COVID-19 Expanded Unemployment Insurance Benefits May Have Discouraged a Faster Recovery

The public debate over the merits of the federal expansion to state unemployment insurance (UI) programs began even before the Coronavirus Aid, Relief, and Economic Security (CARES) Act was passed in March 2020. The controversy over whether expanded and extended UI benefits discourage some recipients from returning to work flares up each time the temporary programs expire and Congress debates whether and how they should be renewed. Following the widespread availability of effective vaccines against COVID-19 beginning in April 2021, the debate shifted to how and when to appropriately end expanded UI benefits. In May, 26 states unexpectedly announced that they would suspend participation in the federal expansion to UI benefits by late June or early July—more than two months before the federally scheduled end date. This unexpected change offers a unique opportunity to estimate the effect of UI changes on workers’ return to employment.


Among other things, the CARES Act created three programs: Federal Pandemic Unemployment Compensation (FPUC), Pandemic Unemployment Assistance (PUA), and Pandemic Emergency Unemployment Compensation (PEUC). FPUC increased the weekly benefits that workers receive (initially by $600 per week, then later by $300 per week). PUA expanded UI to independent workers, who had not previously paid into the UI program and therefore were not otherwise eligible for benefits. PEUC extended program eligibility for those workers who had exhausted their state UI benefits. Later legislation renewed and reduced FPUC’s expanded weekly benefits after the program lapsed and further extended PEUC’s unemployment benefits from 39 weeks to as many as 79 weeks. The programs’ effect on participants’ return to work was likely exacerbated by state-level changes to eligibility rules that no longer required program participants to actively search for reemployment.

UI from its conception has been designed to mitigate its theoretically anticipated effect of reducing participants’ return to work. Program parameters are typically constructed to motivate participants to resume employment as soon as possible. These include:

  • low value of benefits (generally replacing 40 percent to 50 percent of workers’ preemployment income, up to some level),
  • limited benefit duration (generally about 26 weeks—the length of time after which research has shown that unemployed workers who have not found a job tend to depart from the workforce altogether),
  • restricted program eligibility (UI is generally available only to previously employed workers with a recent history of consistent employment and who must actively search for new employment while enrolled), and
  • mandatory program exit (participants must accept a job offer that is appropriate for their skillset).

This approach has received substantial criticism during the COVID-19 pandemic. Many political commenters have insisted that fears of infection and the loss of school-provided childcare have been the primary determinants of workers not returning work, not the potential moral hazard caused by UI. That argument is partly correct, but those reasons for avoiding employment likely had the largest effect at the beginning of the pandemic, when it made sense to err on the side of caution before reliable information was available and effective safety protocols were instituted. Furthermore, extensions to UI can be economically beneficial by providing workers with the ability to wait for a good job match—one characterized by high productivity, satisfactory compensation, and a healthy working environment—as opposed to feeling pressure to take any job at all to replace their lost income.

However, it is also important to understand that UI critics generally focus on how the programs change participants’ incentives. UI, as it is currently designed, relaxes the immediate need to return to work. Periods of unemployment tend to erode human capital and job market connections, and previous research has found that, as the duration of unemployment increases, workers’ future income tends to decrease. Because workers’ quality of life is generally better when they have employment income than when they are making ends meet on limited social safety net benefits, encouraging reemployment (when safe and appropriate to do so) should be understood as being in workers’ best interest.

As time has gone on, the strength of pandemic-based arguments against returning to work has waned:

  • The widespread availability of effective vaccines for adults beginning in April 2021 (and teenagers in May) should have reduced many workers’ concerns regarding the consequences of workplace-caused infection.
  • Recent research by former Council of Economic Advisers Chair Jason Furman, Melissa Kearney, and Wilson Powell III finds that parents did not reduce their amount of work during the pandemic, suggesting that lack of childcare has not affected most parents’ employment status.
  • Vaccine availability for adults and teenagers enabled more childcare facilities to reopen during the summer (although the increased transmissibility of the Delta variant, especially among children, complicated the reopening of typical summer childcare institutions).

As fear of the virus diminishes and weary parents have more access to childcare, federally expanded UI plays a proportionally larger role in explaining why those who are jobless continue to remain so.

Concerns over the employment-discouraging effect of FPUC, PUA, and PEUC climbed even higher when the May release of the April jobs report showed unexpectedly weak employment growth. Just before this news, Montana Governor Greg Gianfonte had announced that the state would reinstitute work search requirements for UI eligibility, provide a return-to-work bonus, and suspend its participation in all three federally expanded UI programs. After the April jobs report, many other states quickly opted to follow Montana’s lead.

The abrupt policy change in 26 states offers an opportunity to estimate the extent to which these federal programs have discouraged workers from returning to employment during the economic recovery.

Economic Theory and Prior Empirical Research

The concern that the federal expansions to UI reduce the likelihood that workers will return to employment is based on the understanding that unconditional monetary grants to unemployed workers tend to raise their reservation wage—the compensation level necessary for the worker to take a job. UI programs are typically designed to mitigate this potential effect by replacing only a portion of workers’ preunemployment income (up to some income limit). However, the additional weekly benefits provided by FPUC (as well as the American Rescue Plan’s exemption of $10,200 of UI benefits from federal income tax) means that many low-wage workers saw no decrease in their weekly income (and some even saw an increase). This program redesign was arguably appropriate when COVID-19 was rapidly spreading and medical systems were in danger of collapse, but it seems less appropriate following the widespread availability of effective vaccines.

The effect of UI on labor markets has been studied extensively, and the empirical research has consistently confirmed the theoretical expectation, finding a negative effect of UI on employment. For example, a literature review by economists Johannes F. Schmieder and Till von Watcher of 13 studies that examine the effect of benefit increases on unemployment duration finds that all 13 studies connected increased UI benefits with longer unemployment durations. Research finds this effect to hold true even in relatively tight labor markets—such as currently exists—where wages are increasing because labor supply is low and employers are eager to hire. The exact size of UI’s effect on labor supply and the social benefits of UI are impossible to know (and they likely change according to context), but there is broad consensus that increased UI benefits discourage some workers’ return to employment.

Three recent studies confirm the findings of previous research. University of Wisconsin professor Noah Williams finds that the states that terminated federally expanded unemployment benefits before the federal deadline have had improved labor market outcomes when compared to those that continued their participation in the expanded UI programs. Employment in the hospitality industry was “especially strong in terminating states,” which is notable, given that this industry was hit hard by the economic downturn and, because of its lower wages, likely experienced the largest employment-discouraging effect from increased UI benefits. Research by Columbia University professor Kyle Coombs and coauthors finds that terminating states saw a 20 percent increase in the job-acceptance rate of UI participants relative to the states that had continued in the federal UI expansion. This expansion was driven by workers who had exhausted their UI benefits. Lastly, University of Chicago professor Peter Ganong and coauthors, using data that runs through April 2021, find clear evidence that federally expanded UI discouraged employment, even before vaccines were widely available and before states terminated federal UI benefits. The size of the effect is smaller than that found in prepandemic research, though.

Measuring the Effect of Federally Expanded Unemployment Insurance Benefits

Here we outline our own analysis of the labor market effects experienced by states that terminated their participation in federal UI programs and provide some preliminary results.


We use a difference-in-differences (DiD) framework to compare data from the period before states’ announcements would have affected such data (May 2021 or the percentage growth from March to May 2021) with data from the period following the full termination of state participation in federal programs (July 2021 or the percentage growth from May to July 2021). This simplified approach using panel data with two-way fixed effects (two time periods × 51 cross-sectional observations, with all treated units receiving the same treatment in the second period) ensures that our methodology does not run afoul of the current DiD credibility crisis facing quantitative research.

We regress the Current Population Survey (CPS) estimate of each state’s employed, unemployed, and not-in-the-labor-force individuals per capita and the Current Employment Statistics (CES) survey estimate of each state’s total payroll employment per capita and leisure and hospitality employment per capita against a set of explanatory variables that identify

  • states that ended all federal UI programs,
  • states that maintained all federal UI programs without other policy changes,
  • contemporaneous state COVID-19 case counts and death counts,
  • the proportion of each state’s population most susceptible to retirement (59 years and older) and most likely to require childcare (15 years and younger), and
  • the percentage of the state’s jobs lost during the pandemic.

We repeat this for the growth rates of the dependent variables from March to May and from May to July.


Our data is the same as those provided by the Bureau of Labor Statistics (BLS) for the monthly jobs report. These data consist of a household survey of labor market participation (conducted by the Census Bureau as part of the CPS) and an establishment survey of business employment records (conducted by the BLS as part of the CES). We draw our household survey data from the IPUMS CPS database and draw our data on payroll employment directly from the BLS’s CES state and metropolitan area database.

We construct an original dataset regarding state labor market policy changes. This is necessary because most previous research appears to have separated states into the 26 that terminated participation in some or all federal UI programs and the 25 (including the District of Columbia) that did not. There is substantial variation, however, within these two groups. Eleven states faced legal challenges to the termination of their participation in the federal programs, and legal injunctions were granted in four states, whereas three lawsuits were rejected and legal proceedings are ongoing in four other states. It seems likely that UI participants whose return to employment is conditional on the provision of UI benefits (or lack thereof) might have perceived the end of UI benefits in these 11 states as being less certain than did similar participants in the 15 states that did not experience legal challenges and ended participation in all federal programs. Furthermore, 4 states ended their participation only in FPUC, whereas 22 ended participation in all three federal expansions to UI. In addition, 10 states offered “return-to-work bonus” programs, which may also have affected the likelihood that jobless workers would return to the labor market. As a result, to clearly identify the effect of terminating participation in federal UI programs, any analysis should first focus on the 11 states that did so without facing legal challenges or instituting return-to-work bonuses contrasted against the 19 states that continued participation without any other policy changes.

Lastly, our data on state-specific COVID-19 case counts and death counts are drawn from the CDC’s database.


Our preliminary results agree with the findings of previous research: the parameter estimates show that higher UI benefits tend to discourage employment, whereas the end of UI eligibility appears to motivate more workers to become employed. However, the results are not one-sided—labor market outcomes clearly vary, even between states that pursued the same policy.

One particular insight that bears mentioning is that the CPS data and the CES data show diverging results. The CPS worker-focused data suggest that states that terminated federal UI program participation experienced employment growth that was twice as fast as states that maintained the programs. By contrast, the CES business-focused data suggest that the states that terminated their participation in federal UI programs experienced employment losses, whereas nonterminating states experienced employment gains.

Given that the CPS data directly address workers’ labor market status, we prefer to emphasize those results, but others might argue that the better data collection of the CES study tells the truer story. Regardless, the important takeaway is that the difference in the datasets might be driving the contradictory narratives currently being debated in the media.


Some pundits seem to have rushed to the defense of federally expanded UI programs when the July jobs report was released, arguing that there was no evidence that the programs discouraged employment. But this perspective cannot be reconciled with decades of labor market research. Furthermore, even research that has been framed as proof that the federal expansion to UI had no employment-discouraging effect itself acknowledges that workers were 20 percent more likely to accept jobs in states that had opted out.

It is entirely possible that jobless workers fear workplace-related COVID-19 infection or face childcare difficulties, and these are legitimate reasons for delaying their return to the job market and availing themselves of UI benefits. However, the question remains regarding why these difficulties would persist despite widespread availability of effective vaccines and the resurgence in travel and entertainment spending.

Our own analysis is a work in progress and is subject to refinement and revision as more data become available. Our preliminary results suggest that states that opted out of the federal expansion to UI saw faster employment growth than states that did not. To the extent that these results contradict other preliminary research, we suggest that differences between the CPS and CES datasets may be driving part of the disagreement. More work remains to be done, and better understanding of the controversy must be developed, because the pandemic-caused recession clearly showed that America’s existing UI system is woefully inadequate to the task of helping workers weather economic downturns.