: 2013. Washington, DC: US Department of Commerce, Economics and Statistics Administration, Bureau of the Census. [Google Scholar]
  39. Sommers B. D., Arntson E., Kenney G. M., Epstein A. M. (2013). Lessons from early Medicaid expansions under health reform: Interviews with Medicaid officials. Medicare & Medicaid Research Review, 3. doi:10.5600/mmrr.003.04.a02 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Sommers B. D. Kenney G. M., & Epstein A. M (2014). New evidence on the Affordable Care Act: Coverage impacts of early Medicaid expansions. Health Affairs, 33, 78–87. doi:10.1377/hlthaff.2013.1087 [DOI] [PubMed] [Google Scholar]
  41. Strumpf E. (2010). Employer-sponsored health insurance for early retirees: Impacts on retirement, health, and health care. International Journal of Health Care Finance and Economics, 10, 105–147. doi:10.1007/s10754-009-9072-4 [DOI] [PubMed] [Google Scholar]
  42. Szinovacz M. E. Martin L., & Davey A (2013). Recession and expected retirement age: Another look at the evidence. The Gerontologist. doi:10.1093/geront/gnt010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Udow-Phillips M. Fangmeier J. Corneail L., & Hirth R (2016). Medicaid expansion in Michigan: The Second CMS Waiver. Health affairs blog post Retrieved January 30, 2016, from http://healthaffairs.org/blog/2016/01/05/medicaid-expansion-in-michigan-the-second-cms-waiver/ on

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Materials
Click here for additional data file. (102KB, docx)

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES