Enhanced Unemployment Insurance Benefits During COVID-19
Summary
Federal Reserve Bank of San Francisco Working Paper 2024-15, "Enhanced Unemployment Insurance Benefits in the United States During COVID-19: Equity and Efficiency," by Robert G. Valletta and Mary Yilma, dated December 2024. The paper assesses the expansion of U.S. unemployment insurance (UI) payments during the pandemic, including the $600 weekly increase available in 2020 and the $300 supplement for most of 2021. Using CPS ASEC household income data adjusted for underreporting of UI benefits, the authors report that UI payments reversed the increase in household income inequality that otherwise would have occurred in 2020 and 2021. They also report that the search disincentive effects of the $600 increase were limited overall and smaller for individuals from lower-income households. The 26-page paper closes with a reference list.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Full text
FEDERAL RESERVE BANK OF SAN FRANCISCO
WORKING PAPER SERIES
Enhanced Unemployment Insurance Benefits in the United States
During COVID-19: Equity and Efficiency
Robert G. Valletta
Federal Reserve Bank of San Francisco
Mary Yilma
Massachusetts Institute of Technology
December 2024
Working Paper 2024-15
https://doi.org/10.24148/wp2024-15
Suggested citation:
Valletta, Robert G., and Mary Yilma. 2024. “Enhanced Unemployment Insurance
Benefits in the United States During COVID-19: Equity and Efficiency.” Federal
Reserve Bank of San Francisco Working Paper 2024-15.
https://doi.org/10.24148/wp2024-15
The views in this paper are solely the responsibility of the authors and should not be interpreted
as reflecting the views of the Federal Reserve Bank of San Francisco or the Board of Governors
of the Federal Reserve System.
Enhanced Unemployment Insurance Benefits in the United States
During COVID-19: Equity and Efficiency
Robert G. Valletta*
Federal Reserve Bank of San Francisco
rob.valletta@sf.frb.org
Mary Yilma
Massachusetts Institute of Technology
myilma@mit.edu
First version: November 20, 2023
Revised: December 2024
*Corresponding author. The work was largely completed while coauthor Mary Yilma was a
Research Associate at the Federal Reserve Bank of San Francisco. Thanks go to Jeff Larrimore
and coauthors for their posted data on adjusted UI recipiency, and to Kenneth Couch for helpful
editorial guidance. This paper was written for the forthcoming volume “Handbook on Inequality
and COVID‐19,” edited by Kenneth Couch and published by Edward Elgar. The views
expressed in this paper are solely those of the authors and are not attributable to the Federal
Reserve Bank of San Francisco or the Federal Reserve System.
Enhanced Unemployment Insurance Benefits in the United States
During COVID-19: Equity and Efficiency
Abstract
We assess the effects of the historically unprecedented expansion of U.S. unemployment
insurance (UI) payments during the COVID-19 pandemic. The adverse economic impacts of the
pandemic, notably the pattern of job losses and earnings reductions, were disproportionately
borne by lower-income individuals. Focusing on household income as a broad measure of well-
being, we document that UI payments reversed the increase in household income inequality that
otherwise would have occurred in 2020 and 2021. We also examine the impacts of the $600
increase in weekly UI benefit payments, available during part of 2020, on job search outcomes.
We find that despite the very high replacement rate of lost earnings for low-wage individuals, the
search disincentive effects of the enhanced UI payments were limited overall and smaller for
individuals from lower-income households. These results suggest that the pandemic UI
expansions improved equity but had limited consequences for economic efficiency.
Keywords: unemployment insurance, COVID-19, inequality, income support, job search
JEL codes: D31, J64, J65
Valletta-Yilma, Unemployment Insurance
Enhanced Unemployment Insurance Benefits in the United States
During COVID-19: Equity and Efficiency
I. Introduction
The COVID-19 pandemic that began in early 2020 prompted widespread curtailment of
social and economic activity in the United States aimed at controlling the spread of the virus.
The resulting labor market shock was unprecedented in its speed and severity. During the first
few months of the pandemic, job losses and unemployment surged to levels not seen since the
Great Depression of the 1930s, and labor force participation sagged. The policy response was
swift, with the CARES Act passed in late March 2020 providing substantial direct aid for
households and individuals affected by the pandemic and ensuing public health responses.
The enhancement of unemployment insurance (UI) benefits was a key element of this
expanded public support intended to provide an economic bridge past the pandemic. The UI
enhancement was historically unprecedented. Most notably, it included a temporary increase in
weekly benefit payments of $600, available from late March through late July 2020. This benefit
increase dwarfed any prior increases in UI payments during economic downturns; it more than
doubled typical weekly UI payments across the United States (Ganong et al., 2020; Petrosky-
Nadeau and Valletta, 2023). After expiration of the $600 benefit followed by partial renewal
later in 2020, the benefits enhancement was reinstated at $300, half its prior level, for most of
2021. The increases in weekly benefits were accompanied by substantial expansions of
maximum benefit durations and eligibility.
In this paper, we contribute to two existing strands of literature on the impacts of the
enhanced pandemic UI benefits. One strand examines the equity or distributional impacts of the
UI payments, generally focusing on the degree to which they replaced lost earnings and how this
Page 3 of 26
Valletta-Yilma, Unemployment Insurance
varies by pre-job loss earnings levels (e.g., Ganong et al., 2020; Cortes and Forsythe, 2023a;
Larrimore et al., 2022). We extend this line of research by focusing on household income rather
than individual earnings. We use microdata from the Current Population Survey Annual Social
and Economic Supplement (CPS ASEC). Due to substantial underreporting of UI payments in
the CPS ASEC data during the pandemic, we use imputed UI amounts derived from tax records
(from Larrimore et al., 2023). We find that UI payments reversed the increase in household
income inequality that otherwise would have occurred in 2020 and 2021.
We also examine the impacts of the $600 increase in weekly UI benefit payments on job
search outcomes. The very large increase in available UI benefits during the pandemic raised
concern about reduced incentives to return to work that could delay the labor market recovery.
Such search disincentive or economic efficiency effects have been documented in a broad
economic literature (see the overview in Petrosky-Nadeau and Valletta, 2023). They represent a
potential adverse consequence of the intended insurance and equity effects of UI benefit
payments. Our empirical framework to assess these economic efficiency effects is from
Petrosky-Nadeau and Valletta (2023). They use a “difference-in-difference” regression design
that exploits the wide variation in the extent to which the UI payments replaced prior earnings,
i.e. the “replacement rate.” We find that despite the very high replacement rate of lost earnings
for low-wage individuals, the search disincentive effects of the enhanced UI payments were
limited overall and smaller for individuals from lower-income households. Overall, our results
suggest that the pandemic UI expansions improved equity but had limited consequences for
economic efficiency.
Page 4 of 26
Valletta-Yilma, Unemployment Insurance
II. Pandemic UI as Income Support
A. The Pandemic Labor Market Shock and UI Policy Responses
The initial effects of the pandemic on the U.S. labor market were rapid, devastating, and
unprecedented. The immediate shutdowns for a wide range of industries pushed millions of
individuals out of their jobs, with limited options for alternative employment. Between February
and April 2020 the official U.S. unemployment rate rose over 10 percentage points, from 3.5
percent to 14.7 percent, the highest level recorded since the Great Depression of the 1930s. This
was accompanied by a sharp drop in labor force participation and widespread misreporting of
employment status, with many survey respondents indicating that they were absent from work
rather than on layoff. As such, the labor market disruption was even larger than indicated by the
increase in the official unemployment rate.1 Viewed more broadly, the number of individuals
employed fell by slightly over 25 million between February and April 2020, representing an
employment decline of about 16 percent and suggesting an unemployment rate near 20 percent.
These adverse effects on the labor market were disproportionately borne by lower-
income and otherwise disadvantaged groups. For example, Cortes and Forsythe (2023b) found
that he pandemic exacerbated pre-existing inequalities, with substantially larger employment
losses in lower-paying occupations and industries. A number of papers also document that
disproportionate impact of the pandemic on the economic status of racial/ethnic minorities and
women (Bitler et al., 2023; Couch et al., 2020, 2022).
The response to the pandemic’s severe and uneven economic shock, via expansion of
transfers and social safety net spending, was equally rapid and unprecedented (Bitler et al., 2020,
1
Bartik et al. (2020) and Cajner et al. (2020) detail the immediate impact of the pandemic on the U.S.
labor market. See U.S. BLS (2020) for discussion of the mismeasurement of labor market status and its
implications for pandemic effects on the U.S. labor market.
Page 5 of 26
Valletta-Yilma, Unemployment Insurance
2023; National Academy of Social Insurance, 2023). The Coronavirus Aid, Relief, and
Economic Security (CARES) Act, passed quickly and made effective in late March 2020,
provided a wide range of near-term income support and stimulus elements aimed at households
and businesses. It included direct aid for individuals and households, such as tax rebates, along
with loan forgiveness and support provisions for businesses. At $2.2 trillion, it was the largest
economic stimulus package ever passed in the United States.
The element of the CARES Act aimed most directly at helping the massive number of job
losers was the enhancement and expansion of UI benefits. The act expanded the eligibility for UI
benefit receipt, most notably to self-employed and gig workers, along with increasing the
maximum duration and weekly amounts payable for UI benefits.
The increase in UI benefit payments was particularly large. At $600 per week, available
from late March through July 2020, it represented an historically unprecedented increase in UI
benefit amounts. With median UI benefit payments at about $400/week, the extra $600/week
more than a doubled typical payment amounts. It was designed to largely replace lost earnings
for individuals displaced by the pandemic shutdowns. The uniform amount paid for all earners,
however, meant much larger replacement rates for lower earners. As a result, most recipients of
UI benefits during this timeframe received weekly UI payments that exceeded their earnings
prior to job loss (Ganong et al., 2020; Petrosky-Nadeau and Valletta, 2023).
The emergency UI policies specified in the CARES Act evolved as the pandemic
progressed (see Whittaker and Isaacs, 2022, for details). As the pandemic’s labor market
disturbances continued, an initial extension of 13 weeks via the CARES Act eventually grew to
49 weeks during much of 2021 (through September). This meant 75 weeks of benefit availability
when combined with 26 weeks of normal UI benefits, a long duration in historical terms but
Page 6 of 26
Valletta-Yilma, Unemployment Insurance
shorter than the 99 weeks available in the aftermath of the Great Recession of 2007-09 (Valletta,
2014). Due to uncertainty and disagreement about the extent of income support needed during
the pandemic, the $600 supplement was not immediately renewed upon expiration at the end of
July 2020. Following a brief renewal at a lower level in August 2020, the supplement was
renewed in early 2021 but at $300/week, half the prior level. This federal policy was in place
until September 2021. However, due to rapidly improving labor market conditions in some states
in early to mid-2021, combined with political disagreement about the tradeoffs inherent to UI
benefit payments, many states opted out of the program early, beginning in June 2021.
B. Data
We now turn to an assessment of the extent to which the UI policies described above
offset the adverse, unequal impacts of the pandemic labor market shock on household income.
For our analyses, we use microdata on household income, employment status, and related
characteristics from the CPS ASEC, which is the source of official government statistics on
income distribution and poverty. The ASEC is administered to monthly CPS respondents
primarily in March, with a smaller share of respondents asked to complete it in other months.2
We use data from the 2020-22 ASEC surveys, which contain detailed information on income
sources and employment status in the preceding calendar years of 2019-2021. We adjust all
income amounts for inflation using annual values of the PCE price index, relative to 2019 as the
base.
2
The monthly CPS survey is the source of official labor force statistics in the United States. The ASEC is
one of a number of supplemental surveys conducted as part of the CPS survey program.
Page 7 of 26
Valletta-Yilma, Unemployment Insurance
The CPS ASEC separately identifies the amount of regular and supplemental UI benefits
received and hence enables us to directly examine their impact on the distribution of household
income. One challenge is that a sharp rise in CPS survey non-response during the pandemic
intensified measurement challenges for income and employment statistics. In particular, recent
work by Larrimore et al. (2023) shows considerable understatements of UI recipiency status and
amounts received within the CPS-ASEC relative to administrative data from the IRS, particularly
for years within our reference period of 2019-2021. For calendar year 2020, CPS ASEC
estimates of total UI recipients and benefits received were respectively about 50% and less than
40% of the amounts reported in administrative data. Misreporting was particularly concentrated
toward the lower end of the income distribution. This has important distributional implications.
For example, Larrimore et al. report that correcting for underreporting of UI benefits reduces
measured poverty by nearly 2 percentage points in 2020, to a record low of 9.6% (versus 10.5%
in 2019)
To adjust for this severe understatement of UI benefits in CPS ASEC and other data
sources, Larrimore et al. (2023) developed and provide adjusted UI recipiency amounts based on
detailed tax records.3 We use their data and imputation procedure, which entails adjusting UI
benefits reported in our CPS ASEC dataset by randomly imputing recipients and amounts
received within income centiles to match the distribution of UI benefits reported in their tax
data.4 Our construction of the income measure used as the basis for the imputation procedure
3
Available at https://www.davidsplinter.com/.
4
The imputation procedure is reliable for assessing distributional impacts that reflect variation across but
not within the 100 income cells used for the calculations. Our analyses below focus on standard income
percentiles (median, 10th and 90th percentile, etc.). This imputation procedure is not appropriate for
analyses conducted at the individual level in microdata. For our analyses of individual job search behavior
in the next section, we use a different procedure for imputing UI benefits in 2020 (see Section III).
Page 8 of 26
Valletta-Yilma, Unemployment Insurance
follows Larrimore et al. (2023) in aggregating different sources at the tax-unit level and splitting
equally between spouses.5
C. Distributional impact of UI benefits in 2020 and 2021
To assess the distributional impact of the UI benefit payments during the pandemic, we
use the simple approach of comparing the distribution of real (inflation-adjusted) household
income with and without UI benefits (adjusted for underreporting) included in the calculation.
We start with a visual representation based on kernel density estimation of the income
distribution. Such density plots are essentially smoothed histogram representations of an
underlying distribution, where the kernel function and bandwidth used for estimating it
determine the exact shape.6
Figure 1 shows the impact of UI benefits on the distribution of real household income in
2020. For visual clarity we only display household income ranges of 0 to $100,000; outside this
range, UI benefit payments have minimal impact. Comparing the black and grey solid lines
shows that the distribution of income shifted out somewhat between 2019 and 2020, indicating
increases in overall income after correcting for underreporting of UI benefits. In a stark contrast,
the dashed line shows what the distribution of household income would look like if no UI
payments had been made in 2020. This counterfactual distribution is shifted sharply to the left
5
Rothbaum and Bee (2021) analyze non-response bias in the CPS ASEC data collected during the
pandemic. They find that lower-income households were less likely to respond, thereby overstating
aggregate household income using the survey weights. They develop alternative individual weights,
which were used by Larrimore et al. in forming their UI imputation data. Household weights adjusted in a
similar manner are not available, so we use the conventional CPS ASEC supplement weights.
6
Daly and Valletta (2006) discuss kernel density methodology in more detail. We use an Epanechnikov
kernel function and 100 bins that are set as identical for the distributions displayed.
Page 9 of 26
Valletta-Yilma, Unemployment Insurance
and has much more pronounced mass toward the bottom tail, for household incomes below about
$30,000.7
Figure 1:
Figure 2 is identical to Figure 1 but provides the income comparison for 2021 rather than
2020. As in Figure 1, the counterfactual distribution with UI payments eliminated in Figure 2
shows a leftward shift and greater mass below the $30,000 level. However, the shift is less
pronounced for 2021 in Figure 2 than for 2020 in Figure 1. This smaller distributional effect of
UI payments in 2021 reflects: (i) the reduction in the weekly UI supplement, from $600 to $300;
(ii) improved labor market conditions that reduced UI recipiency.
7
It would be interesting and informative to distinguish between payments via regular UI and the
expanded pandemic UI programs. Unfortunately, the CPS ASEC data and the tax records used by
Larrimore et al. (2023) do not distinguish between different components of UI income.
Page 10 of 26
Valletta-Yilma, Unemployment Insurance
Figure 2:
Table 1 provides a quantitative summary of the main distributional impacts of UI benefit
payments for the years 2020 and 2021. It lists the median and selected other percentile ratios
from the distribution of household income underlying the density plots depicted in Figures 1 and
2. The selected percentiles used for the ratios, the 10th, 25th, 50th, 75th, and 90th, are commonly
used to summarize differences in income distributions (see for example Daly and Valletta, 2006).
Page 11 of 26
Valletta-Yilma, Unemployment Insurance
Table 1
Household income percentiles (median and ratios), with/without UI
(1) (2) (3) (4) (5)
2019 2020 2020 2021 2021
(ex. UI) (ex. UI)
Median (p50) 69030 70057 64980 69179 66608
p90/p10 12.4 11.4 13.9 12.2 14.1
p50/p10 4.3 4.0 4.6 4.1 4.7
p90/p50 2.9 2.9 3.0 2.9 3.0
p75/p25 3.6 3.4 3.8 3.6 3.8
Source: Authors’ calculations from CPS ASEC data with Larrimore et al.
(2023) UI imputations (years listed are the income reference year, which
is the year prior to the corresponding survey year).
Note: Income for all years adjusted to 2019 terms using the PCE
deflator.
The comparison of the 2019 and 2020 observed distributions in columns 1 and 2 of Table
1 shows an increase of about 1.5 percent in median real household income and a reduction in
inequality (dispersion) of income. The third column shows that the distribution would have been
much different had UI payments not been available in 2020: the median value would have been
about 7 percent lower, and inequality would have been much higher. The latter effect is almost
entirely due to differences in the lower half of the distribution, with a notably larger ratio of
median to 10th percentile earnings when UI benefits are excluded.
The comparison of the final two columns to the prior columns shows a similar but less
pronounced pattern of UI benefit effects in 2021 compared with 2020. Based on reported income
including imputed UI benefits (columns 2 and 4), median real income fell slightly between 2020
and 2021, and bottom-half inequality rose slightly. Absent UI benefits, median real income
would have declined materially from 2020, and dispersion in the lower half of the distribution
would have increased.
Page 12 of 26
Valletta-Yilma, Unemployment Insurance
Overall, these results show that expanded UI fully neutralized the impact of the pandemic
on the distribution of HH income in the years 2020 and 2021: absent UI benefit payments,
inequality in the bottom half of the distribution would have risen substantially. This is consistent
with the large poverty-reducing effects of pandemic UI payments reported by Larrimore et al.
(2023).
III. Job Search Effects
A. Background
By offsetting lost earnings and reducing the near-term costs of job loss, UI benefit
payments make unemployment a less deleterious state for individuals and household members
with whom they share income. A long literature in economics discusses the resulting “moral
hazard” effect of UI benefit payments on job search effort and job acceptance rates (see
Petrosky-Nadeau and Valletta, 2023, for discussion). The $600 UI benefit supplement in 2020
dwarfs any previous benefit enhancement, and it more than doubled typical weekly UI benefit
payments (which averaged slightly under $400 across U.S. states prior to the pandemic). Ganong
et al. (2020) note that absent supplements, the median UI replacement rate across all eligible
workers in the United States hovers slightly below 0.5 (around 50% of prior earnings). They find
that the $600 CARES Act UI supplement raised this number substantially, to a median value of
1.34. This implies that a majority of UI recipients were eligible for benefits exceeding their prior
job earnings. Such large increases in UI benefit levels could substantially offset job search
incentives. We investigate such effects in this section.
Page 13 of 26
Valletta-Yilma, Unemployment Insurance
B. Data
Our analysis of pandemic labor force transitions relies on matched monthly CPS data,
replicating and extending Petrosky-Nadeau and Valletta (2023) (PNV 2023). We link the
monthly files to annual earnings sourced from the ASEC, which limits our matched observations
to the months of January through July 2020. This timeframe is narrow but enables us to focus on
the period when the $600 supplement was in place (April-July), along with the preceding two
months of 2020 (February-March), which are used as a pre-treatment comparison period.
Due to the rotating sampling scheme used for the CPS, surveyed households and
individuals are in the sample for two separate periods of 4 consecutive months (with an
intervening 8-month period spent out of the sample). This enables consecutive month-to-month
matching for around 70% of the sample, which we further validate by ensuring that reported
demographics do not conflict for matched individuals.8 We then identify labor market transitions
by comparing an individual’s labor force status in consecutive months. In this paper, we focus on
transitions out of unemployment (U) to employment (E), a subset of the transitions examined by
PNV (2023).
A well-known concern regarding matched CPS data is the likelihood of spurious
transitions in labor force status arising from inconsistent or error-ridden survey responses rather
than meaningful changes. Therefore, in addition to examining monthly transitions, we also
follow past research by adjusting the data to minimize the incidence of spurious transitions. For
this adjustment, we recode individuals identified as having left unemployment one month and
8
Most of the non-matched observations are from the “outgoing rotation groups” that are exiting the
sample for eight months or permanently (one quarter of each monthly sample). In addition, a modest
fraction of observations is lost because respondent households that move to different geographic locations
are not followed.
Page 14 of 26
Valletta-Yilma, Unemployment Insurance
returned the next month as having had no transition (Rothstein, 2011; Valletta, 2014; Farber and
Valletta, 2015; Farber et al., 2015). If left unaddressed, these errors could bias the estimated
effect of UI payments on labor force transitions downward and decrease the precision of our
estimates. Applying this modification to the data reduces monthly transitions out of
unemployment by about five percentage points (Valletta, 2014). The adjustment results in a
sample we term “two-month matches,” which comprises matched individuals whom we observe
for at least three consecutive months. This reduces the observation count by about one-half
relative to the single-match sample and eliminates July 2020 observations (our final month of
data, for which we cannot check whether a further transition occurred).
As in PNV (2023), our analysis relies on UI replacement rates—the ratio of weekly UI
payments to weekly earnings prior to the job loss resulting in the claim—calculated at the
individual level. We rely on the calculator from Ganong et al. (2020) to estimate these
replacement rates. The calculator requires individual employment and earnings data from prior
quarters, which we obtain from the 2020 ASEC. Because the ASEC does not contain information
on the timing of income and employment, we spread annual earnings evenly across all four
quarters of 2020 before applying the UI benefits calculator.
One notable feature of the distribution of UI replacement rates is that because normal UI
payments generally are determined as a fraction of prior earnings, the uniform $600 supplement
increased replacement rates more for individuals with low versus high prior earnings. To
quantify this, we divided our sample of unemployed individuals into quintiles based on weekly
earnings and calculated their UI replacement rates. The median replacement rate for each group
ranged from about 2.5 in the lowest quintile down to about 0.6 in the highest quintile.
Page 15 of 26
Valletta-Yilma, Unemployment Insurance
This systematic relationship between prior earnings and UI replacement rates raises a
concern for our empirical framework. Because early-pandemic job losses were chiefly within
low-wage sectors, our estimates based on individual variation in replacement rates may be
distorted by sector-specific labor market conditions. PNV (2023) address these concerns in detail
and validate their general empirical approach.
As an illustration, Figure 3 (reproduced from PNV) shows the distribution of replacement
rates across broad occupation categories. The median and mean replacement rates vary notably
across occupations, with higher replacement rates evident in low-wage services occupations.
However, replacement rates vary more within than between occupations, as reflected in standard
deviation spreads within occupations that generally extend to or beyond the range of means and
medians across occupations. This suggests that variation in replacement rates is not closely
related to sector-specific effects of the pandemic.
Page 16 of 26
Valletta-Yilma, Unemployment Insurance
Figure 3:
C. Empirical framework
Our empirical analyses rely on a conventional before/after regression design, essentially a
differences-in-differences analysis that yields an estimate of the average treatment effect of
variation in UI replacement rates. We include imputed UI replacement rates for individuals in
our sample using the procedure described in the preceding section. Our specific regressions take
the following form, using a logit model for estimation:
Pr(Yit = 1) = δRi + (π × Ri × (Apr-Jul)) + γt + βXi,t-1 + λZs,t-1 (1)
Page 17 of 26
Valletta-Yilma, Unemployment Insurance
In this equation, the dependent variable Yit is an indicator for whether an individual i
transitions between unemployment and employment across consecutive months—i.e., whether an
unemployed individual finds a job between months t-1 and t. We report results from our one-
month and corrected two-month matches (described in the preceding section). The underlying
sample contains observations for transitions observed in the months of February through July of
2020, although the two-month matched estimation samples end in June.
The key explanatory variables are the individual’s imputed UI replacement rate (Ri)
under the CARES Act and its interaction with an indicator for observations corresponding to the
months of April through July 2020, when the CARES Act $600 supplement was available. The
replacement rate with the $600 supplement included varies across individuals but not over time
and hence is not the key source of variation in this equation. Instead, the treatment effect of the
$600 CARES supplement is captured by the impact of the replacement rate after the CARES Act
was implemented and the supplemental payments were available. This period began in late
March 2020, between the March and April CPS reference periods. These effects are estimated by
the coefficient (π) on the interaction between the replacement rate Ri and the indicator for the
months of April through July (or June for the two-month matched sample). This represents a
conventional before/after estimation approach with regression controls, with the months of
February and March combined used as the baseline control period.
The regression also includes an extensive set of other controls. The vector Xi,t-1 consists
of individual characteristics observed in the base (pre-transition) month: age (eight categories),
education (five categories), race/ethnicity (five categories), gender by marital status, broad
occupation (10 categories) and industry (13 categories) of prior employment, and duration to
date of the individual’s unemployment spell (10 categories, with the final category indicating
Page 18 of 26
Valletta-Yilma, Unemployment Insurance
duration of longer than one year). The model also includes indicators for calendar months (γt)
and several state/month labor market controls (Zs,t-1): cubics in the state unemployment rate and
three-month employment growth rate.9
We report the parameter estimates from our logit model as average marginal effects. All
estimates are weighted by the longitudinal weights that adjust the sample for the characteristics
of the sequentially matched observations. The analysis is restricted to individuals with non-zero
estimated UI replacement rates under the CARES Act—i.e., individuals who are identified as
eligible to receive UI payments based on their prior earnings history. This is a direct implication
of our before/after design, since individuals who are not eligible to receive UI payments do not
contribute any identifying variation to the estimation.
D. Results
Tables 2 and 3 present the main results of our regression analyses. Table 2 shows the
results for the single-match sample, and Table 3 shows results for the double-match sample that
corrects for likely reporting error. The tables are constructed identically, with full sample results
in the first column and results by tertiles of household income in columns 2-4.
In both tables, the full sample estimates in column 1 show statistically precise negative
effects of UI replacement rates on job-finding during the period when the $600 supplement was
available. This is the main coefficient of interest, which represents the effect of variation in UI
replacement rates due to the $600 benefit supplement (the coefficient π from equation 1). The
9
Unlike the analysis in Petrosky-Nadeau and Valletta (2023), our regression specification does not
include state dummies, which cannot be consistently estimated using our sub-samples defined by
household income groups. This omission is inconsequential: the full sample results excluding state
dummies are nearly identical to those presented in that earlier paper. Our slightly different regression
specification also produces small differences in sample counts.
Page 19 of 26
Valletta-Yilma, Unemployment Insurance
transformed coefficients reported in the tables indicate a 5 to 7 percentage point reduction in job-
finding rates for a 100 percent increase in the UI replacement rate. As discussed in PNV (2023),
the magnitude of the coefficients implies moderate effects of UI replacement rates on job-
finding, with the size of the implied elasticity of job-finding with respect to the replacement rate
toward the low end of past estimates of this elasticity from the relevant literature. The larger and
more precisely estimated UI effect from the double-match sample in Table 3 compared with
Table 2 shows the effects of the measurement adjustment used for the double-match sample.
This likely reflects the impact of increased turbulence in labor market transitions and relaxation
of normal UI eligibility rules during the early pandemic period. Under these conditions, the
estimation is notably more precise when spurious transitions are reduced in the data (Table 3).
Table 2
Regression results: UI replacement rates and job finding,
by income tertile, CPS single match
(1) (2) (3) (4)
Full sample 1st tertile 2nd tertile 3rd tertile
(low) (middle) (high)
UI rep rate 0.024 -0.023 0.068* 0.043
(0.019) (0.037) (0.038) (0.034)
UI rep*(Apr-Jul) -0.047** -0.004 -0.098** -0.050
(0.021) (0.042) (0.043) (0.035)
Observations 5459 1826 1867 1766
Note: Logit coefficient estimates transformed to marginal effects, with standard
errors in parentheses. See the text for the full list of controls.
* p<.10, ** p<.05, *** p<.01.
Page 20 of 26
Valletta-Yilma, Unemployment Insurance
Table 3
Regression results: UI replacement rates and job finding,
by income tertile, CPS double match
(1) (2) (3) (4)
Full sample 1st tertile 2nd tertile 3rd tertile
(low) (middle) (high)
UI rep rate 0.050** -0.022 0.097** 0.081**
(0.022) (0.039) (0.043) (0.041)
UI rep*(Apr-Jul) -0.072*** -0.011 -0.109** -0.108**
(0.026) (0.052) (0.048) (0.044)
Observations 2868 970 941 938
Note: Logit coefficient estimates transformed to marginal effects, with
standard errors in parentheses. See the text for the full list of controls.
* p<.10, ** p<.05, *** p<.01.
Of particular note in Tables 2 and 3 is the variation in this estimated effect of UI
replacement rates across income tertiles in columns 2-4. For both samples, the estimated effect is
indistinguishable from zero for the lowest income tertile. This suggests that job-search behavior
by individuals from lower-income households is unaffected by the amount of UI benefits
available to them, measured relative to their prior earnings. By contrast, the negative and
significant coefficients for individuals from middle-income households in Table 2 and both
middle-income and higher-income households in Table 3 indicate that they respond to higher UI
replacement rates by reducing their job-search intensity or willingness to accept job offers.
Overall, the job-finding regression results suggest moderate disincentive effects of the
higher UI replacement rates arising from the $600 pandemic UI supplement. Also, despite the
very large increase in replacement rates for lower-income individuals, the reduction in job-
finding associated with higher UI benefit payments was larger for individuals from higher-
Page 21 of 26
Valletta-Yilma, Unemployment Insurance
income households. This latter result may seem surprising, given the well-known finding from
Chetty (2008) that UI benefit effects on job search outcomes are larger for liquidity-constrained
households, which tend to have low rather than high income. However, the Chetty finding is not
consistently reproduced. For example, in their analysis of a 1989 UI benefit increase in New
York state, Meyer and Mok (2014) find a larger effect on job search for individuals with higher
predicted net assets, which is contrary to Chetty (2008) and broadly consistent with our findings
by income group. Moreover, other work has found that the job-search response to variation in UI
benefits is more muted when labor market conditions are weak (e.g., Kroft and Notowidigdo,
2016). Labor market conditions during the early pandemic period were especially weak for low-
wage services workers, whose employment in many cases was directly precluded by limits on in-
person contact. This exceptionally weak labor market for low-wage workers may account for the
limited disincentive effects of the $600 UI supplement on their job-search outcomes.
IV. Discussion and Conclusions
The COVID-19 pandemic created a massive negative shock to the U.S. labor market,
with the effects largely concentrated on lower-wage workers. We find that enhanced UI benefits
during the pandemic eliminated the substantial increase in bottom-half income inequality that
otherwise would have occurred, reversing the inequality-increasing effects of the pandemic.
These findings reinforce and extend past findings about the poverty-reducing effects of UI
benefits during economic crises (Larrimore et al., 2023; Rothstein and Valletta, 2017).
We also find moderate disincentive effects of the historically outsized increase in weekly
UI benefits on job-finding outcomes. These search disincentive effects appear larger for
individuals from households above the bottom third of the income distribution, despite the much
Page 22 of 26
Valletta-Yilma, Unemployment Insurance
higher replacement rate of pre-layoff earnings due to the UI benefits for lower-income
individuals. This finding may reflect the exceptionally weak labor market conditions for low-
wage workers early in the pandemic.
Overall, these findings suggest large equity effects from the pandemic UI expansions and
limited adverse consequences for economic efficiency. Our findings are potentially relevant for
UI policy design. Other authors have assessed the possible efficacy of automatic UI duration
extensions during economic downturns, based on analysis of policy responses during prior
downturns including the Great Recession of 2007-09 (Chodorow-Reich and Coglianese, 2019;
Chodorow-Reich et al., 2022). Our results are informative with respect to the design of policies
that include automatic increases in weekly benefits, an issue that may merit further consideration
in future research.
Page 23 of 26
Valletta-Yilma, Unemployment Insurance
References
Bartik, A. W., Bertrand, M., Lin, F., Rothstein, J., & Unrath, M. (2020). Measuring the labor
market at the onset of the COVID-19 crisis. Brookings Papers on Economic Activity,
Summer, 239-326. https://www.brookings.edu/wp-
content/uploads/2020/06/SU20_S4_Bartik-et-al_-final-paper-1.pdf
Bitler, M. P., Hoynes, H. W., & Schanzenbach, D. W. (2020). The social safety net in the wake
of COVID-19. Brookings Papers on Economic Activity, 2, 119–158.
https://www.brookings.edu/wp-content/uploads/2020/06/SU20_S2_Bitler-et-al_-final-
paper.pdf.
Bitler, M. P., Hoynes, H. W., & Schanzenbach, D. W. (2023). Suffering, the safety net, and
disparities during COVID-19.” RSF: The Russell Sage Foundation Journal of the Social
Sciences, 9(3), 32–59. DOI: 10.7758/RSF.2023.9.3.02.
Cajner, T., Crane, L. D., Decker, R. A., Grigsby, J. G., Hamins-Puertolas, A., Hurst, E., Kurz, C.,
& Yildirmaz, A. (2020). The U.S. labor market during the beginning of the pandemic
recession. Brookings Papers on Economic Activity, Summer, 3-33.
https://www.brookings.edu/wp-content/uploads/2020/06/SU20_S1_Cajner-et-al._-
combined.pdf
Chetty, R. (2008). Moral hazard versus liquidity and optimal unemployment insurance. Journal
of Political Economy, 116(2), 173-234.
Chodorow-Reich, G., & Coglianese, J. (2019). Unemployment insurance and macroeconomic
stabilization. In H. Boushey, R. Nunn, & J. Shambaugh (Eds.), Recession Ready: Fiscal
Policies to Stabilize the American Economy (pp. 153–179). Brookings Institution.
Chodorow-Reich, G., Ganong, P., & Gruber, J. (2022). Should we have automatic triggers for
unemployment benefit duration and how costly would they be?” AEA Papers and
Proceedings, 112, 112–116.
Cortes, G. M., & Forsythe, E. (2023a). Distributional impacts of the covid-19 pandemic and the
CARES Act. The Journal of Economic Inequality, 21, 325-49.
Cortes, G. M., & Forsythe, E. (2023b). Heterogeneous labor market impacts of the COVID-19
pandemic. Industrial and Labor Relations Review, 76(1), 30-55.
Couch, K. A., Fairlie, R. W., & Xu, H. (2020). Early evidence of the impacts of COVID-19 on
minority unemployment. Journal of Public Economics, 192, 104287.
Couch, K. A., Fairlie, R. W., & Xu, H. (2022). The evolving impacts of the COVID‐19
pandemic on gender inequality in the US labor market: the COVID motherhood penalty.
Economic Inquiry, 60(2), 485-507.
Page 24 of 26
Valletta-Yilma, Unemployment Insurance
Daly, M. C., & Valletta, R. G. (2006). Inequality and poverty in the United States: the effects of
rising dispersion of men's earnings and changing family behaviour. Economica, 73(289),
75-98.
Farber, H. S, & Valletta, R. G. (2015). Do extended unemployment benefits lengthen
unemployment spells? Evidence from recent cycles in the U.S. labor market. Journal of
Human Resources, 50(4), 873–909.
Farber, H. S., Rothstein, J., & Valletta, R. G. (2015). The effect of extended unemployment
insurance benefits: evidence from the 2012-2013 phase-out. American Economic Review,
105(5), 171–76.
Ganong, P., Noel, P. J., & Vavra, J. S. (2020). U.S. unemployment insurance replacement rates
during the pandemic. Journal of Public Economics, 191(Nov.), 104273.
Kroft, K., & Notowidigdo, M. J. (2016). Should unemployment insurance vary with the
unemployment rate? Theory and evidence. The Review of Economic Studies, 83(3), 1092-
1124.
Larrimore, J., Mortenson, J., & Splinter, D. (2022). Earnings shocks and stabilization during
COVID-19. Journal of Public Economics, 206(Feb.).
Larrimore, J., Mortenson, J., & Splinter, D. (2023). Unemployment insurance in survey and
administrative data. Journal of Policy Analysis and Management, 42(2), 571-579.
Meyer, B. D., & Mok, W. K. (2014). A short review of recent evidence on the disincentive
effects of unemployment insurance and new evidence from New York State. National
Tax Journal, 67(1), 219-251.
National Academy of Social Insurance. (2023). Social insurance during the pandemic: successes,
shortcomings, and policy options for the future. A Report of the COVID-19 Task Force
Policy Translation Working Group. https://www.nasi.org/wp-
content/uploads/2023/06/Social-Insurance-during-the-Pandemic-_REPORT.pdf
Petrosky-Nadeau, N., & Valletta, R. G. (2023). UI generosity and job acceptance: effects of the
2020 CARES Act. FRBSF Working Paper 2021-13 (revised May 2024). Forthcoming,
Journal of Political Economy Macroeconomics. https://doi.org/10.24148/wp2021-13
Rothbaum, J., & Bee, A. (2021). Coronavirus infects surveys, too: survey nonresponse bias and
the coronavirus pandemic. U.S. Census Bureau SEHSD working paper 2020-10.
https://www.census.gov/library/working-papers/2020/demo/SEHSD-WP2020-10.html
Rothstein, J. (2011). Unemployment insurance and job search in the great recession. Brookings
Papers on Economic Activity, 42(2), 143–213.
Rothstein, J., & Valletta, R. G. (2017). Scraping by: income and program participation after the
loss of extended unemployment benefits. Journal of Policy Analysis and Management,
36(4), 880–908.
Page 25 of 26
Valletta-Yilma, Unemployment Insurance
U.S. Bureau of Labor Statistics (BLS). (2020). Frequently asked questions: the impact of the
coronavirus (COVID-19) pandemic on the employment situation for April 2020.
https://www.bls.gov/covid19/employment-situation-covid19-faq-april-2020.htm
Valletta, R. G. (2014). Recent extensions of U.S. unemployment benefits: search responses in
alternative labor market states. IZA Journal of Labor Policy, 3, 1–25.
Whittaker, J. M., & Isaacs, K. P. (2022). Current status of unemployment insurance (UI)
benefits: permanent-law programs and COVID-19 pandemic response. CRS Report
R46687, January. https://crsreports.congress.gov/product/pdf/R/R46687
Page 26 of 26
File and source
- File
- enhanced-unemployment-insurance-benefits-in-the-united-states-during-covid-19-equity-and-efficiency.pdf
- Size
- 672,707 bytes
- SHA-256
- 98bb34fb9216fbf923214dc393ae74660ded76ec7da178dcfdad794c1debdb90
- Original
- doi.org