Employment Effect of Eliminating COVID-19 Related Expanded Unemployment Benefits

By Philip Cornell


In mid-2021 several states stopped paying the additional unemployment benefits spurred by COVID-19. These benefits were provided by the federal government but paid by each state at its discretion.(1)(2) The additional benefits were created in order to ease the economic disadvantages of some workers resulting from the COVID-19 pandemic. Some states eliminated the benefits, as certain legislators from those states said that increased benefits were discouraging workers from seeking employment.

Certainly, the view that the extra COVID-19 payments discouraged work was not universal. Some policymakers believed that work for the unemployed simply did not exist, in some cases because of the pandemic itself. Others argued that that the extra COVID-related payments could actually encourage work by providing seed money to start new businesses more appropriate for a pandemic economy.(3)

My research suggests that the effect of eliminating COVID-related unemployment benefits may in fact increase workers seeking employment, but the effect of the reduction is dependent upon the state’s underlying COVID policies.


In July 2021, a selection of states eliminated the additional federally funded COVID benefits, while other states did not eliminate these same benefits. This contrast provides an opportunity to examine the effects of reducing unemployment benefits.  One group of highly respected economists, Harry J. Holzer, R. Glenn Hubbard, and Michael R. Strain, initially address this question in a National Bureau of Economic Research (NBER) paper.(4) Their findings suggest that ending expanded benefits reduced the unemployment rate in applicable states by approximately 0.3%.

The NBER paper defines employment using the traditional BLS U-3 rate. However, the often-used U-3 number fails to capture the quality of jobs; it considers employment of one hour every two weeks to be equivalent to a full-time, well-paid job and it considers that one is employed irrespective of whether the person earns well below a poverty wage.

An alternative measure, the “True Rate of Unemployment” (“TRU”) was developed by the Ludwig Institute for Shared Economic Prosperity (LISEP). Using data compiled by the federal government’s Bureau of Labor Statistics, TRU tracks those that do not have a full-time job (35+ hours a week) but want one. TRU also tracks those who have no job, and those that do not earn a living wage, conservatively pegged at $20,000 annually, in 2020 dollars before taxes. LISEP calculates this for both the labor force and the whole adult population. Further, LISEP has calculated these metrics by state, allowing us to compare state labor markets. LISEP’s view is that the TRU is a better measure of functional employment.

Using the TRU and the states’ diverging policies, this paper investigates if workers returned to meaningful, living-wage jobs as a result of being cut off from expanded COVID benefits.

My findings suggest that employment in living-wage jobs did increase in the states that eliminated the expanded unemployment benefits on average. In some ways, my findings are aligned with the NBER paper. However, looking deeper, among states with stricter COVID-19 policies (mandatory masks, prohibition on indoor dining, etc.), reducing unemployment benefits had little to no effect. The average effect of increased employment seems to have occurred only in those states with looser COVID protocols. A large majority of the states that ended expanded COVID unemployment benefits early also had loose protocols, thus suggesting that the average effect among all of the states that ended benefits early to be increased employment.

Several additional states reduced unemployment benefits in September 2021 when the funding for the benefits expired. On average, eliminating the extra benefits had no effect on employment levels for these states. Given my findings from the original states that reduced unemployment benefits combined with the fact that in this second group, nearly three-fourths of the states had more stringent COVID protocols, it is of no surprise that the extra benefits had no effect. It aligns with the suggestion that stringent COVID protocols will mitigate positive effects of reducing the expanded unemployment benefits.

The Holtzer, Hubbard, and Strain paper didn’t focus on the fact that 14 out of 17 of the states that reduced unemployment early (in July 2021) were states below the median for stringency of COVID-19 policies (as measured by the Oxford Covid-19 Government Response Tracker (OxCGRT)).

Method and Data

The first method used to analyze the effect of eliminating the federally funded expanded COVID benefits is a difference-in-difference cross-sectional regression (diff-in-diff). I compared the TRU Out of the Population (TRUOP) of the states that eliminated the expanded unemployment benefits in June 2021 with the TRUOP of the states that kept the expanded benefits. The TRUOP was used as opposed to the TRU, which is limited to the labor force, because the TRUOP measures the entire adult population. Someone is in the labor force if they are currently employed or looking for a job in the prior four weeks. The pandemic propelled many people to drop out of the labor force due to various factors, such as childcare issues and worries about workplace safety. Using the entire adult population instead of just the labor force gives a more comprehensive picture of whether overall employment increased.(5)

The diff-in-diff analysis compares the states that dropped the expanded unemployment benefits early (June 2021) with the states that kept the benefits until their official expiration (September 2021) to see if there was any impact on the TRUOP.(6) For all regressions, controls were used to eliminate the possibility of other factors (besides the reduction of COVID unemployment benefits) influencing the results. The first control was used to counter time trends in the data. Because unemployment hit such high levels in April of 2020, nearly every month since then has seen a decrease in unemployment as we slowly recover from the pandemic’s labor shock. To isolate the impact of the reduced COVID unemployment benefits, the bias of the downward trend was eliminated using time-fixed effects. This ensures that steady downward trend is not causally attributed to the impact of reducing the expanded unemployment benefits. Secondly, the health effects of COVID-19 were controlled for by including the number of cases in each state. This eliminated bias from health crises (plural because of the different waves and timing of the waves for each state)  that would have an outside impact on employment. Finally, the stringency of each state’s COVID-19 policies is also included. This helps for analyzing the effect (if any) of COVID policies on employment in the state. The OxCGRT(7) was used to examine confirmed COVID-19 cases and the states’ stringency index.

To confirm the initial results, I then investigated the effect of eliminating the federally provided expanded unemployment benefits in those states that dropped them in September 2021 (when the funding for the benefits expired). Using the same controls as above, I reduced the sample to these states and measured the TRUOP before and after the elimination of the COVID-related federal benefits.(8)


The results (shown in Table 1) suggest that there is some positive effect on functional employment by ending benefits. But there is a big caveat: the effect of ending unemployment benefits is more than counterbalanced by more stringent state-level policies aimed at COVID-19. Indeed, the reason for the extra grants is in part to compensate for the more stringent state rules that are meant to save lives. Eliminating the federal COVID-related benefits only increased functional unemployment (TRUOP) if the state’s COVID-19 protocols were in the bottom half of stringency as measured by the COVID stringency index. In fact, moving from the bottom half to the top half of states by COVID-19 stringency makes any positive effect on employment of benefit reduction completely negligible.  

Ultimately, effectively boosting employment by eliminating the COVID unemployment benefits was largely dependent on the COVID protocols of the state. This result was confirmed by testing the second round of states that eliminated the benefits in September 2021. In these states, the effect of elimination on average was insignificant (shown in Table 2). This was unsurprising considering about three-fourths of the states in this later sample had COVID-19 policies that placed them in the top half of states as measured by their COVID stringency.

March 21, 2022
  1. The Pandemic Unemployment Assistance and Federal Pandemic Unemployment Compensation (PAU) expanded eligibility for benefits, mainly to gig workers and those who classify as self-employed. The Federal Pandemic Unemployment Compensation program (FPUC) increased the value of the benefits.
  2. States ending both FPUC and PUA: Alabama, Arkansas, Georgia, Idaho, Iowa, Mississippi, Missouri, Montana, Nebraska, New Hampshire, North Dakota, Oklahoma, South Carolina, South Dakota, Texas, Utah, West Virginia, Wyoming. States ending neither FPUC nor PUA: California, Colorado, Connecticut, Delaware, District of Columbia, Hawaii, Illinois, Kansas Kentucky, Maine, Massachusetts, Michigan, Minnesota, Nevada, New Jersey, New Mexico, New York, North Carolina, Oregon, Pennsylvania, Rhode Island, Vermont, Virginia, Washington, Wisconsin.
  3. Hombert, J., Schoar, A., Sraer, D., & Thesmar, D. (2014). Can unemployment insurance spur entrepreneurial activity? (No. w20717). National Bureau of Economic Research.
  4. Holzer, H. J., Hubbard, R. G., & Strain, M. R. (2021). Did Pandemic Unemployment Benefits Reduce Employment? Evidence from Early State-Level Expirations in June 2021 (No. w29575). National Bureau of Economic Research.
  5. Using only the labor force could have given the false impression that employment increased when in fact, the unemployed were just no longer being counted because they were dropping out of the labor force all together.  
  6. These observations only include 2021 to exclude the effect of prior labor shocks.
  7. Thomas Hale, Noam Angrist, Rafael Goldszmidt, Beatriz Kira, Anna Petherick, Toby Phillips, Samuel Webster, Emily Cameron-Blake, Laura Hallas, Saptarshi Majumdar, and Helen Tatlow. (2021). “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker).” Nature Human Behaviour. https://doi.org/10.1038/s41562-021-01079-8
  8. I did not use the states that had already reduced their benefits as a comparison because those states wouldn’t satisfy the parallel trends assumption necessary for a diff-in-diff regression. This assumption is that before and after the policy change, the two groups would act in parallel, thus allowing the effect of a difference to be attributed causally. But the states that already struck unemployment could operate differently than a group of states without reducing benefits. Thus, I just regress on the difference, which implies that the states’ labor markets would have continued their trend if not for the policy.