Introduction

The onset of the Coronavirus Disease 2019 (COVID-19) pandemic resulted in a short but deep recession and led to massive job loss in early 2020.¹ Despite a relatively rapid partial recovery, some labor market impacts persisted for several months.² Because a majority of working-age adults (i.e., persons aged 19 to 64) receive their health insurance through an employment setting, some have raised concerns about whether these large and lasting job losses led to a decline in employer-sponsored health insurance (ESI) coverage. Given the unique characteristics of the pandemic job loss, analyzing the extent to which this job loss contributed to ESI declines may help federal policymakers retrospectively evaluate certain COVID-19-related ESI policies and inform the policy debate should similar job loss situations subsequently occur.

This report examines the composition of employment change over the March 2019-March 2021 period—the extent to which net employment loss comprised jobs that included an offer of ESI coverage (ESI jobs) and those that did not (non-ESI jobs)—and how it relates to ESI offer rates among working adults and ESI coverage rates for both workers and the broader working-age population, which includes both workers and nonworkers.³

CRS analysis shows that employment levels fell from 85.0% of the working-age adult population in March 2019 to 82.9% in March 2021 (a decrease of 6.0 million jobs). During this period, the percentage of all working-age adults enrolled in ESI dropped from 62.2% to 61.6% (a decrease of 1.2 million adults with ESI coverage), which is a smaller decline in ESI coverage than studies published early in the pandemic predicted.⁴ Among employed working-age adults, however, ESI coverage rates increased from 73.9% to 74.9%.

This analysis suggests that the rise in the coverage rate among employed adults is in part due to attrition; that is, the ESI coverage rate increased among workers because a sufficiently large share of workers without ESI coverage lost work and therefore were no longer included in the calculation. Of the estimated 6.0 million jobs lost between March 2019 and March 2021, roughly 57% (3.4 million) were jobs that did not include an offer of ESI coverage.

Pandemic conditions led to swift and significant job loss in all industries but affected some sectors more than others. This report considers industry-level differences in the rates at which workers are offered ESI and the extent to which employment loss during the pandemic varied across industries. For example, leisure and hospitality employment fell markedly early in the pandemic and had not fully recovered by March 2021. Such a loss is notable in the context of ESI coverage because the leisure and hospitality sector, historically, has a relatively low ESI offer rate (51.5% in 2019, compared with 74% across all industries in that year). More generally, the four industries with the lowest ESI offer rates in 2019 (all of which were below the overall ESI rate) account for nearly half of the decrease in employment between March 2019 and March 2021.

Job loss during the pandemic was not exclusive to those without ESI offers; overall and within industries, employment declined to varying degrees among both jobs that offered ESI and those that did not. For example, in two of the three industries with the largest declines in employment levels between March 2019 and March 2021, ESI jobs made up 57.6% of the decrease (leisure and hospitality industry) and 80% of the decrease (education and health services), respectively.

The report also examines how ESI offer and coverage rates changed during the pandemic for workers and all working-age adults in selected demographic groups: race, Hispanic ethnicity, sex, and educational attainment. Across the demographic groups, coverage rates generally tended to rise or remain stable among workers and to decline or remain stable among the broader working-age adult population. This analysis reveals that working-age adults with lower levels of educational attainment were particularly vulnerable to loss of ESI coverage.

This report begins with a description of the data used before presenting national, industry-level, and demographic analyses.⁵ It concludes with four appendixes: The first contains a supplementary data table that describes ESI coverage rates by age. The second contains supplementary data tables that describe the composition of job loss (e.g., ESI jobs versus non-ESI jobs). The third provides additional tables that describe ESI coverage rates for several demographic characteristics. The fourth discusses methods used to assess the statistical significance of some estimates presented in this report.

Data Used in This Report

The U.S. Census Bureau's Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC) is the primary data source used in this analysis to examine health insurance coverage in March 2019 and March 2021. The CPS-ASEC is a large-scale household survey of the noninstitutionalized U.S. population (the 50 states and the District of Columbia) conducted annually by the Census Bureau, largely in March.⁶ The ASEC, a supplement to the Census Bureau's "basic" CPS,⁷ is designed to collect detailed information on income, poverty, and health insurance at the individual and household levels. Data from the monthly basic CPS (i.e., without the supplement) are used in this report to examine general labor force patterns; CPS-ASEC data are used to estimate employment levels when presented in data tables that include measures of ESI coverage or offer rates, such as in Table 1.⁸

Among other questions, the CPS-ASEC asks about household members' health insurance coverage at the time of the survey interview and during the preceding calendar year. It also asks whether individuals were receiving coverage through an ESI plan (through either their own employer or that of a spouse or parent) at the time of the interview and whether they had received such coverage in the preceding year. The analysis in this report is based on estimates of respondents' current coverage (i.e., their coverage at the time of the survey interview). The use of this point-in-time measure allows an analysis of changes in ESI between two specific time periods. By contrast, CPS-ASEC questions about ESI coverage during the preceding calendar year measure only if the individual had no coverage or had ESI coverage for at least part of the year.

When collecting information on current coverage, the CPS-ASEC also asks employed respondents whether their employers offer health insurance and, if so, whether the worker is eligible to accept the offer. Another benefit of using the current coverage measure is that it allows for the construction of the ESI offer measure (i.e., the share of workers who are offered and are eligible for ESI).

As a result, in this report, ESI coverage or ESI enrollment refers to individuals' enrollment in ESI (through either their own employer or that of a spouse or parent) at the time survey respondents were interviewed. It does not represent the total number of people who had ESI coverage for the entire year or the total number of people who had ESI at any point in the year. ESI enrollment includes those enrolled in Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA; P.L. 99-272) continuation coverage.

Analysis Period

This report considers the impact of the COVID-19 pandemic on ESI by comparing ESI offer and coverage rates in March 2019 with those rates in March 2021. Although the pandemic started in early 2020, ESI offer and coverage rates in March 2020 are not included in this analysis for two reasons. First, changes in employment status are expected to be a driving force behind COVID-19-related changes to ESI, and the large-scale job losses associated with the pandemic were concentrated in April 2020 (Figure 1). Due to the timing of the CPS-ASEC, which is administered mainly in March, the survey sample of observations measuring current coverage would include responses generally collected before large-scale job losses occurred.

Second, the pandemic disrupted data collection for the 2020 and 2021 CPS-ASEC, resulting in survey response rates that were lower than in months preceding the pandemic. (Similarly, the response rate for the basic CPS was 73% in March 2020 and 76% in March 2021; the pre-pandemic response rate for the basic monthly CPS was regularly above 80%). Census Bureau research reveals that not only were the ASEC response rates muted in 2020 and 2021, but the pattern of nonresponse was not uniform across the earnings distribution (lower-earning households were less likely to respond to the CPS) and other characteristics correlated with ESI coverage (e.g., employment status). Census Bureau analyses indicated that changes in the nonresponse pattern may have affected CPS-ASEC ESI estimates for 2020, meaning the magnitude of change in ESI coverage from the 2019 CPS-ASEC to the 2020 CPS-ASEC was overstated.⁹ The 2021 sample also was found to have nonresponse bias, but to a lesser degree than in 2020.¹⁰ The Census Bureau has indicated that comparisons between the 2021 CPS-ASEC and the 2019 CPS-ASEC provide for "the most consistent comparisons."¹¹

Population Examined in This Report

In large part, this report focuses on the working-age adult population—that is, persons aged 19 to 64 years. This group has relatively high labor force participation rates, and members of this group are more likely than members of other age groups to have ESI, either through their own employers or through a family member's employer (Table A-1).

The report excludes from its analysis individuals under the age of 19 and those over the age of 64. Individuals under the age of 19 with ESI are likely to be a dependent on the ESI policy (e.g., to have ESI provided through a parent's employer); as a result, in most instances, job loss among this younger population would have minimal impact on their ESI coverage rates. Those aged 65 and over generally are eligible for Medicare and may have the option to retire from the workforce, which can create differences in how this population interacts with the labor force and ESI, relative to the working-age adult population.¹²

COVID-19 Health Insurance Policy Landscape

In response to the COVID-19 pandemic and corresponding recession, Congress enacted policies that addressed comprehensive health insurance coverage access and affordability issues. These policies focused on ESI in the context of COBRA continuation coverage, subsidized individual health insurance coverage on the exchanges, and Medicaid.¹³ Collectively, these policies appeared to have had a minimal impact on the ESI estimates over the period analyzed in this report.¹⁴

The CPS-ASEC considers COBRA continuation coverage a type of ESI.¹⁵ As such, the American Rescue Plan Act of 2021 (ARPA; P.L. 117-2) COBRA premium assistance, which fully subsidized COBRA coverage for certain individuals from April 1, 2021, through September 30, 2021, could have affected CPS-ASEC ESI estimates. However, COBRA premium assistance was available only during a small portion of the 2021 survey period. Considering those surveyed for the 2021 CPS-ASEC were interviewed generally in March 2021—with only some interviews conducted in April 2021, when the premium assistance was first available—the ARPA COBRA premium assistance likely had a limited effect on the findings in this report.¹⁶

COVID-19-related policies regarding individual market coverage and Medicaid likely also had little effect on ESI estimates analyzed in this report. In general, individuals who are working and receive an adequate and affordable offer of ESI would not qualify for financial assistance (e.g., premium tax credits, or PTCs) in the individual market on the exchanges.¹⁷ As such, the ARPA expansion of eligibility for and amount of the PTC applicable to certain exchange plans for tax years 2021 and 2022 likely had minimal effect on the ESI estimates used in this analysis.

Similarly, COVID-19 Medicaid policies were not intended to directly incentivize individuals to forego ESI in order to receive Medicaid coverage, though some policies may have resulted in individuals retaining Medicaid coverage (and potentially delaying subsequent ESI enrollment, if available) for longer than they would have absent such policies.¹⁸ The extent to which Medicaid policies affected ESI estimates in this report appears minimal, as the CPS-ASEC found only a 0.4% increase in Medicaid enrollment from March 2019 to March 2021 (among all surveyed individuals).¹⁹

National Employment Levels and ESI Offer and Coverage Rates

The COVID-19 pandemic dramatically disrupted economic activity in early 2020 and led to widespread and massive employment loss. The number of employed working-age adults dropped sharply in April 2020 (Figure 1) in response to the pandemic and the associated recession. Employment among this group of workers rebounded partially between May and October 2020, as many workers on temporary layoff were recalled and other nonemployed workers found new jobs. The pace of employment growth slowed thereafter, and gradually returned to pre-pandemic in early 2022.

Between March 2019 and March 2021, employment levels among persons aged 19 to 64 fell from 143.2 million workers to 137.5 million workers.²⁰ Among non-self-employed workers in the same age group, employment levels fell from 130.2 million workers to 124.2 million workers over the same period.

Figure 1. Number of Employed Workers, January 2019-June 2022 (employment in millions)

Sources: Congressional Research Service (CRS) calculations using monthly data from the U.S. Census Bureau's Current Population Survey (CPS). Recession data are from the National Bureau of Economic Research. Notes: Rates among persons aged 19-64. Data are not seasonally adjusted.

Changes in ESI Versus Non-ESI Jobs

Over half of the jobs lost during the March 2019-March 2021 period were jobs that did not offer ESI to their employees. As shown in Figure 2, the number of workers reporting that their employers did not offer health insurance (or that the workers were not eligible for such offers) fell by 3.4 million, a 10.1% loss. This decline is greater than the decline in the number of workers reporting ESI offers (-2.6 million, a 2.7% loss) between March 2019 and March 2021.

Figure 2. Employed Workers by Employer-Sponsored Insurance (ESI) Offer, March 2019 and March 2021 (employment in millions)

Sources: CRS estimates using the Current Population Survey Annual Social and Economic Supplement (CPS-ASEC) and the CPS-ASEC Research Files (ESI offers in 2019) for 2019 and 2021. Notes: The sample comprises workers aged 19 to 64 who were employed at the time of the survey. Self-employed individuals are excluded. An offer of employer-sponsored insurance means the individual's employer offers health insurance coverage to its employees and the individual is eligible for such coverage.

Changes in ESI Offer Rates for Workers

Among the employed, access to ESI plans through their own employers (ESI offer rates) increased from 74.0% (March 2019) to 75.5% (March 2021) (Table 1).²¹ The rise in ESI offer rates among employed workers could have resulted from a relatively higher rate of job loss among workers who did not receive an offer of ESI through their jobs. Put another way, if ESI jobs make up a smaller share of total job loss than the initial ESI offer rate, the ESI offer rate will rise. For example, if overall employment declined but the number of ESI jobs did not change (i.e., ESI jobs made up 0% of total employment loss), then the ESI offer rate would increase (see text box and Figure 4).

The rise in ESI offer rates also may have been due, in part, to changes in employer behavior (e.g., a higher share of employers offering ESI as a benefit).²² The exit from the market of firms that are less likely to offer ESI (e.g., small employers, firms operating in low-offer-rate industries) could have contributed to a higher share of employers offering ESI.

Survey results published by the Kaiser Family Foundation show a similar share of firms overall offering health insurance in 2019 and 2021.²³ The Bureau of Labor Statistics Employee Benefits Survey found an overall increase in the percentage of private industry establishments offering healthcare benefits from March 2019 to March 2021, with industry-level variation.²⁴

Changes in ESI Coverage Rates

Among the employed, ESI enrollment increased from 73.9% to 74.9% over the March 2019-March 2021 period (Table 2). However, the percentage of all working-age adults enrolled in ESI dropped from 62.2% to 61.6% (approximately 1.2 million adults) over the same period (Table A-1).²⁵ Although 1.2 million adults losing ESI is not insignificant, it was less than some forecasts made in the early months of the pandemic,²⁶ and it was below some estimated losses for various periods in 2020.²⁷ The smaller decline found in this report may be attributed, in part, to the longer period of analysis; by including data for 2021, this report captures more of the partial rebound in employment than earlier studies, which could contribute to the smaller estimated decline in ESI coverage among working-age adults. That said, in comparison to some studies that use data sources other than the CPS-ASEC to examine health insurance enrollment from early 2019 to early 2021, studies that use the CPS-ASEC show a smaller loss in ESI coverage.²⁸

Employment Levels and ESI Offer and Coverage Rates, by Industry

Pandemic conditions led to swift, significant job loss in all industries but affected some sectors more than others. For example, leisure and hospitality employment fell by about 4.6 million workers (aged 19-64) between March and April 2020 (from 10.8 million to 6.2 million), representing more than a quarter of the 18 million workers who lost employment over that period.²⁹ Despite some recovery, there were still 2.1 million fewer workers employed in the leisure and hospitality sector in March 2021 than in March 2020.

These losses are notable in the context of ESI coverage in the current workforce because the leisure and hospitality sector historically has had a relatively low ESI offer rate (51.5% in 2019, compared with 74% across all industries in that year; see Table 1). More generally, the four industries with the lowest ESI offer rates in 2019 (all of which were below the overall ESI rate) account for about half of the decrease in employment between March 2019 and March 2021.³⁰

Changes in ESI Versus Non-ESI Jobs

Industry-level data suggest losses were not consistently concentrated on jobs that did not offer ESI. These data illustrate that employment losses across ESI and non-ESI jobs occurred at different rates across sectors (i.e., losses were not proportional to initial ESI offer rates).³¹ The three industries with the largest decreases in employment between March 2019 and March 2021 were the leisure and hospitality industry, manufacturing industry, and education and health services industry; collectively, these three industries accounted for roughly 87% of the decrease in employment over this period. In the leisure and hospitality industry, jobs without an ESI offer accounted for 42.4% of the decrease in employment from March 2019 to March 2021 (jobs with an ESI offer accounted for 57.6% of the decrease). In the manufacturing industry, jobs without an ESI offer made up 19.9% of the decrease in employment from March 2019 to March 2021 (jobs that offered ESI accounted for 80.1% of the decrease in employment). In the education and health services industry, jobs without an ESI offer accounted for 84.2% of the decrease in employment from March 2019 to March 2021 (jobs that offered ESI comprised only 15.8% of the decrease in employment). (Figure 3 and Table B-1.)

Figure 3. Change in Total Number of Employees With and Without ESI Offers, by Industry, March 2019 and March 2021 (change in thousands)

Source: CRS estimates using the CPS-ASEC and the CPS-ASEC Research Files (ESI offers in 2019) for 2019 and 2021. Notes: Rates among persons aged 19-64. Industries are ordered by the total change in employment. Employees refers to currently employed individuals but excludes self-employed individuals. An offer of employer-sponsored insurance means the individual's employer offers health insurance coverage to its employees and the individual is eligible for such coverage.

Changes in ESI Offer Rates for Workers

Assuming employers did not markedly change the extent to which they offered ESI during the analysis period, the extent to which decreases in employment were concentrated in jobs that did not include an offer of ESI could have impacted industry ESI offer rates among remaining employed individuals. As noted, the loss of large numbers of workers without an ESI offer would lead to a rise in ESI offer rates among the remaining workforce, all else being equal. The inverse is also true: the loss of large numbers of workers with an offer of ESI would lead to a decline in ESI offer rates among the remaining workforce, all else being equal.

Between March 2019 and March 2021, ESI offer rates increased for some industries (e.g., education and health services, wholesale and retail trade, financial services, public administration) and fell for others (e.g., leisure and hospitality, transportation and utilities, information) (Table 1). For the three industries that saw the largest declines in employment levels from March 2019 to March 2021, the change in ESI offer rates over the same period varied in terms of both the magnitude and the direction of change (Table 1). The leisure and hospital industry saw a decrease in its ESI offer rate (-1.8 percentage points), the manufacturing industry saw a slight increase in its ESI offer rate (0.6 percentage point), and the education and health services industry saw a relatively larger increase in its ESI offer rate (2.5 percentage points). Of the three, only the ESI offer rate change for education and health service jobs was statistically significant.

Employment losses in the education and health services sector were concentrated among workers without ESI offers (only 15.8% of the decrease in employment was attributed to jobs that offered ESI), and the industry-level offer rate increased between March 2019 and March 2021 (Figure 3 and Table B-1).

The extent to which job loss between March 2019 and March 2021 contributed to changes in ESI offer rates also depends on industry-level ESI offer rates in 2019 (i.e., the base level, see text box). For example, although about 80% of manufacturing industry employment losses comprised jobs that offered ESI, the ESI offer rate did not decline. In this instance, the offer rate did not fall because the manufacturing ESI offer rate was so high to begin with (85.0% in 2019). For the rate to fall, jobs with ESI offers would have needed to make up more than 85% of total manufacturing jobs lost (see illustrative examples in Figure 4).

In the leisure and hospitality industry, employment losses were slightly more concentrated on workers with an ESI offer—slightly more than half of losses (57.6% of the decrease in employment) were jobs with an ESI offer. Inversely from the manufacturing industry, the percentage of leisure and hospitality job loss that consisted of ESI-offering jobs was larger than the March 2019 ESI offer rate in the industry (51.5%), which contributed to the 1.8 percentage point-decrease in the ESI offer rate from March 2019 to March 2021.

Figure 4. Illustrative Examples of How ESI Offer Base Levels Affect ESI Offer Rate Changes

Source: CRS. Notes: Each square represents one employee. Each industry had 100 employees in 2019. Between 2019 and 2021, each industry saw 70% of the employment change consist of ESI jobs.

Changes in ESI Coverage Rates for Workers

Changes in ESI offer rates across industries do not necessarily translate to similar changes in the ESI coverage rates among workers in those industries. From March 2019 to March 2021, some industries (e.g., education and health services) saw increased offer rates and increased enrollment rates. In other industries (e.g., transportation and utilities), offer rates did not rise and enrollment rates declined during this period (Table 1 and Table 2). Still other industries (e.g., financial services) saw increased offer rates but unchanged enrollment rates. This disconnect may be partially attributable to the potential for an employee's ESI offer to impact the coverage status of employees in two separate industries (e.g., via family coverage offered by one employer to a married couple working in two distinct industries). It also may be attributable to changes in factors that may affect employee decisions to take up an ESI offer (e.g., the ESI becomes too expensive, as judged by the employee; a family member's employer coverage becomes available and/or more attractive due to premium, plan design, or other criteria; the employee prefers receiving higher take-home pay as opposed to using some wages for health insurance premiums). In addition, given the movement of workers between jobs between 2019 and 2021 (e.g., due to reemployment following a permanent job loss in 2020), some workers employed in March 2021 may not yet have met ESI eligibility criteria.

Employment and ESI Offer and Coverage Rates, by Demographic Groupings

Workers' ESI coverage rates vary across demographic groups. To some extent, that variation correlates with demographic differences in industry-level employment.³² For example, whereas 73.9% of all workers (aged 19-64) had ESI coverage in 2019, only 56.1% of employed Hispanic workers and 40.9% of workers without a high school diploma had coverage.³³ In terms of employment, Hispanic workers are disproportionately employed in leisure and hospitality sector jobs and construction jobs—industries with relatively low ESI offer rates (Table 1)—as are workers with lower educational attainment.³⁴ Employment losses also varied across worker groups: Hispanic female workers, non-Hispanic Black female workers, and workers with lower levels of educational attainment, in particular, lost work disproportionately.³⁵ These patterns suggest pandemic-related job loss could exacerbate existing health insurance disparities among demographic groups.³⁶ Against this backdrop, the following section examines how ESI offer and coverage rates changed between March 2019 and March 2021 for adults and for workers in selected demographic groups—namely, race and Hispanic ethnicity, sex, and educational attainment.

ESI Offer and Coverage Rates by Race and Hispanic Ethnicity

Employment declined between March 2019 and March 2021 for Hispanic workers (4.1% loss) and for workers in each racial group considered in this section: non-Hispanic White workers (4.9% loss), non-Hispanic Black workers (6.0% loss), and non-Hispanic Asian workers (2.7% loss).³⁷ Employment losses for Hispanic workers and non-Hispanic Black workers were driven largely by declines in female employment (4.9% loss and 6.6% loss, respectively).³⁸ Employment loss for each race and ethnicity group comprised both jobs that offer ESI (ESI jobs) and those that do not; however, ESI job loss was particularly pronounced for non-Hispanic Black workers.

Changes in ESI Offer Rates for Workers

In both March 2019 and March 2021, Hispanic workers had the lowest ESI offer rates and non-Hispanic White workers had the highest ESI offer rates. ESI offer rates increased for all worker groups (Hispanic and non-Hispanic racial groups) as employment fell. As discussed in the previous section, ESI offer rates rise if the ESI share of employment loss is less than the initial ESI offer rate,³⁹ which was the case for each group of workers. Consequently, ESI offer rates rose to varying degrees between March 2019 and March 2021, despite marked loss in ESI jobs for some groups. About 60% of the reduction in employment for non-Hispanic Black workers comprised jobs that offered ESI. Comparatively and over the same time period, 46.9% of the reduction in jobs among non-Hispanic White working-age adults were jobs that offered ESI and about 40% of the reduction in employment among Hispanic working-age adults comprised jobs that offered ESI (Table 3). ESI employment increased modestly for non-Hispanic Asian workers (resulting in ESI jobs having a negative share of job loss [-0.9%], Table 3).

The loss of ESI-offering jobs for the different racial and Hispanic ethnicity groups reflects, in part, changes in the industrial composition of employment brought about by the pandemic and the extent to which the workers' groups are concentrated in industries. In particular, the level of employment in the leisure and hospitality industry, a low ESI-offer industry, declined by 18% (non-Hispanic Asian workers) to 27% (non-Hispanic Black workers). At the same time, changes in public administration employment levels (a higher ESI-offer industry) varied by race and Hispanic ethnicity; employment in that industry increased by between 4% (non-Hispanic Asian workers) and 13% (Hispanic workers), mitigating the effect of jobs lost in low ESI-offer industries on overall offer rates to varying degrees.

Changes in ESI Coverage Rates

For employed workers in each group, the rise in ESI offer rates was met with a corresponding increase in ESI coverage (Table 4). For workers in aggregate and for some groups (i.e., Non-Hispanic White workers and Non-Hispanic Asian workers), the increase in coverage rates was smaller than the increase in offer rates. A smaller increase is not unexpected, as some workers may decline ESI offers for various reasons (e.g., they are receiving coverage through a spouse's plan, they consider the ESI to be too expensive, they prefer greater take-home pay).

Among the broader adult population (i.e., persons aged 19 to 64, regardless of employment status), ESI coverage declined for most groups (rates increased for non-Hispanic Asian adults). The coverage rate for non-Hispanic Black adults had the largest percentage-point decline. As discussed earlier, large losses in employment, particularly jobs that offered ESI, are expected to bring down ESI coverage among the broader adult population if formerly employed workers (and their dependents) lose their coverage and do not reenroll under another plan (e.g., through an ESI plan offered through their spouse's or parent's employer). ESI losses in a population also can occur for other reasons affecting the employed, such as fewer employers offering health insurance and fewer employees being eligible for, or electing, such coverage.

ESI Offer and Coverage Rates, by Sex

Female employment declined markedly relative to male employment early in the pandemic.⁴⁰ To a large degree, female employment recovered during the pandemic, such that losses over the March 2019-March 2021 period for females and males were both 4.6%.⁴¹ Despite a common employment loss rate over the analysis period, the composition of employment loss (with regard to jobs that offer ESI and those that do not) differed by sex.

Changes in ESI Offer Rates for Workers

For both females and males, ESI jobs lost as a share of total employment loss were below the ESI offer rate in March 2019; consequently, ESI offer rates rose for each group (Table 5). However, jobs with ESI offers (ESI jobs) made up a much larger share of employment loss for males than for females. Specifically, ESI jobs made up about 60% of losses for males and about 26% of losses for females. As a result, the increase in the ESI offer rate for female workers (2.3 percentage points) was larger than that for males (0.7 percentage points).

Changes in ESI Coverage Rates

As ESI offer rates increased among employed females and males from March 2019 to March 2021, so did ESI coverage rates (Table 6). Coverage increased for employed females by 1.3 percentage points and for employed males by 0.7 percentage points during that period. Among the broader adult population (i.e., considering employed and nonemployed adults aged 19-64), ESI coverage declined by less than a percentage point for both groups.

ESI Offer and Coverage Rates by Educational Attainment

The 4.6% decline in employment between March 2019 and March 2021 was driven by large losses among less-educated workers. Employment declined by about 16% among workers without a high school diploma and by about 8% for both workers with a high school diploma and workers with some postsecondary education but no bachelor's degree. Employment increased by nearly 4% for workers with an advanced degree, and the change in employment for those with a bachelor's degree (2.2%) was not significant at the 90% level.

Changes in ESI Offer Rates for Workers

Workers who had not completed high school had the lowest ESI offer rates in both March 2019 and March 2021, and workers with advanced degrees had the highest offer rates in both years. The industry composition of employment across educational attainment groups contributed to these patterns. For example, workers with lower levels of educational attainment (particularly females) are more concentrated in leisure and hospitality jobs and in wholesale and retail trade jobs.⁴² Workers with higher levels of education have relatively high shares of employment in education and health services jobs and professional and business services jobs, both of which had above-average ESI offer rates in March 2019 and March 2021.

Despite large losses in employment, ESI job loss was more or less proportional to ESI offer rates in March 2019 for workers with a high school degree or less; consequently, ESI offer rates for these workers did not change discernably (Table 7). For workers with some postsecondary education (but no bachelor's degree), the ESI job share of employment loss was below the ESI offer rate in March 2019. ESI jobs (and total employment) increased for workers with a bachelor's degree or higher, pushing up offer rates for these workers over the March 2019 to March 2021 period.

Changes in ESI Coverage Rates

Among employed workers, ESI coverage rates increased between March 2019 and March 2021 to a measurable extent for workers with a bachelor's degree (1.1 percentage point increase; see Table 8). Among the broader adult population (i.e., persons aged 19 to 64, regardless of employment status), ESI coverage declined for persons with only a high school diploma (2.6 percentage point decrease) and some postsecondary education (1.8 percentage point decrease). Coverage rates among persons with a bachelor's degree or higher were relatively stable during this period.

Summary of Findings

CRS estimates that over the March 2019-March 2021 period, employment levels fell from 85.0% of the working-age adult population to 82.9% (a decrease of 6.0 million jobs). During this period, the percentage of all working-age adults enrolled in ESI fell, whereas among employed working-age adults, however, ESI coverage rates increased. The analysis in this report suggests that the rise in the coverage rate among employed adults is in part due to attrition; that is, the ESI coverage rate increased among workers because a sufficiently large share of workers without ESI coverage lost work and therefore were no longer included in the calculation.⁴³

Industry patterns appear to contribute to the overall loss of ESI-offering jobs. That is, ESI offer rates and coverage rates vary across industries, and some sectors were affected by pandemic-related job loss more than others. The four industries with the lowest ESI offer rates in 2019 (all of which were below the overall ESI rate) account for nearly half of the decrease in employment between March 2019 and March 2021. However, job loss during the pandemic was not exclusive to those without ESI offers; overall and within industries, employment declined to varying degrees among both jobs that offered ESI and those that did not. For example, in two of the three industries with the largest declines in employment levels between March 2019 and March 2021, ESI jobs made up 57.6% of the decrease (leisure and hospitality industry) and 80% of the decrease (education and health services), respectively.

ESI offer and coverage rates changed to varying degrees for workers and working-age adults in selected demographic groups. These changes appear related to demographic differences in industry-level employment, and the extent to which each industry was affected by the pandemic. Notably, Hispanic workers, non-Hispanic Black workers, and workers without a high school diploma had lower ESI coverage than other groups in 2019. Between 2019 and 2021, however, statistically significant increases in ESI coverage for employed working-age adults were observed only for non-Hispanic White workers, non-Hispanic Asian workers, and workers with a bachelor's degree. These patterns raise concerns that pandemic-related job loss may have exacerbated existing health insurance disparities among demographic groups.

Appendix A. Additional Tables: Coverage Rates by Age Group

Appendix B. Additional Tables: Changes in Number of Employees With and Without Offer of ESI, by Selected Characteristics

Appendix C. Employer-Sponsored Insurance (ESI) Coverage Rates by Race and Hispanic Ethnicity and Sex and by Educational Attainment and Sex

Table C-1 disaggregates the change in ESI coverage rates for the broader population of adult males and females (employed and not-employed adults) by race and Hispanic ethnicity. For both males and females, ESI coverage rates were lowest for Hispanic adults and non-Hispanic Black adults in March 2019 and in March 2021. The largest drop in coverage occurred for non-Hispanic White males and non-Hispanic Black females. For both females and males, coverage rates increased among non-Hispanic Asian adults.⁴⁴

Table C-2 disaggregates the change in ESI coverage rates for the broader adult population (employed and not-employed adults) by educational attainment and sex. For both females and males, ESI coverage rates increased with levels of educational attainment in both March 2019 and March 2021. In March 2019, females with less than a bachelor's degree had lower ESI coverage rates than males in the same education groups and females with a bachelor's or advanced degree had higher coverage rates than males in the same education groups. However, between 2019 and 2021, female ESI coverage declined or remained mostly stable in each education category, whereas male coverage rates fell only for those with less than a bachelor's degree. Consequently, in March 2021, the ESI female coverage rates were below male rates for each education category except advanced degrees.⁴⁵

Appendix D. Statistical Significance of the Estimated Change in Health Insurance and Employment Indicators

The statistical significance of estimated changes in employer-sponsored health insurance offers and coverage between March 2019 and March 2021 is reported at the 10%, 5% and 1% levels.⁴⁶ Such statistical tests require standard error estimates. However, standard methods for incorporating the effect of survey design on variance (and thus standard error) estimates cannot be applied to Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC) data, because the Census Bureau does not publish strata and cluster information in the public use files. This missing information is problematic because stratification and clustering have been shown to affect variance estimates. In general, stratification results in efficiency gains (smaller standard errors) to the extent that it produces a sample with better coverage of subgroups than a simple random draw. Clustering, however, tends to increase standard errors due to intra-cluster correlation (i.e., outcomes and characteristics within a given geographic cluster will tend to be correlated).⁴⁷ In theory, when survey design involves both stratification and clustering, the net effect on standard errors is ambiguous. However, in practice, the efficiency gains from stratification tend to be quite small when compared with losses due to clustering, with the net effect of larger standard errors (relative to simple random sample). As a result, failure to account for the complex survey design used by the Census Bureau can result in downward biased standard errors.⁴⁸ To address these issues, the Congressional Research Service (CRS) calculated standard errors using replicate weights provided by the Census Bureau, which provide needed information about the survey's complex sample design, and the successive difference replication (SDR) method in the statistical software program Stata.⁴⁹

In some instances, significance levels are similarly reported for changes in employment levels between March 2019 and March 2021, based on CPS monthly data. The Census Bureau does not provide replicate weights for the basic monthly files. CRS attempted to minimize the (downward) estimated variance bias by creating proxy variables for the strata and clusters employed in the CPS design. To do so, CRS followed methods developed by the Center for Economic and Policy Research.⁵⁰