How Did Pandemic UI Benefits Affect Employment Recovery in Local Industry Markets?

We analyze the employment recovery of low-wage establishments relative to the employment recovery of high-wage establishments within local labor markets, and we find a slower recovery in low-wage establishments. We associate the difference with the expanded generosity of pandemic unemployment insurance (UI) supplements, which have a larger negative effect on the job-filling rate of low-paying establishments. We use a model of labor search to translate our establishment-level observations into a disincentive effect of pandemic UI benefits at the worker level.

Between March and April 2020, the U.S. economy shed 22 million jobs, a consequence of stay-at-home orders and business restrictions to slow the spread of COVID-19. In response, the U.S. embarked on an unprecedented expansion of unemployment insurance (UI) programs. Beginning with the 2020 CARES Act, the federal government enlarged and extended eligibility criteria for the unemployed to receive UI and supplemental state UI benefits. Thus, for many recipients, UI payments considerably exceeded what they had earned in their previous jobs, as noted in the 2020 paper "U.S. Unemployment Insurance Replacement Rates During the Pandemic."

Naturally, these policy interventions have fueled a vigorous debate about the pros and cons of more generous UI benefits and, in particular, the extent to which they have contributed to labor supply shortages, thereby slowing down the recovery. Nonetheless, existing studies have found only small negative responses of employment, concluding that the disincentive effects of UI benefits are small.¹

How should we interpret the small effects estimated by these studies? Does this mean that the disincentive effects of the UI programs were negligible?

Disincentive and Stimulative Effects of Unemployment Insurance

UI benefits impose two opposing forces on the labor market:

• They act as a disincentive to supply labor.
• They act as an automatic stabilizer that, if sufficiently generous, may stimulate demand by raising disposable income of the unemployed, thereby helping the employment recovery, as noted in the 2021 article "Unemployment Insurance in Macroeconomic Stabilization."

In equilibrium, these two effects may largely offset each other, which would explain why existing studies have found only relatively small negative labor market responses of pandemic UI benefits.

One can argue that the overall effect may be more relevant for policy. Nonetheless, having separate estimates of the disincentive effect is important as well. To evaluate the macro effects and welfare implications of UI programs, labor economists often rely on quantitative models of labor search. In such models, unemployed people choose whether to accept job offers or to continue searching for work, hoping that better offers will arrive.

To discipline and evaluate these models, one needs a credible estimate of the disincentive effect of UI. In contrast, estimates of the overall effect may give the appearance that search models exaggerate the economic responses and, hence, need to be restructured. In addition, independent estimates of the disincentive effect versus the stimulative effect can help us to evaluate different policy alternatives, such as UI versus direct stimulus payments.

A Method to Isolate the Disincentive Effects of Unemployment Insurance

Our working paper "Disincentive Effects of Pandemic Unemployment Benefits" — also co-authored with Lien Ta from Drexel University — proposes a research design to estimate the disincentive effects of pandemic UI benefits. Our empirical methodology consists of two components:

• Comparing differences in the employment recovery of low-paying versus high-paying establishments within narrowly defined local-industry labor markets (for example, high-end restaurants in downtown Manhattan)
• Analyzing whether such employment recovery gaps are more pronounced in local industry markets with more generous UI programs

The main idea is that more generous UI benefits reduce workers' acceptance rates of low-wage job offers. Thus, establishments paying lower wages should experience a slower recovery than establishments paying higher wages for otherwise identical jobs. In local industry markets with relatively generous UI supplements, low-paying stores should have even more difficulty in attracting workers relative to high-paying stores in the same neighborhood. But increased demand arising from the generous UI supplements should affect all establishments in the local industry equally, thus not affecting the relative employment recovery of low-paying and high-paying stores.

This hypothesis is more likely to hold within narrow local industry markets. For example, establishments in a city neighborhood are affected by the average purchasing power of residents in the neighborhood. But establishments in two different cities face different demand conditions, which can influence the employment trajectory.

Another requirement for our methodology is to use establishments that cater to the local market (non-tradable goods and services). Establishments that cater to the national market make it harder to use local geography as a proxy for demand.

To that end, we use data from Homebase (HB), a scheduling and payroll administration provider used by more than 100,000 small businesses in the U.S. The data provide us with a unique worker-establishment matched panel of daily data on employment, hourly wages and hours worked as well as measures of newly posted vacancies and number of applications per vacancy. The HB data is highly representative of the type of low-wage workers in service-sector establishments that were most affected by the pandemic and that benefited most from the pandemic UI benefits.

Did Pandemic Benefits Have Strong Disincentive Effects on the Labor Market?

We start by documenting several facts about the labor market recovery from the pandemic:

• Establishments that paid relatively high wages prior to the pandemic regained employment faster than low-wage establishments.
• Despite the faster employment recovery, hours per worker in high-wage establishments increased by less than in low-wage establishments.
• Hourly wages paid by low-wage establishments grew at a faster pace than the ones paid by high-wage establishments, thus partially closing the pre-pandemic wage difference.

These findings suggest that low-wage establishments faced stronger labor supply constraints and reacted to these constraints by increasing hours worked of their existing employees and by raising wages at a faster pace.

Next, we assess how much of the difference in the employment recovery gap between high-wage and low-wage establishments within a local-industry market can be explained by pandemic UI benefits. We use two sources of variation. For the first source, we exploit the fact that the pandemic benefits were handed out in several rounds: an initial additional $600 in weekly federal unemployment insurance supplement on top of the usual state unemployment benefits, then multiple rounds of supplemental $300 weekly payments once the initial $600 per week program expired. As a second source of variation, we use differences in the replacement rates of low-wage versus high-wage labor markets. Local industries with large wage gaps between establishments also imply a large difference in the UI replacement rates.

Our combined evidence indicates that the pandemic UI supplements had sizable disincentive effects on local employment and slowed down the labor market recovery of low-paying stores relative to the labor market recovery of high-paying stores. Therefore, the effects of UI on employment turn out to be small, because they are confounded with the stimulative effects of UI on local labor markets. Once these local demand shifts are controlled for, the effects turn out to be substantial and negative.

Besides the disincentive effects of UI, public discussion of employment recovery during the pandemic also focused on other return-to-work hurdles such as COVID-19-related health risks and access to child care. Do these factors explain the recovery gap? We find that health risks and school closings have some predictive power for the employment recovery gap, but they do not affect our main conclusion that pandemic UI benefits had strong disincentive effects on the labor market.

A Model of Job Search

We next ask if a model of job search can replicate the disincentive effects we document in the data. It is important to ask this for two reasons:

• We want to know if our models work well with respect to replicating existing empirical patterns.
• We would like to estimate an aggregate employment effect that considers all stores.

The model is a job-search model. Unemployed workers randomly receive job offers which differ in their wages, and they accept the jobs if wages are higher than their reservation wages. Unemployed workers differ in their reservation wages, because their UI benefits depend on wages received when previously employed, as UI benefits increase in accordance with prior earnings. In the model, pandemic UI benefits decrease re-employment rates by increasing workers' reservation wages, leading to lower acceptance rates of job offers (the disincentive effect).

In normal times, unemployment is clearly undesirable. For around 26 weeks, unemployed workers receive just a fraction of their past wages through the UI system. But during the pandemic, UI was so generous that the average unemployed person received a weekly supplemental income even higher than his/her normal wage.

Therefore, unemployment may have not been immediately undesirable, but it was "risky." Unemployed workers declining relatively low-paying job offers to continue receiving higher pandemic UI payments might not be able to secure jobs once benefits were exhausted. We find that our model is consistent with the differential employment recovery we document in the data, provided unemployment is sufficiently risky.

Finally, the model suggests that the disincentive effects affected mostly low-wage establishments. The high-wage labor market recovered quickly even in the presence of the generous supplements. Thus, according to the model, the negative aggregate employment losses from unemployment insurance benefits are concentrated on the low-end of the wage distribution.

Conclusion

Pandemic UI supplements affected the labor market in two ways:

• On one hand, they made it difficult for stores — especially low-paying stores — to find workers.
• On the other hand, they supported local demand by raising the disposable income of the unemployed.

We isolated the disincentive effects by comparing the labor market recovery in low-paying and high-paying markets within narrow geographies and industries and found sizable disincentive effects.

Andreas Hornstein is a senior advisor and Marios Karabarbounis is an economist in the Research Department at the Federal Reserve Bank of Richmond. Andre Kurmann is a professor of economics at Drexel University's LeBow College of Business. Etienne Lale is an associate professor of economics at the University of Quebec at Montreal.

To cite this Economic Brief, please use the following format: Hornstein, Andreas; Karabarbounis, Marios; Kurmann, Andre; and Lale, Etienne. (November 2022) "How Did Pandemic UI Benefits Affect Employment Recovery in Local Industry Markets?" Federal Reserve Bank of Richmond Economic Brief, No. 22-44.

This article may be photocopied or reprinted in its entirety. Please credit the authors, source, and the Federal Reserve Bank of Richmond and include the italicized statement below.

Views expressed in this article are those of the authors and not necessarily those of the Federal Reserve Bank of Richmond or the Federal Reserve System.