Skip to content
View Featured Image

Health Insurance And The COVID-19 shock

Because the large majority of nonelderly U.S. households rely on employer-sponsored insurance (ESI) to pay for health care, any economic shock that destroys jobs also destroys access to health coverage. The COVID-19 crisis is one such shock, causing a near-total shutdown of huge swaths of the U.S. economy in March and April. While there was a large bounce back in job growth in May and June and, to a lesser extent, in July, the overall level of employment remains historically depressed relative to pre-COVID benchmarks.

Although the gold-standard data sources tracking changes in health insurance coverage will not be available until next year, imperfect but available data on job churn and net employment allows us to produce estimates of losses of health insurance coverage since the COVID-19 shock began. These estimates are more accurate than early-crisis estimates, and they account for job gains.

The following are key highlights from the report.

  • In any given month, churn in the labor market—some people losing jobs while other people gain them—means millions of workers newly gain or lose access to employer-sponsored health insurance (ESI) each month. For example, between 2015 and 2019, roughly 2.8 million workers gained access to ESI in each month while 2.7 million workers lost access, leading to a net increase in ESI coverage of just over 100,000 workers each month.
  • Extreme churn after February 2020 has led to very large losses in ESI coverage. In March and April, for example, new hiring led to 2.4 million workers gaining ESI coverage each month, but historically large layoffs led to 5.6 million workers losing coverage each month. This rate of lost coverage—over 3 million workers—dwarfs a similar calculation for the number of workers losing coverage each month during the biggest job-losing period of the Great Recession (September 2008–March 2009).
  • While the data documenting labor market churn data are useful, they do not provide the best estimates of ESI losses because they are not the most timely data, nor do they provide the best net measure of employment changes.
    • Since the onset of the COVID-19 shock to the economy, roughly 6.2 million workers have lost access to health insurance that they previously got through their employer, according to the best measure of net employment change. Our analysis using the monthly, high-quality measure of the total number of jobs in the economy from the Current Employment Statistics (CES) program of the Bureau of Labor Statistics (BLS) is consistent with 9 million workers have lost access to ESI in March and April 2020 but 2.9 million workers having gained coverage between April and July 2020.
  • Not every worker who loses ESI loses health insurance coverage. Public health insurance rolls are expanding to absorb the enormous ESI coverage losses of recent months. However, they have not expanded enough to absorb everybody who lost job-based coverage. A new government survey measuring the economic consequences of the COVID-19 shock in real-time indicates that for every 100 workers who were covered by ESI before losing their job, about 85 retained access to some form of health insurance in the week after they lost their job.
    • It is likely the case that Medicaid is the dominant alternative source of coverage when people have lost ESI in the COVID-19 shock, as Medicaid rolls have likely expanded by more than 4 million since the COVID-19 shock began.

Though we don’t yet know precisely how damaging the COVID-19 shock has been to health insurance access, the shock has laid bare the huge uncertainty that employer-linked health insurance introduces into U.S. families’ lives. Even in normal times millions of U.S. households must manage coverage transitions in a given month. During economic crises, these coverage changes increasingly include transitioning into uninsured status, which puts families’ health and financial security at risk.

Job market churn and the accompanying lapses in ESI provide one more argument for introducing a much larger role for public insurance in the U.S. health insurance system. (Past EPI reports have documented the health and financial benefits to U.S. households of moving away from our current system; see Bivens 2018, 2020.) An expanded public role could greatly reduce noncoverage and destabilizing insurance transitions and help restrain the rapid growth in health care costs (Bivens 2018). Expanded public insurance would also help the labor market function more efficiently during normal times (Bivens 2020).

In the very near term, policymakers at a minimum should make all testing and treatment for COVID-19 free of charge. In the longer term, U.S. families should be offered a much more robust public option for insurance coverage, with the goal of having this public option become the insurer of first resort for most. A bridge from the near term to the long term could be provided by significantly expanding the eligibility criteria for Medicaid during the current crisis.

Background and data on health insurance coverage during the COVID-19 crisis

Very early in the economic crisis caused by the coronavirus, we drew on the best real-time data then available to estimate how many workers likely had lost access to ESI because they were laid off.1 For a number of weeks, we updated our estimates using the weekly initial claims for unemployment insurance (UI), broken down by industry, to make this calculation. Months later in the crisis, there are much better measures of job loss—both gross and net—available to us than these initial UI claims, and so we are updating our methodology.

The gold-standard data source that will allow us to precisely estimate how many people have lost ESI (or any coverage) due to the COVID-19 crisis is more than a year away from being released. Generally, researchers consider the data set released every September from the Annual Social and Economic (ASEC) Supplement to the Current Population Survey (CPS) to be the gold standard for health insurance coverage trends. The ASEC asks households about their income and health insurance coverage over the previous year, so the 2020 data will not be available until September 2021. However, as described below, there are a number of available data sources that allow us to make decent (albeit quite imprecise) inferences about the likely loss of ESI and the likely loss of health insurance coverage of any kind due to the COVID-19 crisis.

Using job openings data to highlight insurance churn in normal times and during the crisis

The first data set we look at comes from the Job Openings and Labor Turnover Survey (JOLTS) from the Bureau of Labor Statistics (BLS). By tracking total new hires and total separations in relatively recent months by industry, this data set lets us look at likely gross ESI losses and gross ESI gains, highlighting the effect of the labor market churn on insurance coverage. (We say “relatively recent months” because the BLS releases the data for a given month not in the following month, but early in the month after that.) The industries we track in this data set constitute the large majority of total employment.

We combine this JOLTS data by industry with industry-level ESI coverage estimates from the most recent CPS-ASEC (2018 data released in 2019) to adjust the JOLTS numbers to account for the fact that not all jobs gained or lost come with health insurance, and that this share varies widely by industry. The full set of results by industry is available in Appendix Table 1. Figure A here shows the likely gross monthly gains and monthly losses in ESI coverage for four periods: the months spanning the end of 2008 and early 2009 that saw the most rapid job losses of the Great Recession (September 2008 through March 2009); the steadily improving labor market of 2015–2019; the period of historically rapid job losses in March and April 2020; and the rapid jobs bounceback in May and June 2020.

Figure A shows that in the five years spanning 2015 to 2019, gross employment gains each month were consistent with roughly 2.8 million workers gaining access to ESI each month, while gross employment losses were consistent with roughly 2.7 million workers losing access to ESI each month. The net of this churn was a steady, small increase of about 100,000 workers each month gaining ESI.In March and April of 2020, gross employment changes were consistent with 2.4 million workers gaining access to ESI each month but 5.6 million workers losing access to ESI, for a net monthly reduction of about 3.2 million workers with access to ESI. This pace of ESI loss dwarfed even that which likely occurred in the worst months of job loss during the Great Recession when net losses averaged 350,000 monthly. Finally, the rapid job bounceback in May and June 2020 likely saw an average of 1 million workers per month gaining access to ESI on net.This analysis tells us two things. First, the COVID-19 shock shows up strongly in this data, with the huge rise in separations (mostly layoffs) leading to very large increases in workers losing access to ESI in March and April of this year. Second, even during months in a steadily improving labor market (such as between 2015 and 2019), there is substantial churn in employer-based health insurance, with millions of working people gaining and losing access each month. All else equal, it seems likely that workers are not well served by this churn even if it does not lead to a net loss of insurance coverage, as it requires time for workers and their families to navigate the new set of benefits and often requires leaving a preferred doctor or set of providers.

Using timely data on jobs in the economy to infer ESI changes during the COVID-19 shock

While the JOLTS data have the useful property of tracking gross employment changes, they are not the most timely data, nor do they constitute the best available source of net employment changes. Instead, the Current Employment Statistics (CES) program of the Bureau of Labor Statistics (BLS) provides a monthly, high-quality measure of the total number of jobs in the economy. The JOLTS data tend to lag the CES data by roughly one month.

Using the CES, we can again match employment by industry with our estimates of ESI coverage by industry from the 2018 ASEC to infer likely changes in ESI coverage during the COVID-19 crisis. A full set of industry estimates is available in Appendix Table 2, but these results are summarized in Figure B. As with our analysis of the JOLTS data, the industries we track in this data set constitute the large majority of total employment.

This figure shows that job losses in March and April were consistent with roughly 9.0 million workers losing access to ESI. Between April and July, however, job gains were consistent with roughly 2.9 million workers gaining access to ESI. Given that the CES is our best timely measure of net employment changes, these inferences are probably our best measure for what the COVID-19 shock has done to ESI coverage. Between February and July (the full timespan of data coverage of the COVID-19 shock), the net result is a reduction of 6.2 million workers with ESI.

As ESI shrinks, is there a safety net for U.S. workers?

Those losing access to health insurance through their own employer may pay for continued temporary coverage under COBRA2 or qualify for public insurance (like Medicaid); switch to another employed family member’s policy or access insurance through the health insurance “marketplace exchanges” established by the Affordable Care Act (commonly known as “Obamacare”).

Data on coverage in public insurance or through the ACA marketplaces are less timely than employment data, so the best inference to be made about the ability of workers losing ESI to transition into these alternative forms of coverage is based on historical experience. Gangopadhyaya and Garrett (2020) look at this historical experience and provide estimates from 2014 to 2018 on health insurance coverage by employment status. Figure C summarizes some of their key findings, breaking down insurance coverage status for adults under the age of 65 who are employed, unemployed, or not in the labor force.

Unsurprisingly, the employed are far more likely to have ESI coverage than the unemployed (with those not in the labor force at all sitting somewhere in between). Unemployed workers and those not in the labor force are also far more likely to be covered by Medicaid. Perhaps surprisingly, the share covered through the ACA marketplace exchanges is pretty similar for all three groups.

These data indicate strongly that it is Medicaid, not the ACA marketplace exchanges, that does the heavy lifting of providing a health insurance safety net for those workers who lose ESI. Gangopadhyaya and Garrett (2020) also note that the share of the unemployed covered by Medicaid is substantially higher in the 35 states (including Washington, D.C., as a “state”) that accepted the Medicaid expansions offered by the ACA (35.8%) than it is in the 16 states that have still not accepted these expansions (16.4%). This 19.4 percentage-point gap in Medicaid coverage explains 97% of the 19.9 percentage-point gap in uninsurance rates for unemployed people in these groups of states (uninsurance among the unemployed averages 22.6% in states that have expanded Medicaid, but 42.5% in states that have not). In short, Medicaid is incredibly important—yet still underpowered—in providing a health insurance safety net for unemployed workers and their families.

Dorn (2020) uses these estimates, along with data on employment and unemployment from the Current Population Survey (CPS), to infer what has happened to overall rates of uninsurance during the COVID-19 shock. He finds that between February and May, 5.4 million workers likely lost health insurance due to layoffs.

Dorn notes a number of reasons why this 5.4 million is an imprecise estimate, largely because the experience of job loss during the COVID-19 shock might have different implications than job loss experienced during the recovering labor market of 2014–2018.3 One issue that might be important is that a large number of people who lost employment in the wake of the COVID-19 economic shutdown have been classified in the CPS data as not in the labor force rather than unemployed. This is likely a mistake even by the normal standards of the CPS reporting, and, from an economic point of view, anyone who lost a job due to COVID-19-related layoffs is almost certainly far more “like” the unemployed as a group than they are “like” those not in the labor force when it comes to how likely they are to be covered by health insurance.4

A new data set constructed precisely to measure the economic consequences of the COVID-19 shock is the Household Pulse Survey (HHPS) from the U.S. Census Bureau. The HHPS ran from the end of April until the middle of July. Its first data point, unfortunately, was collected after the worst of the job losses hit the U.S. economy. Nevertheless, the HHPS has been invaluable in measuring developments since this initial employment trough.

Table 1 presents results looking at a sample of workers who were employed and had access to ESI in the previous week. Among this sample, the survey tracks those who remained employed the next week, those who were no longer employed, and those who affirmed that their job loss was involuntary. It finds that of those who remained employed in the following week, about 99% reported retaining access to health insurance coverage, with 98% of these still-insured having access through ESI.5 Of those who lost jobs involuntarily, 85% report being covered by health insurance in the following week. This 15% loss in access to health insurance is the net result of 17.5% losing ESI and an increase of nearly 12 percentage points in the share of workers reporting access to public insurance (predominantly Medicaid). (Because the HHPS allows people to report more than one form of coverage, these numbers do not add up perfectly.)

The HHPS data is quite noisy and volatile, which is to be expected given its newness, small sample size, and very high frequency (collected weekly). The HHPS also does not identify who is the policyholder, so it is impossible to determine which job losers with ESI coverage have continuing coverage through their former employer or have switched to the policy of another employed family member. However, even in this data, the negative implications for retaining access to health insurance after losing a job are easily detectable.

Besides showing the sharp decline in access to ESI following job loss, the data in Table 1 also show a measurable uptick in the probability of laid-off workers reporting access to public insurance coverage. Figure D shows the share of all respondents to the HHPS who report having public insurance (not just those who were employed and with access to ESI in the previous week). The figure shows four-week moving averages of the share of the over-17 population covered by public health insurance. This measure has been rising steadily for the five weeks since mid-June, by a cumulative 4.3 million people.

This is roughly consistent with an analysis by Aron-Dine, Hayes, and Broaddus (2020). They find that in a subset of states with timely data, which account for roughly two-fifths of total Medicaid enrollments in normal times, Medicaid enrollment had risen by 2.0 million people between February and June 2020. Extrapolated over all the states, this would be roughly 5 million additional enrollees.

On the one hand, it is certainly encouraging that public health insurance rolls are expanding to absorb the enormous ESI coverage losses of recent months. However, they have not expanded enough to absorb everyone who lost job-based coverage. Our estimate from Figure B was that 6.2 million workers likely lost the health insurance that they had from their own employer between February and July. Given that for each person who is covered under their own employer’s ESI plan, roughly two people on average are covered, once spouses and dependents are included, this means that closer to 12 million people have been cut off from ESI coverage due to job losses in recent months.6

Conclusion

The inefficiencies and problems caused by the U.S. system of tying access to health insurance to specific jobs are well known. The downsides of employer-based health insurance access have been made spectacularly visible by the COVID-19 shock—a shock that has cost millions of Americans their jobs and their access to health care in the midst of a public health catastrophe.

Delinking access to health insurance from specific jobs should be a top policy priority for the long term. The most ambitious and transformational way to sever this link is to make the federal government the payer of first resort for all health care expenses—a “single-payer” plan. The federal government already is the primary insurer for all Americans over the age of 65 and for households with incomes low enough to qualify for Medicaid. The advantages of a single-payer system are large, both in ensuring consistent access to medical providers that households prefer and in restraining the often-rapid growth of health care costs.

Absent a once-and-for-all switch to a single-payer system, policymakers can take smaller steps to delink health insurance from specific jobs. They could lower the age of eligibility for Medicare, raise the income thresholds for Medicaid eligibility, and/or incorporate into the ACA marketplace exchanges a public option that enrolls all workers without job-based insurance—even those with access to ESI if they prefer the public option instead. Policymakers could also require that employers either provide comprehensive and affordable insurance or pay a fee to help cover the costs of enrolling their workers in the public option.

Finally, the lowest-hanging fruit in the current crisis is to have the federal government pay all expenses for COVID-19-related testing and treatment. Given the historically rapid increase in uninsurance in the first months of the COVID-19 shock, policymakers should also allow all those without insurance to enroll in Medicaid, regardless of income, for the duration of the crisis.

The COVID-19 shock has exposed just how incomplete and threadbare the U.S. safety net and social insurance system is. We should begin building a better set of systems that provide economic security to U.S. workers.

Appendix

APPENDIX TABLE 1

ESI coverage rates, hires, total separations, and net job changes, by industry, for select periods from 2008 to 2020

Hires, thousands per month Separations, thousands per month Net employment change, thousands per month
ESI coverage rates Sept. 2008–March 2009 2015–2019 average March–April 2020 May–June 2020 Sept. 2008–March 2009 2015–2019 average March–April 2020 May–June 2020 Sept. 2008–March 2009 2015–2019 average March–April 2020 May–June 2020
Mining 74.7% 24.9 29.0 16.0 13.5 31.7 31.8 54.5 29.0 -6.9 -2.9 -38.5 -15.5
Construction 44.5% 353.6 369.9 317.5 588.5 462.1 348.6 795.5 304.0 -108.6 21.3 -478.0 284.5
Manufacturing 69.3% 259.0 317.6 312.5 482.0 414.7 308.6 783.0 335.5 -155.7 9.0 -470.5 146.5
Wholesale 61.5% 125.7 141.5 126.5 174.0 163.9 138.7 266.5 170.0 -38.1 2.8 -140.0 4.0
Retail 40.8% 597.9 742.4 743.5 904.0 682.3 738.6 1,460.5 581.5 -84.4 3.8 -717.0 322.5
Transportation, warehousing, utilities 58.5% 133.4 217.6 211.0 270.5 164.6 201.7 453.5 244.5 -31.1 15.9 -242.5 26.0
Information 61.9% 63.6 86.5 56.5 61.5 79.4 84.2 117.5 57.5 -15.9 2.3 -61.0 4.0
Finance and insurance 70.5% 122.7 135.9 140.0 125.0 145.3 127.6 140.5 111.0 -22.6 8.3 -0.5 14.0
Real estate and leasing 46.5% 63.9 71.7 50.5 70.5 76.9 66.9 184.0 55.5 -13.0 4.8 -133.5 15.0
Professional and business services 61.0% 784.9 1,121.8 951.5 1,065.5 915.4 1,087.6 1,536.5 892.5 -130.6 34.2 -585.0 173.0
Education 61.6% 73.9 95.5 80.5 103.0 72.6 89.4 302.0 56.0 1.3 6.2 -221.5 47.0
Health 56.8% 443.9 578.5 524.0 854.0 414.9 538.2 1,233.0 503.0 29.0 40.3 -709.0 351.0
Arts and entertainment 37.4% 106.6 166.5 79.0 157.5 107.7 159.9 456.5 117.0 -1.1 6.6 -377.5 40.5
Food service and accommodation 23.9% 615.0 871.9 503.0 1,407.0 652.7 846.8 3,210.5 621.5 -37.7 25.1 -2,707.5 785.5
Other services 33.3% 169.3 211.7 166.5 419.0 188.0 206.6 895.0 150.5 -18.7 5.0 -728.5 268.5
Government 70.0% 278.0 351.3 301.0 252.5 282.1 339.6 419.5 268.5 -4.1 11.7 -118.5 -16.0
Total 4,216.0 5,509.1 4,579.5 6,948.0 4,854.3 5,314.8 12,308.5 4,497.5 -638.3 194.3 -7,729.0 2,450.5

Sources: Hires and total separations from BLS’s Job Openings and Labor Turnover Survey; employer-sponsored health insurance (ESI) coverage by industry from the Annual Social and Economic supplement to the Current Population Survey.

APPENDIX TABLE 2

ESI coverage rates and net payroll employment changes by industry, February–July 2020

Cumulative employment changes, thousands
ESI coverage Feb.–April 2020 April–July 2020 Feb.–July 2020
Mining 74.7% -57 -11 -68
Construction 44.5% -811 954 143
Manufacturing 69.3% -1,363 753 -610
Wholesale 61.5% -382 124 -258
Retail 40.8% -2,250 1,604 -646
Transportation, warehousing, utilities 58.5% -596 114 -482
Information 61.9% -274 -29 -303
Finance and insurance 70.5% -55 62 8
Real estate and leasing 46.5% -219 111 -108
Professional and business services 61.0% -2,031 809 -1,222
Education 61.6% -507 -322 -829
Health 56.8% -2,232 991 -1,241
Arts and entertainment 37.4% -1,146 693 -454
Food service and accommodation 23.9% -6,633 3,934 -2,699
Other services 33.3% -1,325 818 -507
Government 70.0% -878 -1,821 -2,699
Total -20,759 8,784 -11,976

Sources: Employment changes from the BLS’s Current Employment Statistics program; employer-sponsored health insurance (ESI) coverage rates by industry from Annual Social and Economic supplement to the Current Population Survey.

Endnotes

1. The methodology for these estimates is described in Bivens and Zipperer 2020. In the classification scheme of this report, those estimates were the gross losses to ESI in those opening weeks of the COVID-19 shock, with initial unemployment insurance (UI) claims serving as a proxy for the gross employment losses. The UI data is by far the timeliest indicator of labor market trends, which is why we (and so many other labor market analysts) relied on it very heavily in the opening weeks of the COVID-19 shock. However, this UI data is low quality as a measure of labor market churn, so as other data sources became available, we stopped using UI-based estimates of gross ESI losses.

2. The Consolidated Omnibus Budget Reconciliation Act, a federal law that concerns the continuation of group health care benefits after termination of employment.

3. For example, a large share of COVID-19-related layoffs so far have been temporary. Some small share of these workers might even continue to have access to ESI through their former employers, if these employers genuinely plan to bring workers back on staff as soon as the shock ends. Cutting the other way, the large number of hires in May and June of 2020 may well have led to fewer offers of ESI than normal, as risk-averse employers are unwilling to incur large fixed costs to rehire workers (i.e., these employers might be worried about the need to potentially shed workers in coming months in the face of virus resurgence).

4. See Shierholz 2020 on the issue of misclassification of unemployed workers in the CPS in recent months. Essentially, we think the very large majority of employment declines in March and April of this year were involuntary job losses, not voluntary withdrawals from the labor force. In this case, we think it makes sense to treat everyone who lost a job in March and April as functionally unemployed, not voluntarily out of the labor force.

5. The table reports that 96.7% have access to ESI the week after remaining employed. Since 98.7% report access to any form of coverage, ESI can explain (96.7/98.7)% of this retention, or 98%.

6. Roughly 160 million people in total were covered by ESI plans in 2018. There were roughly 155 million employed people in the United States before the COVID-19 shock. Common estimates of the share of employed workers who receive ESI through their own coverage hover between 45% and 55%, depending on exactly which groups of workers (full-time or part-time, under age 65 or over, for example) are examined. This implies roughly 80 million workers who have an ESI plan directly from their own employer. Given these numbers, each person covered by an ESI plan through their own employer must, by definition, be covering closer to two people overall.

References

Aron-Dine, Aviva, Kyle Hayes, and Matt Broaddus. 2020. With Need Rising, Medicaid Is at Risk for Cuts. Center on Budget and Policy Priorities, July 2020.

Bivens, Josh, and Ben Zipperer. 2020. “9.2 Million Workers Likely Lost Their Employer-Provided Health Insurance in the Past Four Weeks.” Working Economics Blog (Economic Policy Institute), April 16, 2020.

Bivens, Josh. 2018. The Unfinished Business of Health Reform: Reining in Market Power to Restrain Costs Without Sacrificing Quality or Access. Economic Policy Institute, October 2018.

Bivens, Josh. 2020. Fundamental Health Reform Like ‘Medicare for All’ Would Help the Labor Market: Job Loss Claims Are Misleading, and Substantial Boosts to Job Quality Are Often Overlooked. Economic Policy Institute, March 2020.

Bureau of Labor Statistics (BLS). 2020a. Job Openings and Labor Turnover Survey (JOLTS), interactive data on hires and separations by industry. Accessed August 2020.

Bureau of Labor Statistics (BLS). 2020b. Current Employment Statistics (CES), interactive data on payroll employment by industry. Accessed August 2020.

Census Bureau. Microdata from the Household Pulse Survey. Accessed August 2020.

Current Population Survey. Microdata from the Annual Social and Economic (ASEC) Supplement.

Dorn, Stan. 2020. The Covid-19 Pandemic and Resulting Economic Crash Have Caused the Greatest Health Insurance Losses in American History. Families USA, July 2020.

Gangopadhyaya, Anuj, and Bowen Garrett. 2020. Unemployment, Health Insurance, and the COVID-19 Recession. Urban Institute, April 2020.

Shierholz, Heidi. 2020. “Nearly 11% of the Workforce Is Out of Work with No Reasonable Chance of Getting Called Back to a Prior Job.” Working Economics Blog (Economic Policy Institute), June 20, 2020.

Urgent End Of Year Fundraising Campaign

Online donations are back! Keep independent media alive. 

Due to the attacks on our fiscal sponsor, we were unable to raise funds online for nearly two years.  As the bills pile up, your help is needed now to cover the monthly costs of operating Popular Resistance.

Urgent End Of Year Fundraising Campaign

Online donations are back! 

Keep independent media alive. 

Due to the attacks on our fiscal sponsor, we were unable to raise funds online for nearly two years.  As the bills pile up, your help is needed now to cover the monthly costs of operating Popular Resistance.

Sign Up To Our Daily Digest

Independent media outlets are being suppressed and dropped by corporations like Google, Facebook and Twitter. Sign up for our daily email digest before it’s too late so you don’t miss the latest movement news.