Many politicians, policy makers, and analysts have debated whether the Affordable Care Act (ACA) would have negative effects on the labor market, such as reducing employment, earnings, or hours worked. Building on the existing literature, we investigated how workers' coverage changed under the ACA and whether coverage gains were associated with changes in labor market outcomes across occupations through 2017. We also examined whether occupations experiencing increased coverage through nonemployment sources (i.e., Medicaid or individual plans purchased on the ACA's Marketplaces) also experienced offsetting declines in employer-sponsored insurance (ESI) coverage. Finally, we investigated whether the employer mandate was associated with changes in ESI offers to workers. Among workers in occupations experiencing larger coverage gains under the ACA, we found no evidence that employment, hours worked, or earnings fell relative to workers in occupations that had little change in coverage rates over the same period. Moreover, ESI offers remained stable, even among workers in firms likely subject to the employer mandate. Overall, we found that predictions that the coverage provisions and mandates of the ACA would lead to adverse labor market effects did not materialize.
Even before the Affordable Care Act (ACA) was fully drafted, business groups and others predicted the proposed legislation would cause job losses and harm the broader economy (Chow and Philips 2009; Tuerck et al. 2010). Representative John Boehner and other ACA opponents declared the law a job killer (Adair and Holan 2010; Diamond 2015; Boehner et al. 2011). When Republicans took control of the House of Representatives in 2011, they introduced the Repealing the Job-Killing Health Care Law Act (2011, H.R 2, 110th Congr.). Pundits argued the ACA would harm the same struggling workers and low-income families it was designed to help (Roy 2012). Although certain elements of the ACA could have affected the labor market in potentially negative ways, some analysts argued that these concerns were likely overblown at the time (Holahan and Garrett 2011).
Multiple provisions of the ACA were expected to expand workers' insurance coverage. First, starting in 2015, the employer mandate required firms with more than 100 full-time-equivalent (FTE) workers (defined as those who work at least 30 hours per week or 130 hours per month) to provide health coverage to full-time employees or pay penalties to the federal government if one or more workers claimed premium tax credits to purchase Marketplace coverage.1 In 2016, the employer mandate was fully phased in to apply to all firms with more than 50 FTE workers. Second, the individual mandate imposed a tax penalty on those who could afford coverage (as defined by the ACA's affordability standards) but remained uninsured. (The Tax Cuts and Jobs Act of 2017 eliminated the federal penalty for violations of the ACA's individual coverage requirement in 2019). Third, workers with incomes between 100 and 400% of the federal poverty level (FPL) could qualify for premium subsidies to purchase Marketplace insurance plans (and those with incomes between 100 and 250% also qualify for cost-sharing reductions)2 if they did not have affordable employer-sponsored insurance (ESI) offered to them.3 Finally, depending on their state of residence, workers with incomes below 138% of the FPL gained eligibility for health insurance through Medicaid.
Much of the concern about potential adverse employment effects of the ACA focused on the employer mandate (Puzder 2013; Harrison 2014). To avoid penalties, an employer with more than 50 FTE workers might cut back individual workers' hours or shrink its workforce to fall below the 50-FTE-workers threshold. The Congressional Budget Office (CBO), however, anticipated little effect of the employer mandate on employment. Instead, the CBO expected that the availability of premium subsides for Marketplace plans and increased Medicaid eligibility would provide an incentive for low-income employees to reduce the number of hours worked. The CBO estimated that the ACA would eventually reduce the total number of hours worked by 1.5 to 2.0%, mostly because workers would choose to work less in response to new insurance options presenting an alternative to their employer-based coverage. Another forecast, which garnered substantial attention at the time, held that the ACA would reduce aggregate hours worked by 3%, equivalent to 4 million full-time jobs (Mulligan 2014). Others predicted, based on review of existing evidence, that adverse employment effects would be minimal at most (Garrett and Kaestner 2014).
Another widely cited concern was that the ACA would harm workers by causing large numbers of employers to stop offering health benefits (Goozner 2012). Reflecting this concern in the first 2012 presidential debate, candidate Mitt Romney warned that up to 20 million Americans would lose their health insurance (Jacobson 2012). By providing subsidized insurance options outside of employment, it is possible that employer incentives to offer ESI could be reduced, particularly in smaller firms not subject to the employer mandate (Abraham, Graven, and Feldman 2012). In their main estimate, the CBO and the Joint Committee on Taxation initially predicted the ACA would slightly reduce employer coverage, with about 3 to 5 million fewer people obtaining coverage through an employer (CBO 2012). Revised estimates from the CBO in 2014 predicted that the ACA would, on net, result in about 7 million fewer people with ESI coverage by 2017 (CBO 2014). Urban Institute researchers estimated the ACA would have very little effect on employer coverage (Garrett and Buettgens 2011), while researchers at RAND estimated the law would increase employer coverage by 8 million (Eibner, Hussey, Girosi 2010).
Whether such incentives matter to the affected employers and workers and whether any behavioral effects apply broadly enough to make a meaningful difference to overall labor market outcomes are empirical questions. In this article, written a decade after the enactment of the ACA and using data from four years after the implementation of the major provisions of the ACA, we assess how workers' insurance coverage and work characteristics changed before and after the ACA. We use data from the 2010–17 American Community Survey, the 2010–17 Current Population Survey (CPS)’s basic monthly files, and the 2014–18 Current Population Survey's Annual Social and Economic Supplement files (CPS-ASEC) to investigate how the ACA affected workers' coverage, employment, usual hours worked, ESI and non-ESI coverage, and ESI offer rates. We assess trends in these outcomes among workers overall and across broad occupation categories that were likely differently affected by the ACA.
We find that workers' coverage greatly increased between 2010 and 2017 and differences in coverage rates among workers in different occupations narrowed after implementation of the ACA's major coverage provisions in 2014. We provide supporting evidence that these effects are predominantly driven by the ACA, ruling out factors such as demographic shifts in the population and economic recovery during this period. Despite these large shifts in coverage, we find that most predictions of reduced employment, work hours, and earnings as a result of the ACA did not come to fruition. Moreover, we show that making non-ESI coverage options (i.e., Medicaid or Marketplace coverage) available to nonelderly, nondisabled adults did not lead to fewer people having ESI coverage; in fact, occupations that saw increased non-ESI coverage after ACA implementation also saw increased ESI coverage. Finally, firms subject to the ACA's employer mandate showed little change in ESI offer rates, both overall and relative to firms likely exempt from the mandate.
Changes in Workers' Coverage
One of the primary goals of the ACA was to increase general coverage rates among nonelderly adults. An estimated 20 million people gained coverage following the initial years of implementation (Uberoi, Finegold, and Gee 2016). Among them, an estimated 9.5 million workers under age 65 and 5.2 million of their family members had gained coverage under the ACA through 2015 (Garrett, Gangopadhyaya, and Dorn 2017).
Using data on workers ages 19 to 64 from the American Community Survey, we provide updated estimates of how coverage changed among workers from 2010 to 2017 in table 1.4 Overall, we find that the share of workers with health insurance increased from 82.1% in 2010 to 89.2% in 2017 (a 7.1 percentage-point or 8.6% increase). Because some of this shift could reflect demographic changes, we also estimated changes in coverage adjusting for age, gender, race/ethnicity, education, industry, occupation, and workers' state of residence. The adjusted change in the coverage rate for all workers is similar to the change in the unadjusted rate (7.4 percentage points). Applying the adjusted rate to the total number of workers in 2017, we estimate that 10.6 million workers gained coverage through 2017.
Table 1 also shows changes in workers' coverage by occupation group, ordered from lowest to highest 2010 coverage rates. Coverage rates varied substantially across occupations in 2010, ranging from 48.7% for workers in farming, fishing, and forestry occupations to 96.1% for workers in architecture and engineering occupations. While workers in every occupation group gained coverage from 2010 to 2017, workers in occupations with lower initial coverage rates tended to experience larger coverage gains. For example, coverage for food preparation and serving workers increased by 19.7 percentage points (from 57.1 to 76.7%), while coverage for life, physical, and social science workers increased by only 2.2 percentage points (from 95.2 to 97.4%).
The scatterplot and linear fit in figure 1 demonstrate even more clearly the strong inverse relationship between coverage rates in 2010 (x-axis) and changes in coverage by occupation from 2010 to 2017 (y-axis). As a result, coverage differences across occupations narrowed under the ACA. The pattern also suggests that the coverage provisions of the ACA were well targeted to workers with the highest need. Occupations with lower overall coverage rates in 2010 also tended to have lower ESI coverage, wages, and earnings (Garrett, Gangopadhyaya, and Dorn 2017). Accordingly, workers in occupations with lower coverage rates in 2010 were more exposed to the ACA's coverage provisions (more likely to qualify for Medicaid or subsidies in the Marketplaces and less likely to be excluded from the Marketplaces on account of having an offer of ESI).
Factors other than the ACA might have contributed to greater coverage gains among workers in jobs with lower 2010 coverage rates. An estimated 8.7 million jobs were lost between 2007 and 2010 (CBPP 2019), and the prolonged but steady labor market recovery from the Great Recession could have led to increased coverage among workers and may have differentially increased coverage among those less likely to have had it in 2010. To examine these possibilities, figure 2 shows coverage rates by year for workers categorized into three groups: occupations with low 2010 coverage rates (i.e., fewer than 80% of workers covered), occupations with medium 2010 coverage rates (80 to 90% of workers covered), and occupations with high 2010 coverage rates (more than 90% of workers covered). From 2010 to 2013, coverage rates remained flat for all three groups. There is a sharp increase in coverage rates after 2013 in each group as the ACA's major coverage provisions were implemented in 2014, with the steepest increases occurring among workers with the lowest 2010 coverage rates. Coverage rates continued to increase through 2016, after which time they leveled off.
The patterns in figure 2 are consistent with a causal effect of the ACA. If differences by occupation group were primarily driven by factors related to general economic recovery, we would have expected to see a gradual narrowing in coverage rates across occupations throughout this panel. The unemployment rate peaked at 9.8% in 2010, fell to 7.4% by 2013, and was 4.4% in 2017. Although we observe a slight increase in coverage from 2010 to 2013 for workers in occupations with the lowest 2010 coverage rates, it is very small compared with the change following ACA implementation. Figure 2 clearly indicates that it was the ACA, not the improving economy, that dramatically increased workers' coverage and reduced differences in rates of insurance coverage across job types.
We conduct more formal analysis in appendix C (see online appendixes) using difference-in-differences and event study analysis. These approaches test whether coverage for occupations with higher 2010 uninsured rates increased relative to occupations with lower 2010 uninsured rates after 2014 and whether these differences only emerge after 2014. If occupations with different uninsured rates were diverging in coverage prior to the ACA, this would suggest they are not suited for being compared to one another.5 Results from both of these tests indicate that occupations with higher uninsured rates saw greater gains in coverage after 2014, and that any divergence in coverage before 2014 was very small.
Changes in Employment, Hours Worked, and Earnings
The idea that the ACA would cause employers to reduce workers' hours in large numbers caught fire in the business community. CNBC host Maria Bartiromo declared the ACA was turning the US into a “part-time employment country” (Greenberg 2013; Pickert 2013). The newspaper Investor's Business Daily compiled a list of employers who purportedly reduced worker hours because of the employer mandate (Graham 2014). In a survey conducted by Mercer in 2012, 51% of employers that did not currently offer coverage to all employees working 30 hours or more per week said they would likely change workforce strategy so that fewer employees worked 30 or more hours per week (Watts and Gaertner 2013).
The CBO projected the ACA would result in the loss of about 2.0 million full-time workers and reduce compensation by about 1%, and that these effects would likely be concentrated among low-wage workers. These losses were expected to occur mostly because of a decrease in labor supply—that is, the availability of subsidized Medicaid and Marketplace coverage would induce a large number of workers to choose to work less (CBO 2014).6 The CBO also predicted that the employer mandate could lead some firms to decrease employees' earnings and wages to pay for ESI, further decreasing the incentive to work.7 Alternatively, the employer mandate could lead affected firms to attempt to reduce their workforce below the 50-FTE-worker threshold to avoid penalties.
In total, we estimate that the number of workers increased by 9.6% (from 130.8 million in 2010 to 143.4 million in 2017), even as 9.9 million workers gained coverage (see appendix table A1 in the online appendixes). Employment levels rose in some occupations and fell in others, but the changes bear little relation to the coverage gains workers experienced under the ACA. Panel A of figure 3 shows that there is a small, negative, but statistically insignificant correlation between the changes in coverage rates from 2010 to 2017 and the percent change in the number of workers over the same period.
Average hours worked slightly increased among workers over the period, from 38.9 to 39.2 hours per week. If the ACA reduced work effort, we would expect to see larger declines (or smaller increases) in hours worked among workers in occupations more affected by the ACA. In panel B of Figure 3, we see the opposite. Hours worked per week tended to increase more (or fall by less) in occupations with larger gains in coverage from 2010 to 2017.8 Similarly, in panel C, we see that weekly earnings increased more in occupations with larger coverage gains under the ACA. These findings stand in stark contrast with predictions that the ACA would have an adverse effect on labor supply.
One limitation of these findings is that they do not clearly identify what would have happened in the absence of the ACA. Although we see no significant relationship between employment and the ACA and modest growth in hours worked and earnings after ACA implementation, there may have been stronger employment growth and greater gains in hours worked and earnings absent the ACA. To investigate this possibility, we again compare annual trends in these outcomes across occupations with different 2010 coverage rates.
Panel A of figure 4 shows the percent change in the number of overall workers relative to 2010 for occupations with low, medium, and high coverage rates in 2010. The number of workers in occupations with less than 80% of workers covered in 2010 who were most affected by the ACA grew steadily over the period. The post-ACA trend is a continuation of the pre-ACA trend with no apparent break related to the ACA. Those in occupations with 80 to 90% of their workers covered in 2010 (i.e., less affected by the ACA than the first group) saw stagnant growth in their workforce over the entire period, with no substantial break in trend coinciding with the ACA. Occupations with more than 90% of workers covered saw steady growth that accelerated after 2014.
Though none of the three groups saw adverse trends in employment, it is possible that employment could have increased by even larger amounts in the absence of the ACA. In appendix table C2 (see online appendixes), we test whether growth in the number of workers by occupation was significantly related to 2010 uninsured using a difference-in-differences approach. We find that occupations that were more exposed to the ACA because of their 2010 uninsured rates did have lower growth in the numbers of workers relative to lesser exposed occupations, but this estimate is very small in magnitude and statistically insignificant.9 Moreover, the stagnant growth in employment of the middle group (occupations covering between 80–90% of workers in 2010) relative to the steady growth in employment in jobs with low and high 2010 coverage rates reflects the well-documented long-term secular trend in the hollowing out of middle-skill jobs in the United States (Autor 2019).
Panels B and C of figure 4 show that trends in usual hours worked per week and real weekly earnings remained unchanged over the study period and were similar across occupations with different levels of 2010 coverage rates. Across all three occupational categories, we observe no large differences in the time trends of hours worked per week. The category most exposed to the ACA (with fewer than 80% of workers covered in 2010) even shows a very slight increase in hours worked, rather than the predicted decrease (panel B). Workers in occupations with more than 90% of workers insured, who were less affected by the ACA, saw almost no growth in real earnings during this period, whereas the other two groups saw slight increases (panel C).
Altogether, this evidence shows no indication that workers, in aggregate, chose to work fewer hours in response to the availability of subsidized coverage options, or that firms restricted employees' work hours as a way of circumventing workers' eligibility for ESI under the employer mandate. Furthermore, we find no evidence suggesting that the ACA was associated with a reduction in workers' earnings. While the ACA clearly increased workers' coverage rates, four years into implementation of its main coverage provisions the ACA did not carry the adverse labor market effects that many had predicted.
Other research investigating specific elements of the ACA or specific populations also found very little or no evidence of adverse employment effects. The ACA Medicaid expansion had no impact on employment or hours worked among parents and childless adults (Leung and Mas 2016; Kaestner et al. 2017) or rates of retirement among near-elderly workers approaching Medicare eligibility at age 65 (Levy, Buchmueller, Nikpay 2018), populations where the biggest labor supply responses were anticipated.
Among studies that looked more finely at whether the ACA had any effect on part-time work, there is no clear evidence that workers in firms large enough to be subject to the ACA's employer mandate were more likely to work fewer than 30 hours per week. Moriya, Selden, and Simon (2016), Mathur, Slavov, and Strain (2016), and Garrett, Kaestner, and Gangopadhyaya (2017) present evidence contrary to such an effect, while Dillender, Heinrich, and Houseman (2016) and Even and McPhereson (2018) find the ACA increased involuntary part-time work. Most recently, Valletta, Bengali, and van der List (2018) examined the determinants of involuntary part-time employment. Although they found an elevated level of involuntary part-time work following the Great Recession, they attributed it to slow-changing structural factors, rather than institutional factors such as the ACA, and cite research finding a similar pattern in other countries including the United Kingdom.
Changes in Employer-Sponsored Coverage and Non-Employer Coverage
The ACA increased workers' access to coverage through Medicaid and set up Marketplace coverage that workers could buy, possibly with a subsidy, if their employer did not offer affordable coverage. These sources of non-ESI coverage may have reduced workers' demand for ESI coverage and led some employers to discontinue offering such coverage. On the other hand, the employer mandate may have increased ESI coverage by encouraging employers with more than 50 FTE employees to start offering coverage to avoid financial penalties. Also, the individual mandate may have increased both workers' demand for ESI (possibly leading more employers to offer it) and workers' take-up of ESI offers. The CBO had predicted that by 2017 there would be 7 million fewer people with ESI coverage than there would have been absent the ACA (CBO 2014).
ESI coverage has remained stable since the ACA went into effect (US Census Bureau 2019; Shartzer, Blavin, and Holahan 2018). Previously, we documented that occupations with the most growth in non-ESI coverage also had the largest growth in ESI coverage (Gangopadhyaya, Garrett, and Dorn 2018). We update that analysis in Figure 5, which shows the relationship between the change in ESI and non-ESI coverage for each occupation. Of note is the positive relationship between an occupation's change in ESI coverage and non-ESI coverage. Occupations that most benefitted from the ACA's Medicaid expansion and Marketplace provisions, from a coverage standpoint, experienced greater growth in ESI coverage as well.
Figure 6 shows the trends in ESI and non-ESI coverage from 2010 through 2017 by occupations with different 2010 coverage rates. Occupations with the lowest 2010 overall coverage rates also had the lowest ESI coverage rates in 2010 (panel A); however, there is a clear uptick in ESI coverage starting in 2014. This small increase in ESI coverage rates is not observed for the other two groups of occupations that had higher 2010 coverage rates. In appendix table C3 (see online appendixes), we test the relationship between occupations with different 2010 uninsured rates and changes in ESI coverage rates. We find that occupations with higher 2010 uninsured rates had a positive and statistically significant increase in ESI coverage growth after ACA implementation. However, in panel A of appendix figure C3, this effect appears to be a continuation of a longer secular trend in ESI growth among occupations with higher 2010 uninsured rates relative to occupations with lower 2010 uninsured rates that began before ACA implementation.
Occupations with the lowest 2010 coverage rates are also those with the lowest average weekly earnings—these are the workers most likely to qualify for Medicaid or subsidized Marketplace coverage. It is therefore unsurprising to find that the steepest increase in non-ESI coverage following the ACA's implementation was among occupations with the lowest 2010 coverage rates (panel B). Despite these large changes in non-ESI coverage, we find no evidence that this came at the expense of ESI coverage.
Our findings that overall coverage gains, and both ESI and non-ESI coverage gains, were largest among workers with the lowest initial levels of coverage are consistent with a recent analysis of coverage changes by industry rather than occupation (Agarwal, Goldman, and Sommers 2019). Similarly, focusing on the effect of the Medicaid expansion, Kaestner et al. (2017) found that most estimates of “crowd-out” (i.e., the transition from private to public coverage) during this period were statistically insignificant, though negative in sign (i.e. Medicaid expansion is associated with reduced private insurance coverage) and small in magnitude. Altogether, there is little reason to believe that extending eligibility for Medicaid coverage to low-income nonelderly adults led to a meaningful change in employer-based coverage.
Because of the ACA's employer mandate, employers have limited capacity to decrease ESI offers, particularly in large firms. Smaller firms with fewer than 50 workers are unaffected by the mandate, but, by 2016, all firms with 50 or more employees faced financial penalties if they did not offer affordable ESI plans and if any of their workers received tax credits (premium subsidies) for Marketplace plans as a result. Note, however, that workers are ineligible for premium subsidies and cost-sharing reductions if they are offered affordable ESI coverage. Thus, there were considerable frictions built into the ACA to prevent workers from transitioning from ESI to Marketplace coverage.
We assess how ESI offer rates evolved after the ACA's employer mandate went into effect. Using data from the CPS-ASEC, we examine offer rates by firm size from 2014 (the year the CPS-ASEC started collecting ESI offer information for workers) to 2018 in figure 7. We present rates of ESI offers reported by workers in firms with more than 100 employees (first subject to the employer mandate in 2015), 50 to 99 employees (first subject to the employer mandate in 2016), and fewer than 50 employees (exempt from the employer mandate). Figure 7 shows unchanged ESI offer rates across all three firm sizes, suggesting that the employer mandate was not associated with changes in ESI offer rates. These findings are consistent with other research documenting “largely stable” ESI offer rates following ACA implementation (Abraham, Royalty, and Drake 2016).
Workers have fared well under the ACA. Between 2010 and 2017, workers in all occupations experienced increases in health insurance coverage, and in total we estimate 10.6 million workers gained coverage over this period. Workers in occupations with the lowest coverage rates in 2010, typically low-wage occupations, saw the largest coverage gains, causing coverage differences across occupations to narrow. Well-publicized and influential warnings and predictions that the ACA would substantially reduce employment, work hours, earnings, offers of employer coverage, and employer coverage rates did not materialize. Instead, these labor market outcomes generally improved for most occupations after the ACA's main provisions and mandates were implemented in 2014. Those occupations with the largest coverage gains from 2010 to 2017, the very occupations most affected by the coverage provisions of the ACA, saw no adverse effects on employment, weekly hours worked, or weekly earnings relative to occupations with the smallest coverage gains over the same period.
An analysis of why many of the more research-based predictions of adverse labor market consequences of the ACA did not come to fruition is beyond the scope of this paper. Among three studies that substantially informed the CBO's forecasts of reduced work effort due to the ACA's Medicaid expansions, one was an extreme outlier in a single state where Medicaid eligibility was restricted, causing a decline in enrollment. In weighing the same evidence, Garrett and Kaestner (2014) concluded that the labor market effects of the ACA would be minimal and argued why the estimates of the outlier study were implausible. On the question of whether the ACA affected part-time work, more research is needed to understand why different papers have reached different conclusions.
Some forecasts that the ACA would erode employer-based coverage may not have fully reflected several key considerations discussed by Garrett and Buettgens (2011). First, nearly 80% of workers with their own ESI coverage have incomes above 250% of the FPL. Firms would not have an incentive to drop coverage for these workers and would need to compensate them with higher wages if they did. Additionally, tax incentives continue to play an important role in making it worthwhile to offer ESI post-ACA. Lastly, findings of Gruber and Lettau (2004) suggest the preferences of higher-wage workers in a firm carry more weight and could temper any incentive to drop ESI. Except for small firms comprised largely of low-wage workers, incentives to offer ESI remain.
A common response to early work showing no or minimal adverse effects of the ACA on labor outcomes was that it takes time for these effects to play out. Now, with four years of postimplementation data, any major effects should be apparent. But we continue to see minimal effects. The additional analyses conducted, presented in appendixes B and C in the online appendixes, allow us to better draw causal inferences from the observed trends. From the range of studies we have examined and new analyses presented here, we find broadly improving labor market outcomes, substantial growth in workers' insurance coverage, and virtually no sign of the adverse labor market consequences that previously received so much concern. A decade after the ACA's enactment, millions of workers have gained coverage, while the adverse effects on labor markets forecasted by some analysts have not come to pass.
The authors appreciate helpful comments and feedback from John Holahan, Genevieve Kenney, Rachel Kenney, and Stephen Zuckerman. Support for this research was provided by the Robert Wood Johnson Foundation. The views expressed are those of the authors and should not be attributed to the Robert Wood Johnson Foundation or the Urban Institute, its trustees, and its funders.
For further details on the ACA's employer mandate requirements and penalties, see www.irs.gov/affordable-care-act/employers/employer-shared-responsibilityprovisions.
In states that expanded Medicaid, nonelderly adults with incomes between 138 and 400% of the FPL qualify for premium subsidies for Marketplace plans (those with incomes below 138% qualify for Medicaid).
Under the ACA, ESI affordability is defined as a plan providing minimum essential coverage at a cost that does not exceed an annual affordability threshold established by the IRS (9.78% of household income in 2020).
Worker refers to a person who is currently employed and excludes those who are seeking employment but are not currently employed. For additional detail on our data sources and approach, see appendix B in the online appendixes.
In appendix C, in the online appendixes, we proxy for each occupation's exposure to the ACA's coverage provisions and mandates by their 2010 uninsured rate (rather than their 2010 coverage rate). This is simply for convenience in interpreting the direction of estimates in these models. For example, when looking at changes in coverage, we expect occupations with higher 2010 uninsured rates to experience greater differences in coverage under the ACA relative to occupations with lower 2010 uninsured rates.
The CBO gave a caveat to this theoretical prediction by stating that in the first few years after ACA implementation, it was more likely that wages would not fully adjust and that labor demand effects from the employer mandate would outweigh labor supply effects (see appendix C of CBO 2014). Given that the CBO predicted that the major contributor to changes in overall employment, hours worked, and earnings would be labor supply, it is clear the CBO anticipated that these potential effects of the employer mandate would be relatively small.
Extraction workers were an outlier with a 3.8-hour decline in hours worked per week between 2010 and 2017. Between 2010 and 2017, coal mining employment decreased substantially. Oil and gas extraction peaked in 2014 before falling sharply—both trends may have contributed to reduced work hours in extraction occupations.
In appendix table C2, we find that if an occupation's 2010 uninsured rate was increased by 10-percentage points, thus increasing how targeted the occupation was by the ACA, it would be associated with a − 0.28% change in the number of workers after implementation. The average occupation had about 5.5 million workers in 2010, so this estimate is roughly equivalent to the loss of 15,000 workers. Moreover, this estimate is statistically indistinguishable from zero.