View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

September 2021

The April CPI data in historical context
Douglas Himes
The Consumer Price Index (CPI) data for April 2021 (and other recent months) received much attention in the
media because these data showed larger price increases than had been seen in many years. The changes were
not only higher than what had been expected, but they were also higher than the Federal Reserve’s inflation
target. In “Putting recent inflation in historical context” (Economic Synopses, Federal Reserve Bank of St. Louis,
June 3, 2021), authors Matthew Famiglietti and Carlos Garriga compare the April data with past data and ask if we
should be worried, and if so, how much.
The authors compare the price change in the April CPI report with price changes over the past 60 years.
Compared with the entire 1960–2021 period, the price change reported in April 2021 is well within the normal
range. They then divide those 60 years into the “pre-inflation-targeting era” before 1995 and the “inflation-targeting
era” of 1995 to present. This split is based on the Federal Reserve’s decision to adopt policies to achieve a certain
targeted inflation rate. Compared with 1960–1995, the April 2021 data are again basically normal. But compared
with the post-1995 era, the April 2021 data stand out.
Then the authors look at the “core” inflation, which ignores changes in the relatively more volatile food and fuel
prices. While recent core inflation is unremarkable when compared with pre-1995 data, the April core inflation is
the highest it has ever been during the post-1995 period. This finding shows that the volatile food and fuel prices
are not predominately responsible for the increase in the overall CPI rate. The authors also state that the April
2021 CPI data had a “surprise” factor, in the sense that they were so unlike the immediately preceding months’
data.
With Famiglietti and Garriga concluding that April 2021 is a highly unusual month in the context of the post-1995
inflation-targeting era, how should we respond to it? The data are indeed highly unusual, but April 2021 was an
unusual month. The authors see no need to worry, yet, closing with, “While the April 2021 inflation report is
undeniably historically high, it also follows an unprecedented turbulent period for the economy for which the
historical evidence might provide imperfect guidance.”

1

September 2021

A $15 minimum wage changes more than just
take-home pay
Jonathan Yoe
Does a rise in the minimum wage lead to higher unemployment? Economists have been arguing this question for a
long time, and the “fight for $15” movement has really heated up the debate on how a higher minimum wage
changes employment. According to the basic economics 101 explanation, an increase in the minimum wage
motivates more people to enter the labor market because they will earn more money. At the same time, an
increase in the minimum wage increases firms’ costs and the quantity of labor demanded decreases (firms hire
fewer workers). Now more people are looking for work than there are jobs available, which leads to
unemployment. Although some economists believe this explanation, several studies have found that an increase in
the minimum wage changes employment little, if at all.
In his article “How do firms respond to minimum wage increases? Understanding the relevance of nonemployment margins” (Journal of Economic Perspectives, Winter 2021), economist Jeffrey Clemens argues that
focusing only on how a minimum wage affects employment is a mistake. Clemens explores nonwage aspects of a
job (health insurance, working conditions, schedule flexibility, and production technology) that a firm might adjust in
response to higher costs caused by a minimum wage increase. In his article, he reviews the findings of existing
studies that have examined these nonwage aspects of the job, and he offers his opinion on how policy makers
should use this information to assess using the minimum wage as a tool to improve worker welfare. Clemens
highlights the lack of empirical evidence in many of the areas he discusses and the need for more research.
An increase in the minimum wage increases firms’ costs and lowers their profits. Firms often try to pass that cost
through to the consumer in the form of higher prices for their products. Some goods are made to be consumed
where they are purchased, and others have no viable substitutes—haircuts, restaurant meals, electricity, water,
and construction. Firms can raise the price of these goods and pass the minimum wage-related extra costs on to
the consumer. Clemens also cites a study that argues that minimum wage hikes can be regressive because a
large proportion of low-income households’ budgets go toward products and services made by minimum wage
workers, such as restaurants. He even cites the 2019 Consumer Expenditure Survey to show that food (both at
and away from home) accounts for 15 percent of the budget of low-income homes but only 10 percent of the
budget of high-income homes. All of this means that the increase in the income of those receiving the minimum
wage is thus smaller than it was intended to be.
Other ways that firms can adjust to minimum wage increases is by making noncash benefits (such as pensions,
health insurance, and paid leave) less generous. With a higher minimum wage, health insurance may become less
generous for all employees, even the higher skilled employees not directly affected by the minimum wage. From
their perspective, they would be worse off than before.

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Facing increased labor costs, firms may buy cheaper office equipment or offer fewer training opportunities. Firms
may also replace lower skilled, less-educated workers with higher skilled, more educated workers. If firms were to
hire a greater number of high-skilled workers, low-skilled workers would have more difficulty finding work. Firms
might also have the option to replace people with technology.
Although many studies do not find that an increase in the minimum wage causes a rise in unemployment, Clemens
argues those studies do not use models that are sufficient to capture other changes in social welfare that an
increase in the minimum wage might cause. To compensate for their loss of profit due to their rising labor costs,
firms might manipulate many nonwage aspects of a job, such as less generous health insurance, pensions, and
amenities; and replacement of lower skilled minimum wage earners with higher skilled workers or with technology.

2

September 2021

Am I getting the best price?
Richard Hernandez
With the help of the internet, consumers can easily search to find the lowest possible price of a good from a variety
of retailers. This has led to online retail sales growing at a faster annual rate (16 percent) than offline retail sales (3
percent). Have you ever wondered if the online price is better than the offline price? In a recent working paper,
“The pricing strategies of online grocery retailers” (National Bureau of Economic Research, Working Paper 28639,
April 2021), authors Diego Aparicio, Zachary Metzman, and Roberto Rigobon explain the differences between
uniform pricing and algorithmic pricing and investigate how online grocery stores set prices depending on certain
variables.
Uniform pricing is a strategy in which a retail chain sets a good to a single price at all its stores in a specific area.
For example, the price of dog shampoo is set and sold at one price at any location from the same retail chain
within a geographic area. Alternatively, algorithmic pricing is a strategy that allows a computer to change the price
of a good multiple times a day to optimize profits based on a set of variables that the algorithm learns over time.
For example, after learning how consumers are purchasing gasoline throughout the day, a gas station owner or
manager would lower prices when the station is not busy and increase prices during rush hour. Aparicio and
colleagues study these two pricing strategies to compare price differentiation between grocery firms.
To compare these two pricing strategies, the authors gathered price data multiple times a day across locations and
compared prices within a retailer (online versus offline prices) and across retailers (online and offline prices). The
authors discovered that a retailer changed and varied its online prices more often than the prices of its competitor
and the retailer also was less likely than its competitor to match its offline prices to its online prices. However, the
authors found that the share of private label products of an online grocery store had the same price 90 percent of
the time. The authors believe the difference between prices was because of price stickiness and that a retailer
controls the price of its own manufactured products. Of the two largest online grocery stores, which account for
almost 50 percent of all online grocery sales, the authors found that each changed (either increased or decreased)
the price of any particular product that same day about 7 percent and 8 percent of the time. And as time passes,
the chance of a monthly price change increases to about 74 percent and 50 percent for the largest and second
largest online grocery stores, respectively. This finding clearly indicates an algorithmic pricing strategy.
The authors also investigated how shipping costs could affect online prices. They found that as the delivery
distance increases, the price that the online shopper sees also increases. A 10-mile increase in shipping distance
would add a 0.14-percentage-point increase to the price of a product from the same retailer. Local demographics
(home values, annual income per capita, etc.) play an even smaller part and only moderately influence local
prices.

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

In conclusion, a takeaway from the authors’ research is that online prices of a grocery store are often changing
more rapidly than its offline prices. The retailers’ own products are less likely to have a price change and are more
likely to stay consistently at one price. Delivery distance matters and can add to the cost of a product, and local
demographic information is less relevant to how a grocery store prices its product.

2

September 2021

The COVID-19 small business boom: startups
surge during pandemic
Maureen Soyars Hicks
According to Haltiwanger’s analysis, 2020 was a record year for new business formation, with “the total number of
applications in 2020 . . . the highest by far compared to all years for which the data have been available.” Further,
“the increase from 2019 to 2020 in total applications exceeds 20 percent which is double the growth rate in any
other year.” After a dip in the first half of the year, the entirety of the increase in applications occurred in the last 6
months of 2020 and has continued through May 2021. The author suggests that as more economic data become
available, more small businesses, in total, may have been created during a recession and global pandemic.
The analysis highlights how the pandemic may be accelerating ongoing economic trends. For example,
applications for likely nonemployers increased; nonemployer businesses have been increasing over the last 15
years as employment in the gig economy has increased. Haltiwanger indicates that, for 2020, much of the
nonemployer activity reflects supplemental and stopgap activity.
However, the increase in applications is likely for both employers and nonemployers. Haltiwanger notes that
whether these new businesses will drive job creation remains to be seen, with the increase in new employer
businesses emerging within 4 to 8 quarters of the date of the applications according to historical patterns.
While the surge in new business applications was “uneven across sectors,” the largest increases were in the
nonstore retail sector (alone accounting for 1 out of every 3 new businesses formed during the pandemic) and the
personal services sector. This finding reflects another continuation of a trend: the shift toward ecommerce, with
small businesses now selling goods and services online instead of renting a storefront.
Haltiwanger notes that some substantial restructuring caused by the pandemic will occur, particularly in retail trade
and accommodation and food services, sectors in which existing small businesses have experienced large
declines during the pandemic. He indicates that as the economy recovers, which of these changes that will “stick”
will become clearer.

1

HOME

ARCHIVES

FOR AUTHORS

ABOUT

SUBSCRIBE

Search MLR

BOOK REVIEW
SEPTEMBER 2021

The dangers of ideology and the virtues of pragmatism
Concrete Economics: The Hamilton Approach to Economic Growth and Policy. By Stephen S. Cohen and
J. Bradford DeLong. Boston, MA: Harvard Business Review Press, 2016, 240 pp., $28.00 hardcover.
“It’s a little disquieting you would let your ideals blind you to reality.” This line from the hit Broadway
musical Hamilton pithily captures the ethos of Berkeley economics professors Stephen S. Cohen and J.
Bradford DeLong’s Concrete Economics: The Hamilton Approach to Economic Growth and Policy. If you
feel trapped in the contemporary polarized social and economic discourses, Cohen and DeLong’s book
offers a sobering and timely escape from excessive idealism by providing a reasoned synthesis of what
policies work in bolstering economic growth. The authors articulate the virtues of pragmatic
policymaking, explaining how policy decisions based on precedence have historically buttressed the U.S.
economy and cautioning about the dangers of policy filtered through ideology. As they put it, “In
successful economies, economic policy has been pragmatic, not ideological. It has been concrete, not
abstract.”
The book is inviting to readers of all levels of economic expertise. It is brief, avoids technical jargon, and
offers no new abstract economic theories—all traits consistent with Cohen and DeLong’s quest for
simplicity. It seeks solely to remind readers of what successful economic policy has looked like through
history, concisely illustrating how the United States has lost its way by sacrificing down-to-earth
economic policy. According to the authors, this shift in policy thinking has contributed to the steep
decline in U.S. manufacturing and the hypertrophying of industries such as finance, healthcare
administration, and real estate.
Cohen and DeLong devote the book’s first three chapters to explaining the economic thinking and policies
of the country’s most prominent economic leaders—starting with Alexander Hamilton, the first Secretary
of the Treasury, and moving on to Presidents Abraham Lincoln, Theodore Roosevelt, Franklin D.
Roosevelt, Dwight D. Eisenhower, and Ronald Reagan. In chapter 4, the authors go on to tackle the
intricacies of the East Asian economic model, exploring what the economic booms of Japan, South Korea,
and China can teach us about successful economic policy. And finally, before their closing statements,
Cohen and DeLong include a penultimate chapter explaining the consequences of abstract and ideological
decision making, highlighting what they see as a poorly regulated and inefficient U.S. finance sector. A
more detailed account of each chapter follows.

GO

In chapter 1, Cohen and DeLong debunk the idea that, since its inception, the United States has been a
Jeffersonian, laissez-faire nation, citing the Hamiltonian system of government intervention and policies
as evidence. According to the authors, Hamilton’s system was based on four key economic drivers: high
tariffs, high spending on infrastructure, the assumption of state debt by the federal government, and
reliance on a central bank. Cohen and DeLong assert that these drivers were quintessential to America’s
rise as an industrial and economic powerhouse, and that Hamilton’s policies and vision spurred the U.S.
global dominance in manufacturing and technology. Long after Hamilton’s fatal duel with Aaron Burr—
and despite Hamilton’s opposition party (led by Thomas Jefferson and James Madison) occupying the
White House for nearly four decades after his death—Hamilton’s policies were sustained and have been
immortalized in the U.S. economic framework. Cohen and DeLong argue that it is this basic, yet
benevolent, system established by Hamilton that holds the key to strong economic growth, boldly
asserting that the ideas in Hamilton’s lesser known Report on Manufactures (1791) are more influential
than those in Adam Smith’s The Wealth of Nations (1776).
In chapter 2, the authors outline the U.S. economic redesigns implemented from the post-Civil War era to
the rollout of FDR’s New Deal during the Great Depression. Social and economic planning by the
government was at a historic high during this period, embodied by efforts like the preservation of high
tariffs, the Homestead Act of 1862, and the Morrill Act of 1862, which allowed Americans to own land
and go to school on the federal dime. There were then, of course, the railroads, which were nearly all
subsidized by the government and, as the authors assert, indispensable to the economic burgeoning of the
United States. But this period also brought in its wake unprecedented levels of inequality and xenophobia
against immigrants, among other issues. Cohen and DeLong argue that President Theodore Roosevelt
helped course-correct these unsettling trends through “pragmatic, practical, and concrete change.” The
Federal Reserve, national parks, antitrust laws, unemployment compensation, and federal retirement and
disability benefit programs were all birthed under Teddy’s presidency to combat economic inequities.
Cohen and DeLong conclude the chapter by underscoring the economic redesign of FDR’s New Deal,
describing that program’s strategies to alleviate the Great Depression as “pragmatic experimentalism: try
one thing and then another; what didn’t work was dropped; what worked was quickly expanded.”
Chapter 3 jumps to “the long age of Ike,” delineating the technological advancements and evolving
framework of American life that emerged as a result of the policies implemented under President Dwight
Eisenhower. Challenging the free-market ideas of thinkers like Milton Friedman, Friedrich Hayek, and
Ludwig von Mises, Cohen and DeLong contend that Eisenhower’s massive federal spending and
government interventionist policies were imperative to achieving U.S. global technological dominance.
Under Ike, federal spending as a percentage of gross domestic product was double that of FDR’s New
Deal spending peak, providing funding for measures such as the GI Bill and the National Defense
Highway Act. And military spending—even after the Korean War—was approximately 2.5 times as high
as it was under Presidents Bill Clinton and George W. Bush. Nuclear power, commercial jetliners,
microwave ovens, semiconductors, internet, and computers—nearly all the great technological
innovations of the 20th century—were either subsidized by the federal government or birthed in
government labs. As the authors succinctly summarize, “Mortgage financing for single-family homes;
road building; defense spending and its spin-offs; tight and steady regulation of finance; and government
oversight of big business and big unions. It was all in place by the time Ike came to office and [he]
authoritatively assured its continuation and its massive, smooth, and responsible expansion.”

In chapter 4, Cohen and DeLong highlight the East Asian economic model, citing Japan’s and China’s
extraordinary growth during the 20th century as evidence of the virtues of pragmatic economic
policymaking. For an amalgam of reasons, Japan’s sustained rate of economic growth after World War II
was the highest in modern history, and the authors largely attribute this growth to the country’s
protectionist policies and highly competent, meritocratic government bureaucracy. In the case of China,
Cohen and DeLong ascribe the nation’s impressive economic advancement to Deng Xiaoping, the
“Architect of Modern China” who, after decades of failed campaigns aiming to rear the country out of a
poor peasant economy, “set China on what has proven to be an unimaginably successful trajectory of
structural economic reform.” In both countries—and just like the United States of Hamilton’s time—the
successful reformers sought to shift their nations’ economic comparative advantages from agriculture to
high-value manufacturing.
Lastly, chapter 5—the denouement of Cohen and DeLong’s economic policy synthesis—illustrates the
consequences of supplanting pragmatism with ideological extremity in economic decision making,
highlighting the U.S. finance sector as a case study. Included in FDR’s New Deal policies were safeguards
and regulations of the finance industry that were deemed acceptable and necessary to prevent another
disaster after the Great Depression and to keep banks and other portfolio managers in check. But starting
in the 1970s, a slew of initiatives to deregulate the U.S. finance sector were incrementally rolled out,
representing an idealistic effort to free markets from restrictive government intervention. The repeal of the
Glass-Steagall Act and Regulation Q, as well as the implementation of the Depository Institutions
Deregulation Act and the Garn-St. Germain Act, are just some of the initiatives Cohen and DeLong cite as
reasons for the artificial enlargement of the U.S. finance sector and for subsequent catastrophes such as
the 2007–08 financial crisis. The finance sector of the 1950s constituted only 3.7 percent of the U.S.
economy, compared with over 8.5 percent today, and the authors assert that the more regulated 1950s
economy provided more safety from crisis and economic inequality without hobbling economic growth.
A growing finance sector is supposed to allocate capital across industries more efficiently, but it is not
clear how or where. And despite advancements like faster computers and credit derivatives, some experts,
such as former Federal Reserve Chair Paul Volcker, have concluded that the ATM has been the only
worthwhile financial innovation in the past three decades.
On the last page of the book, Cohen and DeLong outline their call to action: “We propose one change…
shift discussion of economic policy to the concrete, where it had recurrent successes.” As I was reading
the book’s closing pages, I certainly hoped the authors would offer more “concrete” policy solutions to
the economic issues presented to us, but I was unsurprised that they chose not to do so. First, if they did
suggest policy solutions, it may have violated their impartiality. Second, the book was intended as a
conversation starter, not a panacea to all our economic problems. Indeed, the authors suggest that the
answers to these problems may just be hiding in plain sight, concealed in historical precedent and
overshadowed by theoretical abstraction. In the simplest terms, they ask us to look at what worked in the
past and repeat it.
Although I think Cohen and DeLong strive for impartiality, potential counterarguments to their assertions
were found wanting. It is clear that the authors are followers of John Maynard Keynes (they inserted three
block quotes from Keynes in their six-page concluding statements), so I was often left wondering how
Keynes’s greatest philosophical rivals would respond to their assertions. What would Hayek say about the
authors’ claim that FDR’s New Deal was not ideological? What would Friedman, who was skeptical of
the government’s ability to innovate, say in response to all the technological advancements that emerged

from government labs in the last century? Would Mises agree that regulation and government intervention
in the finance sector are necessary? Although these and similar questions remain unanswered, the authors
make stalwart arguments, and I think they are right.
The book coaxes a kind of economic self-awareness; it is easy to be entranced by beguiling ideological
speeches filled with promises of prosperity, but Cohen and DeLong attempt to bring the reader back to
reality. It is also easy to look at the U.S. economy through rose-tinted glasses, but the book’s tough
analysis on the contemporary problems facing the United States is cautionary and enlightening. I hope to
see another down-to-earth piece from Cohen and DeLong soon.

ABOUT THE REVIEWER

Paul Garbarino
garbarino.paul@bls.gov
Paul Garbarino is an economist in the Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.

September 2021

Applying for and receiving unemployment
insurance benefits during the coronavirus
pandemic
During the coronavirus disease 2019 pandemic in the
United States, claims for unemployment insurance (UI)
benefits rose sharply because of the substantial job loss
and the expansion of UI programs. To improve upon UI
administrative data, in this article, we use the Household
Pulse Survey to estimate the number of people who applied
for UI benefits, the number of people who received benefits,
and the success rate of UI applicants (the share of
applicants who received benefits) during the first 9 months
of the pandemic. We examine differences by demographic
group, educational attainment, and prepandemic household
income. In addition, we relate state-level estimates to UI
recipiency before the pandemic, job loss during the
pandemic, and the differential spread of the coronavirus
across states. Compared with individuals who applied for UI

Patrick Carey
carey.patrick@bls.gov

benefits but did not receive them, we find that individuals
who received benefits had greater well-being in a variety of
domains, including household finances, food security, and
mental health.
The coronavirus disease 2019 (COVID-19) pandemic has

Patrick Carey is the Assistant Commissioner for
the Office of Current Employment Analysis, U.S.
Bureau of Labor Statistics.
Jeffrey A. Groen
groen.jeffrey@bls.gov

greatly affected the U.S. labor market starting in March
2020. More than twice as many jobs were lost in the initial
months of the pandemic than during the Great Recession
(2007–09). In addition, only a third of those jobs were
recovered in the subsequent 2 months.1 The large number
of job losses at the beginning of the pandemic caused the
employment–population ratio to plummet from 61.1 percent
in February 2020 to 51.3 percent in April 2020. By
December 2020, it had partially recovered to 57.4 percent.2

1

Jeffrey A. Groen is a research economist in the
Office of Employment and Unemployment
Statistics, U.S. Bureau of Labor Statistics.
Bradley A. Jensen
jensen.bradley@bls.gov
Bradley A. Jensen is an economist in the Division
of Local Area Unemployment Statistics, U.S.
Bureau of Labor Statistics.

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Anne E. Polivka
polivka.anne@bls.gov

In the midst of the initial wave of job losses, the federal
government made several major changes to the
unemployment insurance (UI) program.3 The Families First
Coronavirus Response Act (FFCRA), enacted on March 18,
2020, allowed states to relax several conditions for
applicants to receive UI benefits.4 These conditions include
the requirement that applicants be actively seeking work
and the requirement that applicants left work because of an
employer action, such as a layoff.5 The Coronavirus Aid,
Relief, and Economic Security (CARES) Act, enacted on
March 27, 2020, created three federally funded temporary

Anne E. Polivka is a supervisory research
economist in the Office of Employment and
Unemployment Statistics, U.S. Bureau of Labor
Statistics.
Thomas J. Krolik
krolik.thomas@bls.gov
Thomas J. Krolik is Chief of the Division of Local
Area Unemployment Statistics, U.S. Bureau of
Labor Statistics.

UI programs:6
1. The Pandemic Unemployment Assistance (PUA)
program expanded eligibility for UI benefits to self-employed workers, independent contractors, and parttime workers.
2. The Pandemic Emergency Unemployment Compensation program extended benefits by 13 weeks for those
persons who exhaust their regular UI benefits.
3. The Federal Pandemic Unemployment Compensation (FPUC) program provided a $600-a-week
supplement to UI benefits through July 31, 2020.
After the FPUC program expired, on August 8, 2020, the President issued an executive order that allowed states to
supplement UI benefits by $300 a week, funded by federal disaster relief aid (Lost Wages Assistance), for up to 6
weeks of unemployment.
Because of the expansion of UI programs and substantial job loss, the number of initial claims rose tremendously
during the pandemic. When a person applies for UI benefits, an initial claim is the first claim filed by the person in
determining eligibility for benefits. A state UI office reviews each initial claim and either accepts or rejects it. If the
claim is accepted, benefits are paid. As shown in chart 1, initial claims for benefits under the regular UI program,
which were around 200,000 a week before the pandemic, shot up to about 6 million a week in late March and early
April 2020. After that time, initial claims fell as the pandemic progressed but remained above 2 million a week
through mid-May and above 1 million a week through early August. By the end of December, initial claims were
around 800,000 a week.

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Estimates of the number of people who applied for UI benefits and the number of people who received benefits
during the pandemic are useful for measuring the impact of the UI program. UI benefits are a key part of the
federal government’s policy response to the pandemic.7 With individual-level UI estimates, analysts can precisely
compare the extent of UI payments with the extent of stimulus payments and other less targeted government
interventions that directly support individuals.
In this article, we use the Household Pulse Survey (HPS) to estimate the number of people (and share of the adult
population) who applied for UI benefits and who received benefits during the first 9 months of the pandemic (March
through December 2020). We also estimate a measure of the success rate of UI applicants in obtaining benefits:
the share of applicants who received benefits. To show which groups were more or less likely to receive UI
benefits during the pandemic, we examine differences by demographic group, education, and prepandemic
household income. These estimates can be compared with the pattern of job losses during the pandemic, as
documented in other research (as discussed later in this article).
States are important as program administrators in the federal–state UI system and as geographical areas
capturing the spatial dimension of labor markets and the spread of the coronavirus. Given states’ importance, we
estimate for each state the percentage of the population who applied for UI benefits, the percentage of the
population who received benefits, and the success rate of UI applicants. We relate these state-level estimates to
UI recipiency before the pandemic, job loss during the pandemic, the differential spread of the coronavirus across
states, and the differential use of COVID-19 restrictions by state governments. The relationship between these

3

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

state-level estimates and UI recipiency before the pandemic is intended to partially capture differences that existed
in state UI programs before the pandemic.
Our analysis investigates some empirical consequences of the expanded eligibility for UI benefits to self-employed
workers, low-wage workers, and other workers typically not covered in the regular UI program. The UI outcomes
(application rate, receipt rate, and success rate) for workers with less education and lower prepandemic household
income show the extent to which the UI program helped low-wage workers during the pandemic. Evaluating UI
outcomes for the self-employed shows, in part, how successful the expansion of the UI program to self-employed
individuals during the pandemic was. We also analyze UI outcomes for parents of school-age children who faced
varying levels of disruption to their school routines, which is relevant for the expansion of the range of “good
cause” exceptions for leaving work to include caring for a family member.
Finally, we go beyond measuring the receipt of UI benefits to examine whether receiving benefits improved the
well-being of individuals and their households. This question is critically important during the pandemic, given the
large number of people who received benefits and the expanded benefit amounts (because of the federal
supplement). The survey data provide measures of well-being in a variety of domains, including household
finances, food security, and mental health. To gauge the effect of receiving benefits on well-being, we compare
those who received benefits with those who applied for benefits but did not receive them. Taken together, our
analysis provides information relevant to evaluating how successful the targeted UI program was in mitigating the
effects of the pandemic on individuals who were directly affected by it.

Household Pulse Survey
The Employment and Training Administration (ETA) publishes various reports that summarize UI administrative
data at the state level.8 It is not possible to use the data in these reports to gauge the number of people who
applied for UI benefits and the number of people who received benefits across all UI programs during the
pandemic.9 The main reason we cannot use the ETA data is that over time, a person can apply for and receive
benefits from more than one UI program. Thus, simply summing the claims for each program would result in an
unknown degree of double counting.10 The number of individuals included in more than one UI program may be
substantial during the pandemic because many states have required individuals to exhaust or be denied regular
state UI benefits to be eligible for benefits under the pandemic-related programs. Double counting of individuals
during the pandemic can also occur within programs, particularly when one is trying to determine the number of
people who have applied for benefits.11
Given these issues with the administrative data, we use the HPS to construct individual-level estimates. The HPS
is an experimental rapid-response survey designed to measure how the coronavirus pandemic is affecting U.S.
households from a social and economic perspective. The U.S. Census Bureau conducts the survey in partnership
with other federal agencies, including the U.S. Bureau of Labor Statistics (BLS). The HPS is a 20-minute online
survey conducted in 66 sample areas throughout the nation. The sample frame is based on the Census Bureau’s
Master Address File records and existing email and telephone records. The HPS started on April 23, 2020, and
phase 1 lasted until July 21, 2020. The data for this article were collected during the period August 19–December
21, 2020, which covers all of phase 2 (August 19–October 26) and the first part of phase 3 (which started October
28).

4

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Phase 2 was made up of five data collection periods (each was about 2 weeks), with a separate sample for each
period. In each period, the sample size was about 1.03 million housing units. Households chosen for the sample
were contacted by email and text message and asked to complete the survey online. Across periods, the number
of respondents varied from 88,716 to 109,051, with corresponding response rates between 8.1 percent and 10.3
percent (see appendix table 1). Over the first four data collection periods of phase 3, the sample size was similar
to that for phase 2, but the response rates were lower (5.3 percent to 6.7 percent). The sampling methods and
questionnaire were the same in phase 2 and phase 3. Although the HPS data are timely and relevant, we caution
that the data are labeled as experimental and do not necessarily meet the high standards of other Census Bureau
and BLS data products. For example, how representative HPS respondents are of the entire U.S. population has
not been fully explored.12
In households sampled for the HPS, one adult responds to the survey. These adults report individual experiences
only for themselves, not other members of their household. Throughout our analysis, we use the person weights
created by the Census Bureau.13 The person weights were designed to produce biweekly estimates for the total
number of persons age 18 and older living within housing units. The Census Bureau created these weights by
adjusting the household-level-sampling base weights for various factors to account for nonresponse, adults per
household, and coverage. In addition, the person weights are controlled to independent population controls by
various demographics within each state. The demographic characteristics involved in this process are age, gender,
race, Hispanic origin, and educational attainment.14
The HPS has several advantages for studying UI. It includes questions on whether individuals applied for UI
benefits during the pandemic and, if so, whether they received benefits. The responses to these questions can be
used to produce unduplicated estimates of the number of people who have applied for benefits and received
benefits across all the UI programs in place during the pandemic. The HPS also collects information on
demographics, educational attainment, total household income (from 2019), current work status, and the reason
for not working. Various measures of household and individual well-being can be constructed from the HPS
questions covering several domains, including household finances, food security, and mental health.
The information on who applied for UI benefits and who received benefits comes from two questions that BLS
designed and tested.15 The first question asks about applying for benefits: “Since March 13, 2020, have you
applied for Unemployment Insurance (UI) benefits?” For those who answer “yes” to this question, a follow-up
question asks, “Since March 13, 2020, did you receive Unemployment Insurance (UI) benefits?” March 13 is used
to indicate the beginning of the pandemic in the United States. On this date, the President declared a National
Emergency concerning COVID-19.16 A third question asks about the receipt of benefits at the household level:
“Including yourself, how many people in your household received Unemployment Insurance (UI) benefits since
March 13, 2020?” We do not use responses to the third question in this article.17 Because the questions refer to UI
generically, we interpret the survey responses to encompass both the regular UI program and the pandemic UI
programs.
To measure employment at the time of the survey, the HPS asks, “In the last 7 days, did you do ANY work for
either pay or profit?” For those who answer “yes,” a followup question is asked to determine whether the
respondent is employed by government, a private company, or a nonprofit organization; self-employed; or working
in a family business. For those who are not working, the HPS asks, “What is your main reason for not working for
pay or profit?” The respondent is given a list of 13 potential reasons, and we classify the responses into three

5

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

categories: COVID-19-related reasons that are employer initiated, COVID-19-related reasons that are not
employer initiated, and non-COVID-19-related reasons (see appendix table 2).
Note that the reference period (the last 7 days) for the employment questions is different from the reference period
(between March 13 and the survey date) for the UI questions. The UI questions identify those people who have
received benefits at some point during the pandemic but were not necessarily receiving benefits at the time of the
survey. Similarly, those who were working at the time of the survey may have received benefits during the
pandemic, although not necessarily while they were working. All respondents were asked the UI questions,
regardless of whether they were currently working. The employment information reported in the HPS reflects the
respondent’s situation at the time of the survey but does not necessarily reflect the respondent’s situation at other
points during the pandemic. In addition, the HPS does not provide any information about the respondent’s
employment situation before the pandemic.

Aggregate estimates
We use the HPS responses together with the person weights to estimate the number of people who applied for UI
benefits and the number of people who received benefits during the first 9 months of the pandemic (March 13 to
December 21, 2020). Our estimates are based on the combined sample of 775,788 HPS respondents from August
19 to December 21, 2020. Our estimates refer to the population age 18 and older.
We estimate that 52 million people applied for UI benefits from March 13 to December 21, or 21.0 percent of the
U.S. adult population of 249 million. We also estimate that 40 million people received UI benefits during the
pandemic, or 16.0 percent of the adult population.18 By way of comparison, 84 percent of the population age 18
and older received stimulus payments (i.e., Economic Impact Payments) that were authorized as part of the
CARES Act.19 This comparison demonstrates the extent to which UI benefits were more targeted toward those
experiencing economic hardship than were stimulus payments.
We estimate that 77.2 percent of those who applied for UI benefits since March 13 had received benefits by the
survey date (August 19 to December 21, 2020). This measure, which we refer to as the success rate, reflects both
individual eligibility and the capability of state UI offices to process claims.20 Given that some people applied for UI
benefits before the survey date and started receiving benefits after the survey date, our estimated success rate is
an underestimate of the share of applicants who ultimately received benefits. However, given the long reference
period (back to March 13, 2020) and the time pattern of initial claims (the spike in initial claims was early in the
pandemic), the extent of the understatement is likely not large.

Estimates by demographic characteristics, education, and household
income
Our estimates by demographic characteristics are shown in table 1. In terms of the share of the population who
applied for UI benefits or the share of the population who received benefits, no gender differences essentially exist:
about 21 percent of men and women applied for UI benefits and about 16 percent received benefits. The lack of
gender differences in these estimates makes the pandemic recession stand out from prior U.S. recessions. In prior
recessions, the share of women who applied for (or received) UI benefits was lower than the corresponding share
of men.21 Therefore, our HPS estimates reflect a larger role for employment declines among women in the

6

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

pandemic recession (especially in the initial wave of job losses) than in prior U.S. recessions.22 The success rate
is the same for men and women. Regardless of gender, those with children present in the household were more
likely to receive UI benefits than were those without children present in the household.
Table 1. Number and percentage of people 18 years and older who applied for UI benefits and received UI
benefits, and success rate of UI applicants, by demographic characteristics, March 13–December 21, 2020
Applied for UI
Characteristic

Success rate
Number

Total
Gender
Male
Female
Race
White
Black
Asian
Other
Ethnicity
Hispanic
Not Hispanic
Marital status
Married
Widowed
Divorced
Separated
Never married
Presence of children
Children present
No children present
Age (years)
18 to 24
25 to 34
35 to 44
45 to 54
55 to 64
65 to 74
75 and older

Received UI

Total
Percent

Number

Percent

249,170,916

52,430,773

21.0

39,984,667

16.0

77.2

120,531,610
128,639,306

25,133,339
27,297,434

20.9
21.2

19,190,162
20,794,506

15.9
16.2

77.2
77.2

188,635,899
31,020,064
14,019,197
15,495,756

36,501,427
8,835,766
3,078,922
4,014,657

19.4
28.5
22.0
25.9

28,210,496
6,341,967
2,527,595
2,904,609

15.0
20.4
18.0
18.7

78.2
72.8
82.9
73.2

42,320,445
206,850,471

10,629,962
41,800,810

25.1
20.2

7,924,760
32,059,907

18.7
15.5

75.6
77.6

136,555,176
10,693,257
29,263,521
5,528,055
64,821,419

24,102,096
1,148,908
6,715,409
1,670,992
18,515,480

17.7
10.7
22.9
30.2
28.6

18,986,647
794,199
5,099,306
1,134,385
13,743,083

13.9
7.4
17.4
20.5
21.2

79.9
70.5
77.0
68.5
74.8

98,210,373
150,960,543

23,726,520
28,704,252

24.2
19.0

17,647,146
22,337,521

18.0
14.8

75.3
78.7

26,929,445
45,731,131
43,318,164
41,184,810
43,852,105
35,693,575
12,461,685

6,903,784
12,879,612
10,694,413
9,621,711
8,352,029
3,322,848
656,376

25.6
28.2
24.7
23.4
19.0
9.3
5.3

4,902,683
9,558,623
8,260,628
7,496,943
6,673,140
2,651,893
440,757

18.2
20.9
19.1
18.2
15.2
7.4
3.5

71.4
74.8
78.1
79.1
81.2
81.5
68.5

Note: Other race includes any other race alone and multiple races. Success rate excludes individuals who applied for UI benefits but did not answer the
question about receiving benefits. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

Among racial groups, Blacks had the highest rates of applying for UI benefits (28.5 percent) and receiving benefits
(20.4 percent). However, the success rate was lower among Blacks (72.8 percent) than among Asians (82.9
percent) and Whites (78.2 percent). Hispanics were more likely to have applied for and received UI benefits than
were non-Hispanics. The success rate was somewhat lower among Hispanics (75.6 percent) than among nonHispanics (77.6 percent). The higher rates of applying for UI benefits among Blacks and Hispanics are consistent

7

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

with Black and Hispanic workers being disproportionately affected by job losses, layoffs, and disruptions to small
businesses during the pandemic.23
Estimates by educational attainment and household income are shown in table 2. Among individuals age 25 and
older, those with bachelor’s degree or higher were less likely to have applied for and received UI benefits than
were those with a high school education or some college. The success rate generally increased with the level of
education. However, the rate was lowest among those with less than a high school education (68.8 percent) and
highest among those with a bachelor’s degree (80.8 percent). People with lower household incomes in 2019 were
more likely to have applied for and received UI benefits (see chart 2), which suggests that they had fewer nonwage
income sources to draw on.24 For example, among those with incomes of less than $25,000, 28.1 percent applied
for benefits and 19.6 percent received benefits. By comparison, among those with incomes of $200,000 or more,
9.2 percent applied for benefits and 7.4 percent received benefits. As it did with education, the success rate
generally increased with household income.
Table 2. Number and percentage of people who applied for UI benefits and received UI benefits, and
success rate of UI applicants, by household income and education, March 13–December 21, 2020
Applied for UI
Characteristic

Success rate
Number

People 18 years and older
Total
Household income (2019)
Less than $25,000
$25,000 to $34,999
$35,000 to $49,999
$50,000 to $74,999
$75,000 to $99,999
$100,000 to $149,999
$150,000 to $199,999
$200,000 or more
People 25 years and older
Total
Education
Less than high school
Some high school
High school graduate
Some college, no degree
Associate’s degree
Bachelor’s degree
Graduate degree

Received UI

Total
Percent

Number

Percent

249,170,916

52,430,773

21.0

39,984,667

16.0

77.2

25,045,832
19,757,039
22,492,488
31,800,675
24,397,565
27,519,310
12,727,059
14,008,492

7,041,988
5,504,916
5,553,253
7,108,964
4,664,153
4,401,837
1,680,504
1,290,248

28.1
27.9
24.7
22.4
19.1
16.0
13.2
9.2

4,917,357
4,095,912
4,303,257
5,653,248
3,794,580
3,621,543
1,363,348
1,035,537

19.6
20.7
19.1
17.8
15.6
13.2
10.7
7.4

70.4
75.1
78.3
80.4
82.5
83.1
81.7
80.9

222,241,471

45,526,989

20.5

35,081,984

15.8

78.0

5,923,601
12,457,370
68,430,320
42,223,238
21,797,962
38,969,438
32,439,542

1,476,560
3,133,863
15,925,149
10,179,117
4,943,719
6,552,662
3,315,919

24.9
25.2
23.3
24.1
22.7
16.8
10.2

992,972
2,300,340
12,187,372
7,967,355
3,840,851
5,239,601
2,553,494

16.8
18.5
17.8
18.9
17.6
13.4
7.9

68.8
74.4
77.7
79.0
78.7
80.8
78.0

Note: Success rate excludes individuals who applied for UI benefits but did not answer the question about receiving benefits. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The higher rates of applying for UI benefits and receiving benefits among those with less education and lower
income are consistent with employment declines during the pandemic that were greater among low-wage
workers.25 This pattern may also reflect the higher replacement rates that arose from the $600-a-week federal
supplement (available from April through July 2020), which was independent of a worker’s prior wage and
therefore had the largest effect on replacement rates for low-wage workers.26 Most workers, especially low-wage
workers, could receive more money from the enhanced UI benefits than they received in wages while working.27
Workers with less education and lower income had lower success rates (see chart 3). This finding could reflect that
low-wage workers were less likely to be eligible for regular UI (in part, because of not meeting the minimum
earnings requirement), although the PUA program expanded UI eligibility for individuals lacking sufficient work
history and individuals working part-time hours.28 The pattern of success rates by race may partly reflect the
higher prevalence of low-wage work among Blacks compared with Whites and Asians. In 2019, according to data

9

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

from the Current Population Survey, median weekly earnings for full-time workers were $735 for Blacks, $945 for
Whites, and $1,174 for Asians.29

Estimates by state and relationship to state-level factors
The UI system is a federal–state partnership that is funded by federal and state taxes on employers and
administered by states. States set eligibility, duration, and benefit levels within federal guidelines. Because of state
differences within the UI system, we examine variation by state in the share of the population who applied for UI
benefits, the share of the population who received benefits, and the success rate of UI applicants. In addition to
their role in administering the UI program, states show the geographic dimension to labor markets, COVID-19
spread, and COVID-19 restrictions imposed by governments. During the pandemic, states and local areas have
imposed a variety of restrictions on economic activity to slow the spread of COVID-19, including closing
nonessential businesses, closing schools and daycare facilities, requiring residents to stay at home, canceling
public events, and restricting the size of gatherings.
The estimated share of the population who applied for UI benefits varies substantially across states, ranging from
11.4 percent to 30.3 percent (see appendix table 3). Four states have more than 25 percent of their population who
applied (Nevada, Michigan, Hawaii, and New York). Three states have less than 13 percent who applied (Utah,
South Dakota, and Wyoming). The estimated share of a state’s population who received benefits follows a similar
pattern, ranging from 7.6 percent to 25.5 percent. States with a larger share of their population who applied for

10

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

benefits also have a larger share of their population who received benefits, and the relationship is very tight (see
chart 4).

What accounts for this variation across states? To understand the variation, we examine how the HPS state-level
estimates relate to the extent of UI coverage in a state before the pandemic, state-specific job loss during the
pandemic, and the differential spread of the coronavirus across states. Each of these facets is discussed
separately in the later paragraphs.
As a measure of the extent of UI coverage in a state before the pandemic, we use the UI recipiency rate in 2019.
This measure, which is the share of unemployed workers who received UI benefits, varies widely across states: 7
states had recipiency rates of greater than 40 percent in 2019, whereas 17 states had recipiency rates of less than
20 percent.30 Prior research indicates that variation in state recipiency rates reflects state laws (regarding who is
covered by UI and for how long) and administrative practices, although labor market variables and other factors
are also important.31 States with higher recipiency rates before the pandemic had both a larger share of the
population who applied for UI benefits and a larger share of the population who received benefits (see chart 5a).32
This relationship suggests that despite the temporary expansions of UI coverage during the pandemic, preexisting
aspects of a state’s UI system still mattered for the number of people who received UI benefits during the
pandemic.

11

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

States with a higher average unemployment rate between March and December 2020 had both a larger share of
the population who applied for UI benefits and a larger share of the population who received benefits (see chart
5b). This relationship provides support for the HPS-based estimates, but it is not surprising because many people
who are classified as unemployed receive UI benefits. However, receiving UI benefits and being unemployed are
not the same.33 For instance, some people who are unemployed do not qualify for UI benefits (e.g., new entrants
to the labor force and those who do not have sufficient work experience), some people who are eligible for UI
benefits do not apply for benefits, and some people who are receiving UI benefits are not considered unemployed
(e.g., if they are not searching for work).34 In addition, some people are classified as employed but still eligible for
UI benefits (e.g., those whose hours were reduced by their employers or who were participating in a work-sharing
program).

12

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The share of the population who applied for UI benefits and the share of the population who received benefits are
also positively correlated with the cumulative number of COVID-19 deaths per 100,000 population (see chart 5c).
This effect presumably operates through the disruptions to the labor market caused by the coronavirus and by
countermeasures taken by states and local areas to combat the virus’s spread. Another COVID-19 measure that
focuses on states’ countermeasures is the Stringency Index compiled by the Oxford COVID-19 Government
Response Tracker.35 This index measures the strictness of states’ closure and containment policies that primarily
restrict people’s behavior. This measure is also positively correlated with the shares of the population who applied
for or received UI benefits during the pandemic, and the magnitude of the correlation is stronger than that for the
COVID-19 death rate (see chart 5d).

13

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

14

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The success rate of UI applicants varies from 63.9 percent (Arkansas) to 87.1 percent (Rhode Island). States with
higher recipiency rates before the pandemic have a higher success rate (see chart 5a). Beyond that, one concern
is that states with a larger extent of job loss during the pandemic may have experienced “crowding effects” in their
UI systems—whereby some applicants could not have their applications processed promptly—leading to declines
in the success rate. If this crowding hypothesis were true, the success rate would be negatively correlated with
demand for UI benefits. However, the scatter plots of the drivers of demand for UI benefits show the opposite
pattern: if anything, the success rate is positively correlated with the unemployment rate, COVID-19 deaths per

15

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

capita, and the COVID-19 Stringency Index (see charts 5b–5d). As a complementary way of looking at the data,
the success rate is positively correlated with the share of the population who applied for UI benefits (see chart 6).
Although this evidence does not support the crowding hypothesis, our approach is not overly sensitive to crowding
issues early in the pandemic because the HPS data we use were collected from August 19 to December 21, 2020,
and the questions on applying for and receiving UI benefits have a reference period going back to March 13, 2020.

Given the possibility of multiple factors driving the variation in HPS state-level estimates, we now turn to a
multivariate approach that allows us to explore the effects of a given factor while controlling for one or more other
factors. In the results reported in table 3, we estimate linear regressions with a given HPS state-level estimate as
the dependent variable and different combinations of the factors as independent variables. Across regressions, a
consistent finding is that the recipiency rate before the pandemic is positively correlated with the share of the
population who applied for UI benefits, the share of the population who received benefits, and the success rate of
UI applicants. When we hold the prepandemic recipiency rate constant, the unemployment rate is positively
correlated with the share of the population who applied for UI benefits and the share of the population who
received benefits but is not correlated with the success rate. These results suggest that, among states with a
similar recipiency rate, when the unemployment rate increases, the increased number of people applying for UI
benefits can be accommodated without reducing the success rate.

16

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 3. Relationship of UI outcomes (percentage who applied for UI benefits, percentage who received UI
benefits, and success rate of UI applicants) to state factors, regression results
Independent variable

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Dependent variable: percent who applied for UI benefits
Recipiency rate (2019)

Unemployment rate

COVID-19 death rate

COVID-19 Stringency Index

R2

—
—
—
0.176[1]
(0.047)
—
—
—
[3.064]
—
—
—
— 1.685[1]
—
—
— (0.157)
—
—
— [4.260]
—
—
—
—
0.020
—
—
— (0.014)
—
—
— [1.012]
—
—
—
— 0.228[1]
—
—
— (0.050)
—
—
— [3.148]
0.22
0.70
0.04
0.30

0.073[2]
(0.030)
[1.272]
1.546[1]
(0.160)
[3.907]
—
—
—
—
—
—
0.74

0.167[1]
(0.049)
[2.904]
—
—
—
0.011
(0.013)
[0.528]
—
—
—
0.23

0.125[1]
(0.044)
[2.167]
—
—
—
—
—
—

0.102[2]
(0.045)
[1.767]
—
—
—

0.184[1]
(0.049)
[2.550]
0.40

0.203[1]
(0.049)
[2.812]
0.43

0.172[1]
(0.042)
[2.995]
—
—
—
0.013
(0.011)
[0.651]
—
—
—
0.31

0.135[1]
(0.038)
[2.354]
—
—
—
—
—
—

0.110[1]
(0.038)
[1.915]
—
—
—
0.022[2]
(0.009)
[1.111]

0.172[1]
(0.042)
[2.385]
0.47

0.193[1]
(0.041)
[2.672]
0.53

0.224[1]
(0.062)
[3.895]
—
—
—
0.019
(0.016)
[0.927]
—
—
—
0.27

0.186[1]
(0.059)
[3.239]
—
—
—
—
—
—

0.153[2]
(0.060)
[2.665]
—
—
—

0.193[1]
(0.066)
[2.666]
0.36

0.220[1]
(0.065)
[3.042]
0.41

0.020[3]
(0.011)
[1.012]

Dependent variable: percent who received UI benefits
Recipiency rate (2019)

Unemployment rate

COVID-19 death rate

COVID-19 Stringency Index
R2

—
—
—
0.184[1]
(0.041)
—
—
—
[3.193]
—
—
—
— 1.423[1]
—
—
— (0.166)
—
—
— [3.598]
—
—
—
— 0.023[3]
—
—
— (0.012)
—
—
— [1.151]
—
—
—
— 0.219[1]
—
—
— (0.045)
—
—
— [3.035]
0.29
0.60
0.07
0.33

0.102[1]
(0.030)
[1.768]
1.229[1]
(0.162)
[3.107]
—
—
—
—
—
—
0.68

Dependent variable: success rate of UI applicants
Recipiency rate (2019)

Unemployment rate

COVID-19 death rate

COVID-19 Stringency Index
R2

—
—
—
0.240[1]
(0.060)
—
—
—
[4.177]
—
—
—
— 0.774[2]
—
—
— (0.355)
—
—
— [1.957]
—
—
—
— 0.032[3]
—
—
— (0.017)
—
—
— [1.578]
—
—
—
— 0.258[1]
—
—
— (0.068)
—
—
— [3.561]
0.24
0.09
0.06
0.23

[1] Statistically significant at the 0.01 level.
[2] Statistically significant at the 0.05 level.
[3] Statistically significant at the 0.10 level.
See footnotes at end of table.

17

0.216[1]
(0.065)
[3.758]
0.362
(0.345)
[0.914]
—
—
—
—
—
—
0.26

0.029[2]
(0.015)
[1.451]

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Note: The number of observations for each regression is 51 (50 states and the District of Columbia). Standard errors are in parentheses. Numbers in brackets
are the coefficients multiplied by (p75 – p25), where p75 and p25 are the 75th and 25th percentiles of the distribution (across states) of the relevant
independent variable, respectively. UI outcomes are based on the period March 13–December 21, 2020. Percentage who applied for UI benefits, percentage
who received UI benefits, and the success rate of UI applicants are for people 18 years and older. Recipiency rate is percentage of unemployed workers who
received UI benefits, 2019. Unemployment rate is average unemployment rate from March through December 2020. COVID-19 death rate is cumulative
deaths (through December 21, 2020) per 100,000 population, on the basis of reports from state and local health agencies, and U.S. Census Bureau
population estimates for July 1, 2019. COVID-19 Stringency Index is average of daily values from March 13 through December 21, 2020. Dashes indicate no
data. COVID-19 = coronavirus disease 2019. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020; Employment and Training Administration, “Recipiency rates, by state,” Section A.13,
Unemployment Insurance Chartbook, https://oui.doleta.gov/unemploy/chartbook.asp; U.S. Bureau of Labor Statistics, “Employment status of the civilian
noninstitutional population, not seasonally adjusted, statewide data,” monthly series, https://www.bls.gov/web/laus/ststdnsadata.zip; The New York Times,
“Coronavirus (Covid-19) data in the United States,” https://github.com/nytimes/covid-19-data; U.S. Census Bureau, “Annual population estimates, estimated
components of resident population change, and rates of the components of resident population change for the United States, states, and Puerto Rico: April 1,
2010 to July 1, 2019,” https://www2.census.gov/programs-surveys/popest/datasets/2010-2019/national/totals/nst-est2019-alldata.csv; and Thomas Hale,
Noam Angrist, Rafael Goldszmidt, Beatriz Kira, Anna Petherick, Toby Phillips, Samuel Webster, Emily Cameron-Blake, Laura Hallas, Saptarshi Majumdar, and
Helen Tatlow, “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker),” Nature Human Behaviour, vol. 5, no. 4, April
2021, pp. 529–538, https://doi.org/10.1038/s41562-021-01079-8.

Instead of looking at the effect of the unemployment rate, we look at the effects of the COVID-19 measures.
Because the COVID-19 measures capture what might be causing the higher unemployment rates, we do not
include the unemployment rate in the same model with the COVID-19 measures. When we consider both
COVID-19 measures together and control for the recipiency rate, each COVID-19 measure is positively correlated
with the share of the population who applied for UI benefits, the share of the population who received benefits, and
the success rate of UI applicants. However, the magnitude of the effect is more than twice as large for the
Stringency Index than the death rate.36 This finding suggests that although both COVID-19 measures explain
variation in the state-level HPS estimates, the Stringency Index contributes more than the death rate.

Estimates by employment income, work status, and children’s school
routines
A measure of whether an HPS respondent lives in a household where a member’s labor earnings were negatively
affected during the pandemic comes from the question, “Have you, or has anyone in your household experienced
a loss of employment income since March 13, 2020?” In the HPS data for August 19 to December 21, 2020, 46.5
percent of individuals responded “yes.” Those who responded “yes” did not necessarily lose their job during the
pandemic or experience a reduction in hours at work. Respondents could have answered “yes” for several
reasons, including that someone else in their household lost employment. However, the set of respondents who
said “yes” likely includes people who experienced either job loss or a reduction in work hours during the pandemic.
Among adults in households that lost employment income, 41.2 percent applied for UI benefits and 31.8 percent
received benefits (see table 4).
Table 4. Number and percentage of people 18 years and older who applied for UI benefits and received UI
benefits, and success rate of UI applicants, by current work status and whether household lost
employment income during the pandemic
Applied for UI
Characteristic

Received UI

Total

Success rate
Number

See footnotes at end of table.

18

Percent

Number

Percent

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 4. Number and percentage of people 18 years and older who applied for UI benefits and received UI
benefits, and success rate of UI applicants, by current work status and whether household lost
employment income during the pandemic
Applied for UI
Characteristic

Success rate
Number

Total
Household lost employment income
Household did not lose employment
income
Total at work (last 7 days)
Private
Government
Self-employed
Nonprofit
Employed in family business

Received UI

Total
Percent

Number

Percent

249,170,916
115,886,322

52,430,773
47,723,344

21.0
41.2

39,984,667
36,842,757

16.0
31.8

77.2
78.1

131,875,465

4,651,902

3.5

3,105,128

2.4

67.7

143,161,639
85,114,321
20,829,780
16,846,500
13,359,379
4,168,710

26,229,773
17,716,946
1,948,237
3,634,195
1,570,867
732,885

18.3
20.8
9.4
21.6
11.8
17.6

20,074,771
14,029,861
1,353,946
2,568,220
1,132,450
522,524

14.0
16.5
6.5
15.2
8.5
12.5

77.9
80.5
71.3
72.2
73.2
72.8

Note: Success rate excludes individuals who applied for UI benefits but did not answer the question about receiving benefits. “Total at work” includes those
who did not answer the question about class of worker (private, government, etc.). UI information covers March 13–December 21, 2020. Current work status is
based on 7 days before survey date (August 19–December 21, 2020). UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

We examine current work status using the question on whether the respondent was doing any work for pay or
profit in the last 7 days. For this analysis, the “not working” category includes those who are not working for a
variety of reasons, including stay-at-home parents and those who are retired. To focus the analysis on individuals
for which usual employment rates are relatively high, we restrict the sample for this part of our analysis to primeworking-age (25–54) people. For those not working, we classify the main reasons for not working into three
categories: COVID-19-related reasons that are employer initiated, COVID-19-related reasons that are not
employer initiated, and non-COVID-19-related reasons (see appendix table 2 for a fuller explanation of this
categorization).
Among prime-working-age persons who received UI benefits at some point during the pandemic, 51.3 percent
were at work in the last 7 days, 30.9 percent were not working in the last 7 days because of a COVID-19-related
reason initiated by their employer, and 4.4 percent were not working because of a COVID-19-related reason not
associated with their employer (see table 5). By comparison, among all prime-working-age individuals regardless
of whether UI benefits were received, 71.4 percent were at work. Thus, although more than half of UI recipients
had either returned to work or never lost their jobs (e.g., had their hours reduced), the extent of job loss among UI
recipients (on the basis of employment status at the time they were surveyed) was still substantial. Among primeworking-age individuals not working in the last 7 days because of a COVID-19-related reason initiated by the
employer, 61.3 percent had received UI benefits. By contrast, only 26.6 percent of individuals not working because
of other COVID-19-related reasons had received benefits.

19

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 5. Current work status of those who were UI applicants and recipients, ages 25–54
Number
Work status

UI

UI

applicants

recipients

130,234,105

33,195,736

25,316,195 100.0

92,006,883

17,032,532

13,115,441

37,466,674

16,128,139

12,344,361

All
Total
At work (last 7
days)
Not working
COVID-19-related
reason, employer
initiated
COVID-19-related
reason, not
employer initiated
Non-COVID-19related reason

Percent within column
All

UI

Percent within row

UI

applicants recipients

All

UI

UI

applicants recipients

100.0

100.0 100.0

25.5

19.4

71.4

50.8

51.3 100.0

18.5

14.3

12,179,149

28.0

49.0

48.6 100.0

43.0

32.5

9,347,643

7,563,091

9.3

28.9

30.9 100.0

75.7

61.3

4,531,103

1,832,296

1,203,011

2.9

5.1

4.4 100.0

40.4

26.6

19,693,323

4,881,467

3,368,860

15.1

14.8

13.1 100.0

24.8

17.1

Note: For classification of reasons for not working, see appendix table 2. UI information covers March 13–December 21, 2020. Current work status is based
on 7 days before survey date (August 19–December 21, 2020). COVID-19 = coronavirus disease 2019. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

In the rest of this section, we look at people of all ages. Among those who were working in the last 7 days, 18.3
percent had applied for UI benefits and 14.0 percent had received benefits since March 13, 2020 (see table 4).
There is wide variation by class of worker in the share of workers who had received UI benefits. This share is 6.5
percent among workers in government, 8.5 percent among workers in nonprofit organizations, 12.5 percent among
those employed in a family business, 15.2 percent among the self-employed, and 16.5 percent among wage and
salary workers in the private sector. One factor in this pattern is the success rate among UI applicants, which is
higher among wage and salary workers in the private sector than among workers in the other classes.
The variation by class of worker in the share of workers who received UI benefits may also be related to
differences by class of worker in the extent of employment disruption during the pandemic. Employment estimates
from the Current Population Survey are broadly consistent with this explanation: employment losses during the first
9 months of the pandemic were greater for wage and salary workers in the private sector than for workers in
government and the self-employed. From February to April 2020, employment fell 17.9 percent for wage and
salary workers in the private sector but only 8.8 percent for workers in government and 7.8 percent for the selfemployed. These differences persisted to some extent during the recovery: by December 2020, employment was
down 5.9 percent (compared with February 2020) for wage and salary workers in the private sector but only 3.2
percent for workers in government and 4.2 percent for the self-employed.37 Consistent with the pattern of
employment losses, the share of wage and salary workers in the private sector who applied for UI benefits was
more than double the share of workers in government who applied. For the self-employed, the share of people
who applied for UI benefits was higher than is suggested by the employment losses. This finding might reflect that
many of the self-employed continued to work during the pandemic but experienced reduced income from
employment.

20

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Our analysis of UI application and receipt by class of worker provides a perspective on the effect of UI expansions
during the pandemic. The PUA program, created under the CARES Act, expanded eligibility for UI during the
pandemic to self-employed workers, independent contractors, and part-time workers.38 Among those who were
working in the last 7 days and were self-employed, 21.6 percent had applied for UI benefits and 15.2 percent had
received benefits (see table 4). These rates are similar to those for wage and salary workers in the private sector
who worked in the last 7 days. However, the success rate was much lower for self-employed workers (72.2
percent) than for private wage and salary workers (80.5 percent). This finding suggests that the PUA program
expanded UI coverage for self-employed workers during the pandemic, but a gap in coverage remained, compared
with private wage and salary workers.
With another policy change that broadened eligibility for UI benefits during the pandemic, the FFCRA allowed
states the flexibility to expand the range of “good cause” exceptions for leaving work to include caring for a child,
parent, or spouse/partner.39 This change is relevant for parents of school-age children who contended with the
shift to remote learning during the pandemic. With children at home instead of in school buildings during the day,
some parents had to stop working or reduce their hours of work to supervise and support their children with remote
learning and care.40 From the perspective of the UI system, it is of interest whether the changes to school routines
affected parents’ use of UI benefits during the pandemic. For households with any children enrolled in a public or
private school in kindergarten through 12th grade, the HPS asked, “How has the coronavirus pandemic affected
how the children in this household received education for the 2020–2021 school year?”41
The responses to this question show that parents whose children experienced more disruptions to their school
routine were more likely to have applied for UI benefits and more likely to have received benefits during the
pandemic (see table 6). Among parents of students whose classes normally taught in person at the school were
canceled, 21.5 percent received benefits. Among parents of students whose classes normally taught in person at
the school were moved to a distance-learning format (using either online resources or article materials), 19.2
percent received benefits. Among parents of students who did not experience any change in their school routine
(that is, their classes were taught in person at the school), 14.1 percent received benefits.
Table 6. Number and percentage of people 18 years and older who applied for UI benefits and received UI
benefits, and success rate of UI applicants, parents with children enrolled in school (K–12), by school
routine change of children
Applied for UI
School routine change

Success rate
Number

Total
Classes canceled
Classes moved to distance
learning
Classes changed in some other
way
No change

Received UI

Total
Percent

Number

Percent

52,469,508
13,521,499

12,825,212
3,902,190

24.4
28.9

9,774,425
2,908,652

18.6
21.5

76.9
75.4

37,140,865

9,290,310

25.0

7,135,603

19.2

77.4

6,567,109

1,411,664

21.5

1,067,410

16.3

76.3

5,535,674

1,031,502

18.6

781,418

14.1

76.4

Note: Success rate excludes individuals who applied for UI benefits but did not answer the question about receiving benefits. UI information covers March 13–
December 21, 2020. School routine change is based on 2020–21 school year. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

21

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

We also look at the current work status of parents of children enrolled in school (see table 7). Parents whose
children experienced more disruptions to their school routine were less likely to be at work in the last 7 days.
Among parents of students whose classes normally taught in person at the school were canceled, 63.0 percent
were at work. By contrast, among parents of students who did not experience any change in their school routine,
72.6 percent were at work. Furthermore, among parents who were not working, those whose children experienced
more disruptions to their school routine were more likely to choose “I am/was caring for children not in school or
daycare” as their main reason for not working.
Several possible reasons exist for the correlations among disruption to school routines, parental employment, and
parents receiving UI benefits. One reason is that changes to school routines caused parents to leave their jobs and
apply for UI benefits. Another is that a third factor (the coronavirus or responses to it at the local level) caused both
school disruptions and parental job losses. Our analysis does not allow us to distinguish between these two
explanations.
Table 7. Number and percentage of parents with children enrolled in school (K–12), by current work status
and school routine change of children
Reason for not working =
At work
School routine change

caring for children not in
school or daycare

Number
Total
Classes canceled
Classes moved to distance learning
Classes changed in some other way
No change

Not working

Total

52,469,508 35,549,640
13,521,499 8,523,813
37,140,865 25,179,183
6,567,109 4,575,905
5,535,674 4,017,586

Percent

Number

67.8 16,857,027
63.0 4,973,892
67.8 11,931,002
69.7 1,985,016
72.6 1,509,538

Percent
32.1
36.8
32.1
30.2
27.3

Number

Percent

3,349,141
1,092,927
2,534,912
377,821
224,462

6.4
8.1
6.8
5.8
4.1

Note: Current work status is based on 7 days before survey date (August 19–December 21, 2020). School routine change based on 2020–2021 school year.
Source: Household Pulse Survey, August 19–December 21, 2020.

Effect of receiving UI benefits on well-being
What is the impact of UI benefits on the well-being of recipients and their households? One of the fundamental
goals of the UI system is to provide a source of income to workers during periods of unemployment. Research
using prepandemic data shows that UI benefits support unemployed workers by replacing lost income, increasing
spending, and increasing food consumption.42 Furthermore, more generous UI benefits increase health insurance
coverage and use, and reduce mortgage defaults.43 Before the pandemic, some adults were so financially
vulnerable that they would have struggled to pay for an emergency expense as small as $400 or cover expenses if
they lost their job.44 During the pandemic, the effects of receiving UI benefits on well-being might be stronger
because of the federal supplement of $600 a week. During the period when the supplement was available, Peter
Ganong and colleagues found that spending of the unemployed after job loss rose substantially above
prepandemic levels. Spending of unemployed households also rose compared with the spending of employed
households, which is opposite of the normal pattern.45

22

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

We use the HPS data to construct six measures of the well-being for individuals and households. Each measure is
an indicator for whether the individual or household experienced distress in a particular domain (see appendix
table 4 for details):
1. In the last 7 days, the household had difficulty paying for usual household expenses.
2. In the last 7 days, the household had experienced food insecurity.
3. The household is not current on mortgage or rent payments.
4. The household is not confident in being able to pay its next mortgage or rent payment.
5. Over the last 7 days, the individual experienced symptoms of anxiety.
6. Over the last 7 days, the individual experienced symptoms of depression.
The anxiety and depression measures are designed to match concepts in surveys sponsored by the National
Center for Health Statistics.46
Prior research that used the HPS and other data indicates that in the population overall, levels of distress during
the pandemic were much higher than before the pandemic. Using data from the COVID Impact Survey and
comparable data from the National Health Interview Survey (NHIS), Marianne P. Bitler, Hilary W. Hoynes, and
Diane Whitmore Schanzenbach estimated that food insecurity increased sharply from 11 percent in 2018 to 23
percent in April 2020.47 HPS data for May 14–19, 2020, indicate that 28.2 percent of adults had symptoms of
anxiety disorder and 24.4 percent had symptoms of depressive disorder.48 These estimates are 3 to 4 times larger
than comparable estimates for January–June 2019 from the NHIS, which indicated 8.2 percent of adults had
symptoms of anxiety disorder and 6.6 percent had symptoms of depressive disorder.
To gauge the effect of receiving UI benefits on well-being, we compare those who had received benefits (at some
point during the pandemic) with those who applied for benefits but did not receive them.49 In our HPS data for
August 19 to December 21, 2020, the well-being measures show substantially less distress among UI recipients
than among unsuccessful applicants (see table 8 and chart 7). For instance, 18.9 percent of UI recipients were
experiencing food insecurity, compared with 29.1 percent of unsuccessful applicants. In addition, 45.2 percent of
UI recipients were experiencing anxiety symptoms, compared with 53.2 percent of unsuccessful applicants. The
well-being measures for UI applicants (both successful and unsuccessful) are substantially higher (indicating more
distress) than for the general population, consistent with the higher incidence of job loss among UI applicants.50
Table 8. Household and individual well-being of UI applicants, by whether they received UI benefits (in
percent)
Applied for UI benefits
Characteristic

All

Difference
Received

Well-being measure
Having difficulty with household expenses
Experiencing food insecurity
Not current on mortgage or rent
Not confident on upcoming mortgage or rent payment
Having symptoms of anxiety
Having symptoms of depression

33.7
11.4
12.7
18.7
34.1
26.7

See footnotes at end of table.

23

55.9
18.9
20.1
31.7
45.2
36.7

Not received

66.6
29.1
27.4
40.4
53.2
45.5

–10.7
–10.2
–7.3
–8.7
–8.0
–8.8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 8. Household and individual well-being of UI applicants, by whether they received UI benefits (in
percent)
Applied for UI benefits
Characteristic

All

Difference
Received

Household income (2019)
No answer
Less than $25,000
$25,000 to $34,999
$35,000 to $49,999
$50,000 to $74,999
$75,000 to 99,999
$100,000 to $149,999
$150,000 to $199,999
$200,000 or more
Education (age 18+)
Less than high school
Some high school
High school graduate or equivalent
Some college, but degree not received
Associate’s degree
Bachelor’s degree
Graduate degree

Not received

28.7
10.1
7.9
9.0
12.8
9.8
11.0
5.1
5.6

28.0
12.3
10.2
10.8
14.1
9.5
9.1
3.4
2.6

31.6
17.4
11.5
10.1
11.6
6.8
6.2
2.6
2.1

–3.6
–5.1
–1.2
0.7
2.5
2.7
2.8
0.8
0.5

2.5
5.7
30.9
21.1
9.4
17.1
13.2

2.7
6.3
34.6
24.6
10.5
14.8
6.5

4.1
7.8
35.8
23.9
9.8
12.3
6.3

–1.4
–1.6
–1.3
0.8
0.7
2.5
0.2

Note: UI information covers March 13–December 21, 2020. Difference = received minus not received. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

24

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The difference in well-being between UI recipients and unsuccessful applicants may not necessarily be the causal
effect of receiving UI benefits on well-being. Prepandemic differences could exist between those who received UI
benefits and unsuccessful applicants. For instance, differences in prepandemic earnings patterns may be
associated with eligibility for UI during the pandemic. As a result, those who received UI benefits may have had
higher incomes and higher savings before the pandemic than did unsuccessful applicants. As such, they may have
been in a better position to weather losing their job, apart from the UI benefits they received. We find that UI
recipients had higher household incomes in 2019 and more education than unsuccessful applicants (see table 8).
To assess the causal effect of receiving UI benefits on well-being, we estimate linear regressions with controls for
variables that might differ between UI recipients and unsuccessful applicants. In each regression, the dependent
variable is one of the well-being measures and the key independent variable is an indicator for receiving UI
benefits. The sample is limited to those who had applied for benefits, so the estimated coefficient on the indicator
for having received benefits distinguishes between UI recipients and unsuccessful applicants. Without any
controls, the estimated coefficients in column 1 of table 9 match the differences between groups in table 8. As
controls for household income (in 2019), education, and demographics (gender, race, and ethnicity) are added, the
estimated coefficients on the indicator fall in magnitude—indicating that some of the differences in well-being
between the groups in table 8 are due to prepandemic differences rather than receipt of UI benefits.51

25

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Controlling for household income, education, and demographics reduces the differences between groups by 15
percent to 30 percent, depending on the well-being measure. However, when we control for prepandemic
differences in these factors, the differences in well-being between UI recipients and unsuccessful applicants are
still large. This finding suggests that receiving UI benefits during the pandemic substantially improved the wellbeing (reduced the distress) of individuals and households. Using our preferred estimates in table 9 (column 4),
one can see that receiving UI benefits reduces the probability of having difficulty with household expenses by 7.8
percentage points and reduces the probability of experiencing food insecurity by 7.5 percentage points. For the
housing measures, receiving UI benefits reduces the probability of an applicant not being current on mortgage or
rent by 5.6 percentage points and reduces the probability of not being confident on paying the upcoming mortgage
or rent payment by 5.8 percentage points. For the mental health measures, receiving UI benefits reduces the
probability of an applicant having symptoms of anxiety by 7.2 percentage points and reduces the probability of
having symptoms of depression by 7.5 percentage points. These effects are large in relation to the mean among
unsuccessful applicants (see table 9).
Table 9. Effect on well-being of applicants receiving UI benefits
Effect of receiving UI benefits
Well-being measure and controls

Mean
(1)

Controls
Household income (2019)
Education
Demographics (gender, race, ethnicity)
Well-being measure

(2)

(3)

N

(4)

—
—
—

X
—
—

X
X
—

X
X
X

—
—
—

—
—
—

Having difficulty with household expenses

–0 .107[1]
(0.003)

–0 .084[1]
(0.003)

–0 .083[1]
(0.003)

–0 .078[1]
(0.003)

Experiencing food insecurity

–0.102[1]
(0.003)

–0.081[1]
(0.003)

–0.079[1]
(0.003)

–0.075[1]
(0.003)

Not current on mortgage or rent

–0.073[1]
(0.003)

–0.060[1]
(0.003)

–0.059[1]
(0.003)

–0.056[1]
(0.003)

Not confident on upcoming mortgage or rent
payment

–0.087[1]
(0.004)

–0.063[1]
(0.004)

–0.061[1]
(0.004)

–0.058[1]
(0.004)

Having symptoms of anxiety

–0.080[1]
(0.004)

–0.072[1]
(0.004)

–0.072[1]
(0.004)

–0.072[1]
(0.004)

Having symptoms of depression

–0.088[1]
(0.004)

–0.075[1]
(0.004)

–0.075[1]
(0.004)

–0.075[1]
(0.004)

0.666
—
0.291
—
0.274
—
0.404
—
0.532
—
0.455
—

126,553
—
117,535
—
86,951
—
86,779
—
109,106
—
109,014
—

[1] Statistically significant at the 0.01 level.
[2] Statistically significant at the 0.05 level.
[3] Statistically significant at the 0.10 level.

Note: UI information covers March 13–December 21, 2020. Standard errors are in parentheses. Each estimated effect comes from a linear regression with a
given well-being measure as the dependent variable; the independent variables are an indicator for successful applicants and the variables for the controls
indicated. Regressions are weighted by the Household Pulse Survey person weight. X = control included in the regression. N = number of observations used
in the regression, which is 129,901 (102,303 successful applicants and 27,598 unsuccessful applicants) minus observations not in the universe for the
particular well-being measure. Dashes indicate no data. Mean = mean among unsuccessful applicants. UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

26

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Conclusion
In this article, we explored who applied for UI benefits and who received benefits during the pandemic, whether
state factors and worker characteristics could explain differences in the success rate of UI applicants, and the
effect of receiving benefits on applicants’ well-being. We estimate that, through the first 9 months of the pandemic
(March through December 2020) in the United States, 21.0 percent of adults applied for UI benefits and 16.0
percent of adults received UI benefits. Among adults in households that lost employment income during the
pandemic, 41.2 percent applied for UI benefits and 31.8 percent received benefits. By way of comparison, 84
percent of adults received stimulus payments that were authorized as part of the CARES Act. This comparison
shows the extent to which UI benefits are more targeted than stimulus payments.
We found that demographic groups disproportionately affected by job loss during the pandemic, as shown by other
sources, were more likely to receive UI benefits. These groups include Blacks, Hispanics, and women. Although
similar to the share of men who received UI benefits, the share of women who received benefits was higher than in
prior recessions, when women were less likely than men to receive benefits. We also found that states with higher
unemployment rates during the pandemic had a larger share of their population who received UI benefits.
Overall, among those who applied for UI benefits, 77.2 percent received benefits. Although workers with less
education and lower income were more likely to apply for UI benefits, the success rate of UI applicants was lower
among those with less education and lower income. This finding could reflect that low-wage workers were less
likely to be eligible for regular UI (because of minimum earnings requirements). Success rates were similar for
women and men but varied by race. Among racial groups, Blacks had the lowest success rate and Asians the
highest, with a difference of 10 percentage points. This pattern could be due, in part, to Blacks being more likely to
work in low-wage jobs and Asians being less likely.
In addition to illustrating differences in economic hardship across demographic groups, we found that success
rates are also relevant for understanding the extent to which the expansions of the UI program during the
pandemic were successful. The PUA program expanded UI eligibility during the pandemic to cover self-employed
workers, low-wage workers, and other workers typically not covered in the regular UI program. Despite the
expanded eligibility for individuals lacking sufficient work history, workers with less education and lower income
had lower success rates than workers with more education and higher income. A similar pattern emerged for selfemployed workers: the success rate for self-employed workers was lower than for private wage and salary
workers. These patterns suggest the expansions of UI eligibility under the PUA program, while beneficial to
workers not normally covered by the regular UI program, did not eliminate gaps in coverage.
The persistence of differences in UI coverage also emerged from our cross-state analysis. Across states, the
success rate is positively correlated with the prepandemic UI recipiency rate, even when we controlled for
pandemic factors such as the unemployment rate and COVID-19 measures. Moreover, the prepandemic
recipiency rate is positively correlated with the share of the population who applied for UI benefits and the share of
the population who received benefits. Taken together, these findings imply that differences in UI recipiency across
states before the pandemic contributed to differences in UI recipiency across states during the pandemic.
Despite the persistence of state differences in UI recipiency from before the pandemic, we found that the UI
program provided benefits to individuals who were directly affected by the pandemic in the labor market. In
addition to providing income support, the UI program improved the well-being of individuals and households.

27

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Compared with individuals who applied for UI benefits but did not receive them, individuals who received UI
benefits had greater well-being in a variety of domains, including household finances, food security, and mental
health.
ACKNOWLEDGMENT: We are grateful to Elizabeth Handwerker and Mark Loewenstein for their helpful
comments. We have also benefited from the comments of seminar participants at the Department of Labor’s Chief
Evaluation Office.

Appendix: Additional documentation of the Household Pulse Survey
and results regarding unemployment insurance applicants and
recipients during the coronavirus pandemic
Appendix table 1. Household Pulse Survey sample size, number of respondents, and response rate,
August 19–December 21, 2020
PhaseWeek
2
2
2
2
2
3
3
3
3

13
14
15
16
17
18
19
20
21

Start date

End date

August 19, 2020
September 2, 2020
September 16, 2020
September 30, 2020
October 14, 2020
October 28, 2020
November 11, 2020
November 25, 2020
December 9, 2020

August 31, 2020
September 14, 2020
September 28, 2020
October 12, 2020
October 26, 2020
November 9, 2020
November 23, 2020
December 7, 2020
December 21, 2020

Sample
size

Number of respondents

1,032,959
1,033,494
1,034,047
1,034,605
1,035,186
1,035,752
1,036,354
1,036,968
1,037,606

Response rate (percent)

109,051
110,019
99,302
95,604
88,716
58,729
71,939
72,484
69,944

10.3
10.3
9.2
8.8
8.1
5.3
6.6
6.7
6.5

Source: U.S. Census Bureau, “Source of the data and accuracy of the estimates for the 2020 Household Pulse Survey—phase 2,” week 17, October 14–26,
2020; and phase 3, week 21, December 9–21, 2020.

Appendix table 2. Classification of reasons for not working
COVID-19-related reason: employer initiated
• My employer experienced a reduction in business (including furlough) due to coronavirus pandemic.
• I am/was laid off due to coronavirus pandemic.
• My employer closed temporarily due to the coronavirus pandemic.
• My employer went out of business due to the coronavirus pandemic.
COVID-19-related reason: not employer initiated
• I am/was sick with coronavirus symptoms.
• I am/was caring for someone with coronavirus symptoms.

28

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

• I was concerned about getting or spreading the coronavirus.
Non-COVID-19-related reason
• I did not want to be employed at this time.
• I am/was caring for children not in school or daycare.
• I am/was caring for an elderly person.
• I am/was sick (not coronavirus related) or disabled.
• I am retired.
• Other reason, please specify.
Note: We considered classifying the reason “I am/was caring for children not in school or daycare” as
COVID-19-related because parents of children enrolled in school may have had to stop working because
their children were learning from home during the pandemic and required supervision during the school day.
However, this reason could also be used by parents of children who are younger than school age, including
parents who choose to stay at home for reasons unrelated to COVID-19. As a result, we classified the
reason as non-COVID-19-related. COVID-19 = coronavirus disease 2019.
Source: Household Pulse Survey, August 19–December 21, 2020.

Appendix table 3. Percentage of people 18 years and older who applied for UI benefits and received UI
benefits, and success rate of UI applicants, by state, March 13–December 21, 2020
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana

Percent applied for UI benefits

Percent received UI benefits
17.33
21.13
17.93
16.73
25.26
17.30
21.80
19.37
18.04
18.84
21.09
27.93
14.47
19.03
19.02
17.97
14.30
21.86
23.50

See footnotes at end of table.

29

Success rate
11.37
15.94
12.57
10.55
20.70
13.97
17.29
13.47
12.28
13.44
15.01
21.24
10.80
14.27
14.83
13.74
10.38
15.90
17.53

66.45
75.72
70.70
63.87
82.99
81.26
80.37
70.31
69.14
72.15
72.29
77.31
75.73
75.63
78.59
77.46
73.22
73.60
75.64

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Appendix table 3. Percentage of people 18 years and older who applied for UI benefits and received UI
benefits, and success rate of UI applicants, by state, March 13–December 21, 2020
State

Percent applied for UI benefits

Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming

Percent received UI benefits
17.76
19.06
23.36
30.25
22.36
21.47
17.46
18.48
13.19
30.31
19.73
25.31
20.06
27.14
19.05
14.66
19.95
14.64
21.80
23.46
25.28
18.34
12.11
17.12
18.43
11.39
21.74
17.51
21.59
16.82
18.09
12.47

Note: UI = unemployment insurance.
Source: Household Pulse Survey, August 19–December 21, 2020.

Appendix table 4. Construction of well-being measures
Well-being measure
Having difficulty with household expenses
Response is coded as “yes” if response is “Somewhat difficult” or “Very difficult.”

See footnotes at end of table.

30

Success rate
14.53
14.71
19.52
25.51
18.04
16.56
12.75
13.75
10.01
21.71
15.65
20.57
15.19
23.29
12.85
11.13
13.71
9.76
15.63
16.96
21.72
12.86
8.33
12.83
13.44
7.63
18.08
12.89
16.23
11.68
12.55
9.70

82.94
77.79
84.51
85.23
81.57
78.11
74.05
74.80
76.20
72.31
80.34
82.25
76.75
86.97
68.79
76.96
69.66
67.38
72.37
73.07
87.08
70.89
69.11
75.78
73.73
67.42
83.79
74.78
76.11
70.03
70.05
78.66

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Appendix table 4. Construction of well-being measures
Well-being measure
Q19a: “In the LAST 7 DAYS, how difficult has it been for your
household to pay for usual household expenses, including but not
limited to food, rent or mortgage, car payments, medical expenses,
student loans, and so on?”

1. Not at all difficult
2. A little difficult
3. Somewhat difficult
4. Very difficult

Experiencing food insecurity

Response is coded as “yes” if response is “Sometimes not enough to eat” or “Often not enough to eat.”

Q24: “In the LAST 7 DAYS, which of these statements best describes
the food eaten in your household?”

1. Enough of the kinds of food
(I/we) wanted to eat
2. Enough, but not always the
kinds of food (I/we) wanted to
eat
3. Sometimes not enough to
eat
4. Often not enough to eat

Not current on mortgage or rent

Response is coded as “yes” if Q40c = “no” (when home is owned, with mortgage) or Q40b = “no” (when home is rented).

Q40c: “Is this household CURRENTLY caught up on mortgage
payments?”

1. Yes

Q40b: Is this household CURRENTLY caught up on rent payments?

1. Yes

2. No
2. No

Q39: “Is your house or apartment . . . ?”

1. Owned free and clear
2. Owned with a mortgage or
loan (including home equity
loans)
3. Rented
4. Occupied without payment
of rent

See footnotes at end of table.

31

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Appendix table 4. Construction of well-being measures
Well-being measure
Not confident on upcoming mortgage or rent payment

Response is coded as “yes” if “No confidence” or “Slight confidence” when home is owned, with mortgage, or rented.

Q41: “How confident are you that your household will be able to pay
your NEXT RENT OR MORTGAGE PAYMENT on time?”

1. No confidence
2. Slight confidence
3. Moderate confidence
4. High confidence
5. Payment is/will be deferred

Having symptoms of anxiety

Response is coded as “yes” if the sum of Q32 and Q33 is 5 or more (where the numerical value for a question is the number
associated with the response category).

Q32: “Over the LAST 7 DAYS, how often have you been bothered by
the following problems . . . Feeling nervous, anxious, or on edge?
Would you say not at all, several days, more than half the days, or
nearly every day?”

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day

Q33: “Over the LAST 7 DAYS, how often have you been bothered by
the following problems . . . Not being able to stop or control worrying?
Would you say not at all, several days, more than half the days, or
nearly every day?”

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day

Having symptoms of depression

Response is coded as “yes” if the sum of Q34 and Q35 is 5 or more (where the numerical value for a question is the number
associated with the response category).

Q34: “Over the LAST 7 DAYS, how often have you been bothered
by . . . Having little interest or pleasure in doing things? Would you
say not at all, several days, more than half the days, or nearly every
day?”

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day

Q35: “Over the LAST 7 DAYS, how often have you been bothered
by . . . Feeling down, depressed, or hopeless? Would you say not at
all, several days, more than half the days, or nearly every day?”

1. Not at all
2. Several days
3. More than half the days
4. Nearly every day

Source: Household Pulse Survey, August 19–December 21, 2020.

SUGGESTED CITATION

32

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Patrick Carey, Jeffrey A. Groen, Bradley A. Jensen, Anne E. Polivka, and Thomas J. Krolik, "Applying for and
receiving unemployment insurance benefits during the coronavirus pandemic," Monthly Labor Review, U.S.
Bureau of Labor Statistics, September 2021, https://doi.org/10.21916/mlr.2021.19.
NOTES
1 Elizabeth Weber Handwerker, Peter B. Meyer, Joseph Piacentini, Michael Schultz, and Leo Sveikauskas, “Employment recovery in
the wake of the COVID-19 pandemic,” Monthly Labor Review, December 2020, https://doi.org/10.21916/mlr.2020.27.
2 The Employment Situation—December 2020, USDL-21-0002 (U.S. Bureau of Labor Statistics, January 8, 2021), https://
www.bls.gov/news.release/archives/empsit_01082021.pdf.
3 The unemployment insurance (UI) system is a federal–state partnership that is funded by federal and state taxes on employers and
administered by states. States set eligibility, duration, and benefit levels within federal guidelines.
4 Families First Coronavirus Response Act, Pub. L. No. 116–127 (March 2020).
5 Maurice Emsellem and Michele Evermore, “Understanding the unemployment provisions of the Families First Coronavirus
Response Act,” policy brief (National Employment Law Project, March 2020), https://www.nelp.org/publication/understanding-theunemployment-provisions-of-the-families-first-coronavirus-response-act/.
6 “COVID-19: urgent actions needed to better ensure an effective federal response,” GAO-21-191 (Government Accountability Office,
November 30, 2020), https://www.gao.gov/products/gao-21-191.
7 “The effects of pandemic-related legislation on output” (Congressional Budget Office, September 18, 2020), https://www.cbo.gov/
publication/56537; and Krista Ruffini and Abigail Wozniak, “Supporting workers and families in the pandemic recession: results in
2020 and suggestions for 2021,” Brookings Papers on Economic Activity, Spring 2021, BPEA conference draft, March 25, 2021.
8 The Employment and Training Administration (ETA) publishes these data reports, which are based on summary information reported
by states, and accompanying documentation. For more information, see https://oui.doleta.gov/unemploy/DataDownloads.asp. These
ETA reports contain information by UI program (e.g., regular UI, extended benefits, and pandemic-related programs).
9 Alex Bell, Thomas J. Hedin, Geoffrey Schnorr, and Till von Wachter, “An analysis of unemployment insurance claims in California
during the COVID-19 pandemic,” policy brief (California Policy Lab, November 19, 2020), https://www.capolicylab.org/wp-content/
uploads/2020/11/Nov-19th-Analysis-of-CA-UI-Claims-During-the-COVID-19-Pandemic.pdf; Tomaz Cajner, Andrew Figura, Brendan M.
Price, David Ratner, and Alison Weingarden, “Reconciling unemployment claims with job losses in the first months of the COVID-19
crisis,” Finance and Economics Discussion Series 2020-055 (Washington, DC: Board of Governors of the Federal Reserve System,
July 13, 2020); and “COVID-19: urgent actions needed to better ensure an effective federal response,” GAO-21-191.
10 With access to individual-level microdata of the UI claims and benefits records, one could uniquely identify individuals and avoid
double counting. However, these microdata are not widely available to researchers. The California Policy Lab has used this approach
to analyze UI claims in California during the pandemic. See Bell et al., “An analysis of unemployment insurance claims in California
during the COVID-19 pandemic.”
11 In addition, reports of fraud also occurred in applications for UI during the pandemic, which could lead to inflation in the number of
claims. See “COVID-19: urgent actions needed to better ensure an effective federal response,” GAO-21-191.
12 The U.S. Census Bureau conducted an initial investigation into the potential for nonresponse bias by using 2020 Household Pulse
Survey (HPS) data and American Community Survey estimates. See Sandra Peterson, Norilsa Toribio, James Farber, and David

33

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Hornick, “Nonresponse bias report for the 2020 Household Pulse Survey” (Demographic Statistical Methods Division), Sample Design
and Estimation, version 1.0 (U.S. Census Bureau, March 24, 2021).
13 “Source of the data and accuracy of the estimates for the 2020 Household Pulse Survey—Phase 2,” week 17 (U.S. Census
Bureau, October 14–26, 2020), https://www2.census.gov/programs-surveys/demo/technical-documentation/hhp/
Phase2_Source_and_Accuracy_Week_17.pdf. For consistency, we use the person weights even when tabulating responses to
questions about the household.
14 Although weighting addresses some of the issues surrounding the potential of respondents being nonrepresentative, weighting, in
of itself, cannot solve all the issues.
15 The U.S. Bureau of Labor Statistics (BLS) was also involved in designing and testing HPS questions that focus on household
spending. See Thesia I. Garner, Adam Safir, and Jake Schild, “Receipt and use of stimulus payments in the time of the Covid-19
pandemic,” Beyond the Numbers: Prices and Spending, August 2020, https://www.bls.gov/opub/btn/volume-9/receipt-and-use-ofstimulus-payments-in-the-time-of-the-covid-19-pandemic.htm; and Thesia I. Garner, Adam Safir, and Jake Schild, “Changes in
consumer behaviors and financial well-being during the coronavirus pandemic: results from the U.S. Household Pulse Survey,”
Monthly Labor Review, December 2020, https://doi.org/10.21916/mlr.2020.26.
16 “Timeline of events related to the COVID-19 pandemic,” January 3, 2020, to June 29, 2021 (Fraser, Federal Reserve Bank of St.
Louis), https://fraser.stlouisfed.org/timeline/covid-19-pandemic.
17 We did not use data for this question because of issues with weighting.
18 In computing rates of people applying for UI and receiving UI, we use the population because it is the most appropriate
denominator available in the HPS. As an alternative denominator, we considered measures of the prepandemic labor force (employed
and unemployed). This denominator could not be constructed from the HPS because the survey does not contain any questions
about employment activities or job search before the pandemic. Estimates of the prepandemic labor force are available from the
Current Population Survey (CPS), and we considered using estimates for February 2020. However, we were concerned that some
people who applied for UI benefits during the pandemic were not in the labor force in February. This result could arise from the
expanded eligibility for UI benefits during the pandemic, in addition to normal movement in and out of the labor force. We also were
concerned about other measurement differences between the HPS and the CPS.
19 Garner et al., “Receipt and use of stimulus payments in the time of the Covid-19 pandemic.” Stimulus payments authorized by the
CARES Act were made to adults whose income was less than $99,000 (or $198,000 for joint filers). Garner et al.’s estimate is based
on the HPS data for June 11–16, 2020 (week 7). The estimate reflects both those people who had received a stimulus payment by the
survey date and those who had not received a stimulus payment but expected to receive one.
20 Estimates from a 2018 supplement to the CPS, conducted in May and September 2018, provide a point of comparison for the
success rate of UI applicants before the pandemic. Among unemployed people age 16 and older who worked in the 12 months before
the supplement and applied for UI benefits since their last job, 65.8 percent received benefits. See Characteristics of Unemployment
Insurance Applicants and Benefit Recipients—2018, USDL-19-1692 (U.S. Bureau of Labor Statistics, September 25, 2019), https://
www.bls.gov/news.release/uisup.nr0.htm.
21 Heidi Hartmann, Ashley English, and Jeffrey Hayes, “Women and men’s employment and unemployment in the Great Recession,”
briefing paper, IWPR Publication C373 (Washington, DC: Institute for Women’s Policy Research, February 2010).
22 Titan Alon, Matthias Doepke, Jane Olmstead-Rumsey, and Michèle Tertilt, “This time it’s different: the role of women’s employment
in a pandemic recession,” IZA Discussion Paper 13562 (Bonn, Germany: Institute of Labor Economics, August 2020); and Matthew
Dey, Mark A. Loewenstein, David S. Piccone Jr., and Anne E. Polivka, “Demographics, earnings, and family characteristics of workers
in sectors initially affected by COVID-19 shutdowns,” Monthly Labor Review, June 2020, https://doi.org/10.21916/mlr.2020.11.

34

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

23 Robert Fairlie, “The impact of COVID‐19 on small business owners: evidence from the first three months after widespread social‐
distancing restrictions,” Journal of Economics and Management Strategy, vol. 29, no. 4, Winter 2020, pp. 727–740; David Dam,
Meghana Gaur, Fatih Karahan, Laura Pilossoph, and Will Schirmer, “Black and White differences in the labor market recovery from
COVID-19,” Liberty Street Economics (Federal Reserve Bank of New York, February 9, 2021); and Brad J. Hershbein and Harry J.
Holzer, “The COVID-19 pandemic’s evolving impacts on the labor market: who’s been hurt and what we should do,” Working Paper
21-341 (Kalamazoo, MI: W. E. Upjohn Institute, February 11, 2021).
24 The question about the 2019 household income had a relatively high rate of nonresponse. Overall, 28 percent of HPS respondents
did not answer the question about household income.
25 Dey et al., “Demographics, earnings, and family characteristics of workers in sectors initially affected by COVID-19 shutdowns.”
Using 2019 data, Dey et al. show that workers with lower education and lower family income were disproportionately represented in
the industry sectors most susceptible to employment losses during the initial stage of the pandemic. These sectors were not
considered essential and provided goods and services requiring considerable interaction between workers and customers. See also
Michael Dalton, Jeffrey A. Groen, Mark A. Loewenstein, David S. Piccone Jr., and Anne E. Polivka, “The K-shaped recovery:
examining the diverging fortunes of workers in the recovery from the COVID-19 pandemic using business and household survey
microdata,” The Journal of Economic Inequality, 2021, https://doi.org/10.1007/s10888-021-09506-6; and Hershbein and Holzer, “The
COVID-19 pandemic’s evolving impacts on the labor market.”
26 Guido Matias Cortes and Eliza C. Forsythe, “Impacts of the Covid-19 pandemic and the CARES Act on earnings and inequality,”
unpublished paper, University of Illinois at Urbana-Champaign, May 12, 2021), http://publish.illinois.edu/elizaforsythe/files/2021/05/
Cortes_Forsythe_Inequality_May2021.pdf. Cortes and Forsythe find that with the federal supplement, UI benefits were successful at
protecting the income of low-wage workers from the greater extent of job loss. However, after the federal supplement expired at the
end of July 2020, the income of low-wage workers (earnings and UI benefits combined) fell sharply.
27 Peter Ganong, Pascal J. Noel, and Joseph S. Vavra, “US unemployment insurance replacement rates during the pandemic,”
Journal of Public Economics, vol. 191, no. 104273, November 2020, https://doi.org/10.1016/j.jpubeco.2020.104273. Ganong et al.
estimate that with the federal supplement, about 76 percent of workers eligible for regular UI had replacement rates above 100
percent—that is, they were eligible for UI benefits that exceeded their lost wages. Without the supplement, typical replacement rates
would be around 45 percent to 55 percent.
28 According to Government Accountability Office (GAO), low levels of UI receipt among unemployed low-wage workers may be
explained, in part, by state UI eligibility rules, including (1) the base period for meeting the minimum earnings requirement often
excludes the latest calendar quarter and (2) family obligations may not be considered “good cause” for leaving employment. For more
information, see “Unemployment insurance: low-wage and part-time workers continue to experience low rates of receipt,”
GAO-07-1147 (Government Accountability Office, September 2007).
29 Roxanna Edwards and Sean M. Smith, “Job market remains tight in 2019, as the unemployment rate falls to its lowest level since
1969,” Monthly Labor Review, April 2020, https://doi.org/10.21916/mlr.2020.8.
30 Christopher J. O’Leary and Stephen A. Wandner, “An illustrated case for unemployment insurance reform,” Working Paper 19-317
(Kalamazoo, MI: W. E. Upjohn Institute, January 22, 2020). We use the state recipiency rate constructed by ETA (“Recipiency rates,
by state,” Section A.13, Unemployment Insurance Chartbook, https://oui.doleta.gov/unemploy/chartbook.asp). The numerator for the
2019 recipiency rate is the average weekly number of people who received UI benefits in 2019 through the regular UI program,
according to UI administrative data. The denominator is the average monthly unemployment level for 2019, based on state
unemployment estimates from BLS (Local Area Unemployment Statistics).
31 In “How low can we go? State unemployment insurance programs exclude record numbers of jobless workers,” Briefing Paper 392
(Washington, DC: Economic Policy Institute, March 9, 2015), Will Kimball and Rick McHugh find that states that cut the maximum
duration of UI benefit recipiency saw their recipiency rates decline relative to other states. In addition, in “Low benefit recipiency in

35

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

state unemployment insurance programs,” Report for the Office of Workforce Security, Employment and Training Administration, U.S.
Department of Labor (The Urban Institute, June 2001), Wayne Vroman finds that states with high recipiency rates made more
accommodations for non-English speakers in filing for UI benefits and had requirements for monetary eligibility that were easier to
satisfy. Labor market variables that were related to state recipiency rates include the unionization rate and the job loser share of new
unemployment spells.
32 This relationship is consistent with Eliza Forsythe’s findings from her draft paper “Understanding unemployment insurance
recipiency during the Covid-19 pandemic” (University of Illinois at Urbana-Champaign, March 1, 2021). Forsythe finds that, during the
pandemic, the share of unemployed individuals who reported receiving UI benefits in the last 14 days is positively correlated with the
2019 UI recipiency rate in the individuals’ state of residence. She uses data from the Understanding America Study from March 2020
through February 2021.
33 “How the government measures unemployment,” technical documentation (U.S. Bureau of Labor Statistics, June 2014), https://
www.bls.gov/cps/cps_htgm.pdf; and “Did you know official unemployment estimates are NOT from unemployment insurance counts?”
Commissioner’s Corner (U.S. Bureau of Labor Statistics, October 17, 2019), https://blogs.bls.gov/blog/2019/10/17/did-you-knowofficial-unemployment-estimates-are-not-from-unemployment-insurance-counts/.
34 Before the pandemic, those who were not searching for work typically would not have been eligible for UI benefits. (Individuals on
layoff from an employer are eligible for UI benefits if they expect to be recalled, even if they are not searching for work.) However,
during the pandemic (per the Families First Coronavirus Response Act), states were allowed to relax the requirement that applicants
had to be actively seeking work.
35 Thomas Hale, Noam Angrist, Rafael Goldszmidt, Beatriz Kira, Anna Petherick, Toby Phillips, Samuel Webster, Emily CameronBlake, Laura Hallas, Saptarshi Majumdar, and Helen Tatlow, “A global panel database of pandemic policies (Oxford COVID-19
Government Response Tracker),” Nature Human Behaviour, vol. 5, no. 4, April 2021, pp. 529–538, https://doi.org/10.1038/
s41562-021-01079-8.
36 This statement is based on the standardized coefficients reported in brackets in table 3. These numbers are the coefficients
multiplied by (p75 – p25), where p75 and p25 are the 75th and 25th percentiles of the distribution (across states) of the relevant
independent variable, respectively.
37 These calculations were the authors, and they used estimates from the CPS that were not seasonally adjusted. For more available
data, see https://www.bls.gov/webapps/legacy/cpsatab8.htm and https://www.bls.gov/webapps/legacy/cpsatab9.htm.
38 Individuals covered by the Pandemic Unemployment Assistance program include “self-employed, individuals seeking part-time
employment, individuals lacking sufficient work history, or those otherwise not qualified for regular UI, extended benefits under state or
federal law, or PEUC.” See “Pandemic unemployment assistance (PUA) implementation and operating instructions,” Attachment I to
Unemployment Insurance Program Letter No. 16-20 (U.S. Department of Labor, ETA, April 5, 2020), https://wdr.doleta.gov/directives/
attach/UIPL/UIPL_16-20_Attachment_1.pdf.
39 Emsellem and Evermore, “Understanding the unemployment provisions of the Families First Coronavirus Response Act.”
40 Handwerker et al., “Employment recovery in the wake of the COVID-19 pandemic"; and Misty L. Heggeness, “Estimating the
immediate impact of the COVID-19 shock on parental attachment to the labor market and the double bind of mothers,” Review of
Economics of the Household, vol. 18, no. 4, December 2020, pp. 1053–1078.
41 In “Parents in a pandemic labor market” (Federal Reserve Bank of San Francisco, February 2021), Olivia Lofton, Nicolas PetroskyNadeau, and Lily Seitelman use HPS data from this question to estimate a monthly school disruption index at the state level.

36

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

42 Marianne Bitler and Hilary Hoynes, “The more things change, the more they stay the same? The safety net and poverty in the
Great Recession,” Journal of Labor Economics, vol. 34, no. S1, January 2016, pp. S403–S444; Chloe N. East and David Simon,
“How well insured are job losers? Efficacy of the public safety net,” Working Paper No. 28218 (Cambridge, MA: National Bureau of
Economic Research, December 2020); Peter Ganong and Pascal Noel, “Consumer spending during unemployment: positive and
normative implications,” American Economic Review, vol. 109, no. 7, July 2019, pp. 2383–2424; and Jonathan Gruber, “The
consumption smoothing benefits of unemployment insurance,” American Economic Review, vol. 87, no. 1, March 1997, pp. 192–205.
43 Elira Kuka, “Quantifying the benefits of social insurance: unemployment insurance and health,” The Review of Economics and
Statistics, vol. 102, no. 3, July 2020, pp. 490–505; and Joanne W. Hsu, David A. Matsa, and Brian T. Melzer, “Unemployment
insurance as a housing market stabilizer,” American Economic Review, vol. 108, no. 1, January 2018, pp. 49-81.
44 “Report on the economic well-being of U.S. households in 2019, featuring supplemental data from April 2020” (Washington, DC:
Board of Governors of the Federal Reserve System, May 2020), https://www.federalreserve.gov/publications/files/2019-reporteconomic-well-being-us-households-202005.pdf; and Brooke Helppie-McFall and Joanne W. Hsu, “Financial profiles of workers most
vulnerable to coronavirus-related earnings loss in the spring of 2020,” Financial Planning Review, vol. 3, no. 4, December 2020, p.
e1102, https://onlinelibrary.wiley.com/doi/10.1002/cfp2.1102.
45 Peter Ganong, Fiona Greig, Max Liebeskind, Pascal Noel, Daniel M. Sullivan, and Joseph Vavra, “Spending and job search
impacts of expanded unemployment benefits: evidence from administrative micro data,” Working Paper No. 2021-19 (Chicago, IL:
Becker Friedman Institute, February 11, 2021).
46 The anxiety measure is designed to match the concepts for the two-item Generalized Anxiety Disorder (GAD-2) scale. The
depression measure is designed to match the Centers for Disease Control and Prevention concepts for the two-item Patient Health
Questionnaire (PHQ-2) scale. The HPS questions are based on symptoms experienced over the last 7 days, rather than the typical 14
days. More details are available at https://www.cdc.gov/nchs/covid19/pulse/mental-health.htm.
47 Marianne P. Bitler, Hilary W. Hoynes, and Diane Whitmore Schanzenbach, “The social safety net in the wake of COVID-19,”
Brookings Papers on Economic Activity, Summer 2020 (COVID-19 and the economy: part one), pp. 119–145.
48 “Early release of selected mental health estimates based on data from the January–June 2019 National Health Interview Survey,”
National Health Interview Survey Early Release Program (National Center for Health Statistics, May 2020), https://www.cdc.gov/nchs/
data/nhis/earlyrelease/ERmentalhealth-508.pdf.
49 Researchers have used unsuccessful applicants as a comparison group for benefit recipients in other contexts—for example, the
Social Security Disability Insurance program in articles by John Bound, “The health and earnings of rejected disability insurance
applicants,” American Economic Review, vol. 79, no. 3, June 1989, pp. 482–503; and Till von Wachter, Jae Song, and Joyce
Manchester, “Trends in employment and earnings of allowed and rejected applicants to the Social Security Disability Insurance
Program,” American Economic Review, vol. 101, no. 7, December 2011, pp. 3308–3329. In our analysis, people who applied for UI
benefits but did not receive them is not as clean of a comparison group as in these other studies because it includes those who
applied for benefits before the survey date and started receiving benefits after the survey date. However, given the long reference
period (back to March 13, 2020), we believe this issue is minor.
50 Among all adults, 34.1 percent were experiencing anxiety symptoms and 26.7 percent were experiencing depressive symptoms.
These numbers are consistent with tabulations of HPS data for the same period reported in Anjel Vahratian, Stephen J. Blumberg,
Emily P. Terlizzi, and Jeannine S. Schiller, “Symptoms of anxiety or depressive disorder and use of mental health care among adults
during the COVID-19 pandemic—United States, August 2020 to February 2021,” Morbidity and Mortality Weekly Report, vol. 70, no.
13, April 2, 2021, pp. 490–494.
51 We control demographics to capture differences across groups in the well-being measures and as a predictor of earnings (given
the presence of missing data on household income).

37

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

RELATED CONTENT

Related Articles
Employment recovery in the wake of the COVID-19 pandemic, Monthly Labor Review, December 2020.
The cost of layoffs in Unemployment Insurance taxes, Monthly Labor Review, October 2020.
Geographic impact of COVID-19 in BLS surveys by industry, Monthly Labor Review, August 2020.

Related Subjects
Unemployment Insurance
Employment
Layoffs

Benefits
COVID-19

Survey methods
Recession

Labor force

38

Separations

Unemployment