Excess Savings During The Covid 19 Pandemic Feds Note

Summary

The Social Safety Net in the Wake of COVID-19, a paper by Marianne P. Bitler, Hilary W. Hoynes and Diane Whitmore Schanzenbach published in Brookings Papers on Economic Activity, Summer 2020. It reviews the response under the Families First Coronavirus Response Act and the CARES Act, including a $600 per week unemployment insurance supplement, a onetime payment of $1,200 per adult and $500 per dependent, higher SNAP payments and Pandemic EBT. The authors offer three explanations for continued unmet need: delayed payments, the modest size of payments outside UI, and coverage gaps. The paper reports that the official unemployment rate reached 14.7 percent in April 2020, with larger job losses for those with less education. It recommends extending emergency UI and SNAP measures, raising maximum SNAP benefits by 15 percent and adding automatic countercyclical triggers.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Full text

MARIANNE P. BITLER
University of California, Davis

HIL ARY W. HOYNES
University of California, Berkeley

DIANE WHITMORE SCHANZENBACH
Northwestern University

The Social Safety Net in the Wake
of COVID-19

ABSTRACT The COVID-19 crisis has led to spiking unemployment rates
with disproportionate impacts on low-income families. School and child-care
center closures have also meant lost free and reduced-price school meals.
Food prices have increased sharply, leading to reduced purchasing power for
families with limited income. The Families First Coronavirus Response Act and
the Coronavirus Aid, Relief, and Economic Security Act constituted a robust
response, including expansions to unemployment insurance (expansions in
eligibility and a $600 per week supplement), a onetime payment of $1,200
per adult and $500 per dependent, an increase in SNAP payments, and the
launch of the Pandemic EBT program to replace lost school meals. Despite
these efforts, real-time data show significant distress—notably, food insecurity
rates have increased almost three times over the pre-COVID-19 rates and
food pantry use has also spiked. In this paper, we explore why there is so
much unmet need despite a robust policy response. We provide evidence

Conflict of Interest Disclosure: The authors did not receive financial support from any firm
or person for this paper or from any firm or person with a financial or political interest in
this paper. Marianne P. Bitler was paid by the California Department of Public Health for
research and analysis of the WIC Program as it relates to vendors; however the analysis did
not focus on WIC participation and food spending behavior because of economic shocks,
but rather on how changes among vendors impact participant shopping behavior. Hilary W.
Hoynes is a member of the board of directors for MDRC and the California Budget and
Policy Center. They are currently not officers, directors, or board members of any organiza-
tion with an interest in this paper. No outside party had the right to review this paper before
circulation. The views expressed in this paper are those of the authors, and do not necessarily
reflect those of Northwestern University, the University of California, Berkeley, or the
University of California, Davis.

119
120 Brookings Papers on Economic Activity, Summer 2020

for three explanations: (1) timing—relief came with a substantial delay, due
to overwhelmed unemployment insurance (UI) systems and the need to
implement new programs; (2) magnitude—payments outside UI are modest;
and (3) coverage gaps—access is lower for some groups, and other groups
are statutorily excluded.

T he COVID-19 crisis has hit low-income families especially hard. As
unemployment rates have spiked overall, they have risen even higher
for those with lower levels of education, and for Black and Hispanic
individuals. Other aspects of the crisis have a disproportionate impact on
low-income families as well; for example, low-income families are more
likely to be headed by a single mother, and a higher share of women have
lost jobs than during prior recessions. Closures of schools and child-care
centers have meant that large numbers of low-income children have lost
access to free or reduced-price meals. Food prices have increased sharply
leading to a reduction in the purchasing power of families’ limited income.
Two pieces of legislation, the Families First Coronavirus Response Act
and the Coronavirus Aid, Relief, and Economic Security (CARES) Act,
include important provisions to respond to these historic job losses. Four
elements are particularly relevant in our context. First, there were substan-
tial expansions to unemployment insurance (UI): a $600 per week uni-
versal supplement, a thirteen-week extension of eligibility, and expanded
eligibility for self-employed and gig economy workers and those without
sufficient earnings for normal UI. Second, a onetime payment of $1,200 per
adult ($2,400 for a married couple) plus $500 per dependent child under
seventeen was implemented (with phaseouts for high-income families).
Third, all Supplemental Nutrition Assistance Program (SNAP) payments
were raised to the maximum benefit level, averaging a $165 increase in
monthly benefits for households receiving increases. Fourth, a new pro-
gram, Pandemic EBT (P-EBT), was launched to provide direct payments
to the millions of families whose children lost access to free and reduced-
price meals while their schools were closed.
Despite these efforts, many individuals and families are suffering. Food
insecurity rates have increased sharply over the pre-COVID-19 rates with
almost a quarter of families reporting their food “just didn’t last” and
they did not have money to buy more. Seven percent of adults reported
receiving help from a food pantry in the prior week, with Feeding America
(the national organization of food pantries) reporting a 60 percent increase
in need and many news outlets documenting long lines of individuals
BITLER, HOYNES, and SCHANZENBACH 121

waiting to obtain food assistance.1 Adverse mental health conditions
have worsened, with rates of depression and anxiety much higher than
pre-COVID-19 levels. While it will be many months before we have a
clear picture of how family incomes are changing, it is evident from the
available real-time data that there currently remains tremendous unmet
need.
Why do we see so much need and distress despite a policy response of
unprecedented magnitude? In this paper, we examine this question and pro-
vide evidence for three explanations. First, there is the timing of the response;
many relief payments, especially to low-income families, came with a sub-
stantial delay, and the income shock could not be weathered without hard-
ship (or emergency charity aid) for those who lacked savings or access to
credit. Payment delays have been driven by overwhelmed UI systems, the
need to engineer new programs, and application requirements for the most
disadvantaged families built into the delivery system. To the extent that
these are factors, we should see improvements as administrative capacity
and payments increase across time, though of course hardship may increase
once again when emergency payments are rolled back. Second, outside of
the UI system, the magnitude of payments made to low-income families
was relatively modest—averaging $30 to $40 per week—and may not have
been sufficient to offset increased need. Third, there are coverage gaps in the
response, and some who were hit by the economic shock had no recourse
from existing safety net programs. Importantly, despite expansions intended
to make UI coverage more universal than it has traditionally been, the limited
real-time data suggest that there are still many unemployed workers who
are not receiving UI.
Furthermore, and more structurally, over the past several decades the
United States has steered its social safety net, which has always been less
far-reaching and less funded compared to other rich countries, to focus
on work. Through the shift from cash assistance to earnings supplements,
and through adding work requirements to programs designed to meet basic
food and healthcare needs, the United States has built a social safety net
that delivers less insurance and has placed more emphasis on incentivizing
work and topping up low earnings. The current system may meet need

1. Feeding America, https://www.feedingamerica.org/take-action/coronavirus; “Feeding
America Network Stays Resilient during COVID-19 Crisis,” press release, May 12, 2020,
https://www.feedingamerica.org/about-us/press-room/feeding-america-network-stays-resilient-
during-covid-19-crisis.
122 Brookings Papers on Economic Activity, Summer 2020

during times of low unemployment, but it is ill-suited to protect against job
loss and high unemployment. Cash welfare payments for the nondisabled
are extremely limited and are either not countercyclical or only very
slightly so (Bitler and Hoynes 2016; Bitler, Hoynes, and Iselin 2020).
While SNAP payments typically can quickly increase in response to rising
need, the benefits are modest, and recent policy changes—tying SNAP
receipt to work for some groups and making it more difficult for immi-
grants to participate—will dampen SNAP’s countercyclical impact if not
waived. As a result, there are many who are likely falling through holes in
the safety net.
This analysis leads us to two sets of recommendations. In terms of
policies that need to be addressed now, the emergency policies expanding
UI and SNAP and replacing missed school meals should be extended and
adapted to the ongoing crisis. In addition, following the successful policies
of the 2009 stimulus, it would be advisable to increase maximum SNAP
benefits by 15 percent. Because UI and SNAP only serve a limited subset
of those in need, another round of stimulus payments may also be in order,
potentially targeted more narrowly to low-income families.
Second, there must be more structural policy changes to our work-based
social safety net that enable it to function more effectively in economic
downturns. The UI system should be updated to reach a larger share of
unemployed workers, including the self-employed and those with incon-
sistent work histories. Because the level and coverage of programs should
be expanded during recessions, we recommend building more effective
countercyclicality into these key safety net programs, with policy changes
automatically triggered by increases in the unemployment rate and shutting
off when economic recovery takes place. Federal and state data systems
should be harmonized to facilitate automation of relief payments to all
eligible recipients.

I. The COVID-19 Shock to Economic Well-Being
To begin, we deploy the available data to monitor the current, real-time
measures of household well-being, with particular attention to the dis
advantaged population.2
To understand who is at risk under COVID-19 for needing new or
increased access to the social safety net, we start by describing the extent

2. Han, Meyer, and Sullivan (2020) and Parolin, Curran, and Wimer (2020) use available
data to estimate real-time measures of poverty.
BITLER, HOYNES, and SCHANZENBACH 123

of job loss. We use the monthly Current Population Survey (CPS) to docu-
ment increases in unemployment across education groups (Blau, Koebe,
and Meyerhofer 2020; Montenovo and others 2020) pooling the data for
twenty-four months ending in June 2020, limiting the sample to age 18–64,
and estimating a model with calendar month dummies (to control for
seasonality) and month dummies for the four months beginning in March
2020. In online appendix table 1 panels B-E, we present the estimated
coefficients on the COVID-19 month dummies (March, partially treated;
April; May and June); each provides estimates for the effect of the crisis
on labor market outcomes, and net of typical seasonal patterns.3 As has
been widely discussed, the current crisis has made it difficult to measure
unemployment, and the Bureau of Labor Statistics has documented a spike
in the share recorded as having jobs but not being at work and also in
those not in the labor force but wanting work, many of whom should likely
be classified as unemployed instead.4 In light of this, in online appendix
table 1 we present five outcome measures for the estimated COVID-19
shock, each showing changes relative to February 2020 (netting out the
previous year): unemployed (column 1); unemployed or having a job and
not at work (column 2); unemployed, having a job and not at work, or not
in the labor force (column 3); has a job and not at work (column 4); and
not in the labor force (column 5).5 Our preferred measure is the most
expansive and is shown in column 3. Overall, by April 2020 there was a
14.1 percentage point increase in the share unemployed or with a job but
not at work or not in the labor force (or an 8 percentage point increase in
unemployed) and an 11.2 percentage point increase for those unemployed
or with a job but not at work. The labor market shock has been signifi
cantly greater for those with lower levels of education. The increase in
April unemployment (for our preferred measure) was 17.8 percentage
points for those with high school or less compared to 8.8 percentage points
for those with a college degree or more. Because children’s exposure to

3. The baseline comparison we suggest is to February 2020, but of course, the regression
results would be the same as long as the omitted month is not during March–June.
4. https://www.bls.gov/cps/employment-situation-covid19-faq-may-2020.pdf. The BLS
has documented that some share of those reporting they have a job but are not at work likely
are unemployed given ideal definitions of these measures and also notes similar concerns
for those not in the labor force due to COVID-19. Some who would like to have work but
are not measured as in the labor force reached record levels during the crisis, likely due to
closures, stay at home orders, and concerns about engaging in the labor market (also noted
in the BLS FAQ).
5. For completeness the table also shows estimates for has a job and not at work (column 4),
and not in the labor force (column 5).
124 Brookings Papers on Economic Activity, Summer 2020

economic shocks has been shown to have long-lasting health and eco-
nomic consequences (Hoynes and Schanzenbach 2018), we also analyze
changes in children’s exposure to the crisis as measured by changes in labor
market status for adults age 18–64 in their household.6 As shown in
online appendix table 2, children in households with a household head with
high school degree or less experienced a 10.1 percentage point increase in
the likelihood they lived with an adult who was unemployed, with a job but
not at work, or not in the labor force in April; compared to 6.9 percentage
points for children with a household head with a college degree. These
striking inequalities in the extent of the economic shock across education
groups continue through May and June 2020 and are evident for all of the
labor market measures. This result—that recessions increase unemploy-
ment more for lower education groups than higher education groups—
is a recurring feature of US business cycles (Hoynes, Miller, and Schaller
2012; Aaronson and others 2019).
Also important to the underlying context is that these economic indi
cators increased more and did so more quickly during the COVID-19
crisis, compared to the Great Recession (see online appendix figures 1a
and 1b).7 The (official) unemployment rate spiked to 14.7 percent in April
2020 and has remained above 10 percent through July during COVID-19,
while it reached 10 percent for only a single month in the Great Reces-
sion. Prices for food at home have increased quickly during COVID-19
driven in large part by the largest single-month increase in nearly forty-five
years in April.8
Next, we move beyond labor market outcomes to examine real-time
measures of family economic well-being. We start by analyzing food
insecurity, a summary measure indicating that a household does not have
reliable access to the food they need due to lack of resources. Usually,
a household’s food insecurity status is categorized based on their answers
to an eighteen-item questionnaire, ranging from how often the household
worried that their food would run out before there was money to buy more,
to whether a child in the household has gone for a day without eating

6. Note that unlike measures about own labor force participation and employment status,
these measures are not mutually exclusive, as a child living with more than one adult can live
with adults with various employment outcomes.
7. The online appendix figures differ in when the series documenting unemployment
rates and price changes in the Great Recession begins, with 1a starting at the beginning of the
Great Recession and 1b showing the run-up to the unemployment peak.
8. These price increases do not include increased time and hassle costs of obtaining food
for many families during COVID-19.
BITLER, HOYNES, and SCHANZENBACH 125

due to lack of money for food. Food insecurity rates can be thought of as
a measure of economic (lack of) well-being, and the time series pattern is
highly correlated with unemployment rates (Schanzenbach and Pitts 2020).
During the COVID-19 pandemic, surveys collecting real-time data have
not asked the entire battery of food security questions, but instead have
asked only a few questions drawn from the survey. We show estimates
from three waves of the COVID Impact Survey, which asked respondents
whether the following statement was often true, sometimes true, or never
true for their household over the past 30 days: “The food that we bought just
didn’t last, and we didn’t have money to get more.” We code a respondent
as being food insecure if they report that the statement was often or some-
times true. To compare food insecurity rates during COVID-19 to the past,
we calculate the share answering yes to the same question in the National
Health Interview Survey (NHIS). The NHIS asks the full food security
questionnaire, but we limit the analysis to responses to the single item
asking whether the respondent agrees that their food “just didn’t last.”9
Figure 1 displays trends in food insecurity rates for households over-
all and for those with children.10 For respondents overall, rates of food
insecurity increased sharply from 11 percent in 2018 (the latest available
NHIS estimate) to 23 percent in April 2020. Low-income families with
children have been hit particularly hard during this period, between the loss
of free and subsidized school meals due to school closures and particu-
larly elevated unemployment rates among women. This is reflected in even
greater elevation in food insecurity among respondents with children, from
13 percent in 2018 to 34 percent in April 2020.11 The large increase in
(seasonally adjusted official) unemployment, from 3.5 percent in February
to 14.7 percent in April—an out of sample prediction with strong linearity

9. Like the COVID Impact Survey, the NHIS also asks about experiences in the past
30 days. To make the data series comparable, we weight the NHIS at the respondent level;
the COVID Impact Survey only provides respondent-level weights. In general, in the NHIS
the share answering that their food “just didn’t last” is consistently 1.24 (overall) to 1.27
(with children) times the food insecurity rate based on the full questionnaire; see online
appendix table 3.
10. Online appendix figure 2 shows increases in food hardship measures using the
Census Bureau’s Household Pulse Survey compared with the Current Population Survey’s
Food Security Supplement. The Household Pulse Survey asks a different question from the
food security questionnaire and inquires about the past seven days. Results are qualitatively
similar.
11. Karpman, Zuckerman, and Gonzalez (2018) find that food insecurity rates are
higher in self-administered online surveys than they are in telephone or in-person interviews,
which they theorize is in part due to reduced social desirability bias, suggesting that the
self-administered versions might be more accurate descriptions of respondents’ well-being.
126 Brookings Papers on Economic Activity, Summer 2020

Figure 1. Food Insecurity Rates, 2011–2018 and during COVID-19
Percent of respondents

20 Respondents with children

10 Respondents overall

2011 2012 2013 2014 2015 2016 2017 2018 April May June
2020 2020 2020

Source: Authors’ tabulations from the National Health Interview Survey (NHIS) and the COVID
Impact Survey.
Note: The solid (dashed) line is the annual average share of respondents (respondents with children)
reporting that over a thirty-day period it was sometimes or often the case that their “food just didn’t last”
and that they didn’t have money to get more, calculated from the NHIS 2011–2018. The three connected
round (square) dots are share of respondents (respondents with children) reporting monthly, calculated
from the COVID Impact Survey collected April 20–26, May 4–10, and May 30–June 8. Statistics are
respondent-weighted.

assumptions to be sure—explains more than half of the increase in food
insecurity. Some of the remaining unexplained increase in food insecurity
may be due to the sharp increase in food prices (online appendix figure 1)
or loss of free or reduced-price school meals due to school closures.12 Food
insecurity rates remain elevated but have come down somewhat from their
April peak, with overall rates of 22 percent in May and 20 percent in June
(32 percent and 27 percent for respondents with children, respectively).
Other measures of real-time hardship are also elevated. Figure 2 displays
the share of households reporting receipt of emergency food from a food
bank, food pantry, or church, based on an annual time series 2002–2018
drawn from the CPS-Food Security Supplement collected each December
that asks about receipt of emergency food over the past month. The solid

12. While many schools continued to offer grab-and-go meals, according to our calcu
lations from the Census Household Pulse Survey fewer than 10 percent of households with
children report receiving “free meals through the school or other programs aimed at children.”
Ananat and Gassman-Pines (2020) find that 11 percent of low-income families reported picking
up a grab-and-go meal at their child’s school in the first weeks of school closures. Usually
58 percent of students are eligible for free or reduced-price meals at school.
BITLER, HOYNES, and SCHANZENBACH 127

Figure 2. Households Receiving Food from a Food Bank/Pantry or Church
Percent

Food bank use
4.0 last week 3.8%

3.5
Households with children
3.0 2.7%
2.5

2.0 Households overall
1.5

1.0

0.5

2002 2004 2006 2008 2010 2012 2014 2016 2018 May
2020

Source: Authors’ tabulations of CPS Food Security Supplement (CPS-FSS) and Census Pulse Survey.
Notes: The share of households (solid) or households with children (dashed) who reported using a food
bank, pantry, or church sometime in the last month from the CPS-FSS for December 2002–2018. The
square (circle) plots the share of households (households with children) who received a meal from a food
pantry, food bank, or church in the past week, based on the Census Pulse Survey pooled across April 23–
May 26. Statistics are weighted to be representative of all US households, using household weights in the
CPS-FSS and calculating pseudo-household weights in the Census Pulse Survey by dividing the
respondent weight by the number of adults in the household.

and dashed lines present trends for households overall and for those with
children. The previous peak, in 2014, showed 2.8 percent of households
receiving emergency food (3.6 percent for households with children) per
month. The point estimates for the COVID-19 period represent responses
from the Census Household Pulse Survey (averaged across months May
through July 2020), which asked respondents to report on emergency food
from these sources over the past week. Comparing across data sources,
weekly receipt of free food is at or above its previous peak monthly rate
reaching 4.3 percent of households (6.3 percent of those with children).13
In addition, measures of mental health are also being tracked in real
time during COVID-19 and show elevated rates of distress across three

13. The COVID Impact Survey also asks about receipt of food over the past seven days
from a food pantry and finds even higher estimates—6.8 percent for respondents overall and
8.3 percent among those with children, averaged across their three waves of data collected
from April to June.
128 Brookings Papers on Economic Activity, Summer 2020

categories: whether the respondent had little interest in doing things;
whether the respondent felt down, depressed, or hopeless; or whether the
respondent felt nervous, anxious, or worried. During COVID-19, the share
of adults reporting mental health problems in the past week has increased
compared with rates from 2017–2018, suggesting serious distress.14 Rates
are generally higher among those with lower levels of education, and this
gradient persists during COVID-19 (see online appendix table 4).
The Census Household Pulse Survey also asks respondents to rate
their confidence in their ability to pay for basic needs in the coming weeks.
In May, more than half of respondents indicated they are not “very confi-
dent” in their ability to pay for the food they need in the next four weeks,
with 9 percent indicating they are “not at all confident.” These rates are
uniformly higher among respondents with children and are higher among
respondents with lower levels of education (see online appendix table 5).
Among those who have a rent or mortgage payment, 43 percent overall
and 51 percent of those with children did not have “high confidence” that
they could make their next payment. Together, the evidence suggests that
households and individuals are struggling across a variety of dimensions
during COVID-19.

II. The Policy Response: How Much Money Is Going
to Whom and When?
Between the Families First Coronavirus Response Act (passed March 18)
and the CARES Act (passed March 27), more than $1 trillion have been
allocated in relief and assistance nationally. Four elements are particu-
larly important for lower-income families: expansions to SNAP, the new
P-EBT program that provides payments to compensate for missed school
meals, expansions to UI, and the onetime economic impact payments (EIP).
As we will show, these four policies account for about $600 billion and are
the main response of direct payments to households. Here we track what we
know about the magnitude of these benefits, who they went to, and the timing
of their activation.
By design, and even without congressional action, SNAP is structured
to respond quickly to increased need. Households that newly become
eligible due to unemployment or other loss of income can apply for SNAP
and generally receive benefits within thirty days. Indeed, across states, SNAP

14. The 2017–2018 data measures are for the past two weeks.
BITLER, HOYNES, and SCHANZENBACH 129

participation increased more between February and April in states with
larger increases in unemployment rates (see online appendix figure 3)
following the pattern found in prior downturns (Bitler and Hoynes 2016).15
Additionally, during COVID-19 Congress made temporary changes that
increased both participation and (for many participants) benefit levels.
Usually, SNAP benefits are reduced as a household’s income increases, with
a maximum monthly benefit in fiscal year 2021 of about $170 per person
reduced by 30 cents for each additional dollar in income (after allowable
deductions).16 While state and federal health emergencies are in progress,
states can award all SNAP participants the maximum benefit (a provision
known as the Emergency Allotment). This increases SNAP spending (hold-
ing participation constant) and provides an average increase in benefits of
40 percent to those on SNAP with higher incomes, such as the working
poor (for whom SNAP tops up earnings) who have been at particular risk
for job loss. To date there has been no benefit increase for the most dis
advantaged SNAP recipients who were already receiving the maximum
benefit. Additionally, states are temporarily allowed to extend eligibility
periods for currently participating households for six months—under
normal circumstances recipients are required to reapply for benefits every
6 to 12 months—so offices already stretched by health-related office
closures and the need to socially distance could concentrate on screening
new applicants. This temporary policy change increased SNAP partici-
pation by reducing the flows out of the program during the pandemic.
As a result, SNAP spending and participation are increasing with
unprecedented speed, as shown in figure 3, but as we show below, the
magnitude is small relative to UI and the economic impact payment.
Although national data on SNAP participation only come with some lag,
the figure presents the percentage increase in SNAP participation (dark
solid line) across forty-three states that have released their data for April or
May (these states account for 97 percent of SNAP participation). Relative
to February, SNAP participation increased by 12 percent in April, and by
17 percent by May. For comparison, SNAP participation increases during
the Great Recession are shown as the dark dotted line. It took 9 months
to see the same SNAP participation increase during the Great Recession,

15. Worth noting, Florida experienced the largest increase in SNAP participation, likely
due in part to their strong administrative system for SNAP developed to quickly deploy
Disaster-SNAP after hurricanes. Rosenbaum (2020) provided SNAP data.
16. The maximum benefit for a family of four in fiscal year 2021 is $680 or $170 per
person (USDA 2020).
130 Brookings Papers on Economic Activity, Summer 2020

Figure 3. Percentage Increase in SNAP Participation and Spending Since Business
Cycle Peak: COVID-19 versus Great Recession

Percent increase in SNAP participants, spending since business cycle peak

100

80 Spending,
COVID-19

40 Spending,
Great Rec.
Participation,
20 COVID-19

Participation, Great Rec.

1 2 3 4 5 6 7 8 9 10 11
Months since business cycle peak

Source: Authors’ calculations of Great Recession spending and caseload data, and February 2020
caseload data, from USDA, Food and Nutrition Service, SNAP Data National Level Annual Summary.
Notes: Growth in caseloads in March–May 2020 calculated from states that have reported caseload
data as of July 31, 2020. Forty-three states released April SNAP participation (42 states in May), and
these states made up 97 percent of all SNAP participation in February. Growth in SNAP spending in
2020 is reported in Daily Treasury Statements through July 31, 2020. All series are plotted as growth by
month since the business cycle peak, which was December 2007 for the Great Recession and February
2020 for the COVID-19 recession.

but of course unemployment also grew more slowly during that recession.17
SNAP spending (light solid line) is calculated using daily Treasury state-
ments and compares spending on SNAP by month through July relative
to spending in February. Some of the spending increase is due to the new
P-EBT program, which provides benefits patterned after SNAP to families
who lost access to free or reduced-price meals due to school closures.
By the end of July, SNAP spending has more than doubled. Our calcula-
tions suggest about 20 percent of the increase is explained by increases
in participation, 40 percent is due to paying all participants the maximum

17. Online appendix figure 4 shows the growth of SNAP spending and participation for
the twelve months leading up to the unemployment rate peak during the Great Recession.
The patterns are qualitatively similar.
BITLER, HOYNES, and SCHANZENBACH 131

benefit, and 40 percent is from P-EBT payments. Some of this increase
will end once state and federal health emergencies end. Spending grew
much more slowly during the Great Recession (light dotted line) and
increased substantially when the 15 percent increase in maximum SNAP
benefits authorized by Congress as part of the American Recovery and
Reinvestment Act stimulus package was implemented.
The congressional policy response also included large expansions
to UI, including a $600 per week supplement, a 13-week extension of fully
federally funded benefits, and an expansion of eligibility for self-employed
and gig-economy workers and other patches to reach workers who were
previously excluded from eligibility (under the new Pandemic Unemploy-
ment Assistance or PUA program).18 The number of UI participants has
increased to record levels, with 34.5 million total continuing claims through
the week ending July 4, as shown in online appendix figure 5. After their
early May peak, regular continuing claims have started to decrease while
PUA claims, after considerable delay in initiation, started to increase.
The onetime economic impact payments included in the CARES Act
provide $1,200 per adult ($2,400 for a married couple) and $500 per
dependent under 17. This was structured as a fully refundable tax credit,
phased out beginning at annual incomes of $150,000 for married couples,
$112,000 for head of household filers, and $75,000 for single filers. The
Treasury provided automatic payments for all who filed federal taxes in
tax years 2018 or 2019 as well as many elderly or disabled individuals
receiving payments through Social Security or Veteran’s Affairs programs.19
However, entire families that included any immigrant adult without a Social
Security number were ineligible, thus excluding many citizen children
and spouses (if not in the military). The initial payments were made to
those with direct deposit information during the week of April 17 and paper
checks followed more slowly after that.
Putting this all together, figure 4 shows weekly spending on economic
impact payments, UI, and SNAP (including P-EBT) calculated from
daily Treasury statements.20 The increase in UI payments has averaged

18. The federal government also is funding the “waiting” week for UI, so benefits get
out more quickly, and most states suspended search requirements for obtaining UI during the
health crisis through May at least.
19. Some of the Social Security Administration groups had to submit forms to receive
dependent payments.
20. Here we follow Tedeschi (2020), who estimates economic impact payments and
UI payments by calculating year-over-year changes by week. We also use this approach for
SNAP spending.
132 Brookings Papers on Economic Activity, Summer 2020

Figure 4. Weekly Spending on Unemployment Insurance, Economic Impact Payments,
and SNAP
New spending per week, billions

Unemployment insurance
140 Economic impact payments
120 SNAP

100

40
CARES Act
20 passed

-M
27 ar
-M
3- r
10 rA
-A
17 pr
-A a
p
24 pr
-A
1- pr
M
8- y
M
15 ay
-M
22 ay
-M a
29 ay
-M
5-
12 n
19 n Ju
-J ay
u
2026 n-J
-J
3-
10 l
17 l
24 l
31 l
u
un
Ju
-J
-J
-J
-J u
u
u
ul

Source: Authors’ tabulations of Daily Treasury Statements through July 31 for SNAP, unemployment
insurance benefits, and IRS tax refunds to individuals.
Notes: We difference expenditures from the inflation-adjusted same-week payments in 2019 to net out
the seasonality in payments and to separate economic impact payments from usual tax refunds. We
censor economic impact payments at zero prior to the week of April 17.

$23.5 billion per week from May through July. We estimate $131 billion in
economic impact payments were made in mid-April when the direct
deposit payments were made, with smaller amounts paid in subsequent
weeks as the paper checks rolled out. Increases in SNAP, the only program
with payments narrowly targeted to low-income families, hover around
$1 billion per week, with some weekly fluctuation due to variation across
states in the timing of monthly SNAP benefit payments and disbursal
of P-EBT benefits. Between these three categories of spending, nearly
$600 billion in new expenditures occurred between April and July—almost
$360 billion through UI, $220 billion through economic impact payments,
and just over $16 billion in new spending came through SNAP.21
There is some emerging evidence that these payments are helping
alleviate hardship. For example, unemployed workers who report receiv-
ing UI have lower levels of food insecurity than do those who unsuccess-
fully attempted to receive UI. Food insecurity rates reported in the COVID

21. Online appendix figure 6 shows cumulative weekly spending using the same data.
BITLER, HOYNES, and SCHANZENBACH 133

Impact Survey dropped from 23 percent in April to 20 percent in June for
respondents overall, and from 34 percent to 27 percent among respondents
with children (figure 1). Furthermore, new evidence finds that receipt
of P-EBT payments decreases measures of food hardship (Bauer and
others 2020). Despite noteworthy improvements, these measures are still
extremely elevated, and are generally worse for families with children, and
for Black and Hispanic respondents.

III. With This Policy Response, Why Is There Need?
Given the policy response to date, why do we see such large unmet economic
need? There are three driving factors: delays in the receipt of payments
that were authorized, modest benefit levels (for programs other than UI),
and holes in coverage. In this section, we describe elements of the policy
implementation, including slow rollout, cumbersome administrative pro-
cesses, as well as more structural deficiencies.
The available real-time evidence shows that despite high levels of
aggregate claims, many workers, especially those with low levels of educa-
tion, are not receiving UI. We establish this finding from survey and admin
istrative data sources, and it is consistent with experiences during previous
recessions. Panel A of table 1 presents data from week 3 of COVID Impact
Survey data collected May 30–June 6.22 We tabulate data on receipt of UI and
SNAP among workers reporting being on furlough. The survey asks “In the
past 7 days, have you either received, applied for, or tried to apply for any
of the following forms of income assistance, or not?” and the interviewer
asks about UI and SNAP. The table presents the responses separately for
those with a high school education or less, some college, or a college degree
or more. The results show striking disparities in access to UI payments;
among furloughed persons with a high school degree or less, 42 percent
were receiving UI compared to 52 percent for those with a college degree
or more. And this disparity in access to UI is consistent with prior reces-
sions. Panel B of table 1 presents a similar gradient for the Great Recession
using the 2008 Panel of the Survey of Income and Program Participa-
tion (SIPP). Using the sample of individuals in short-term unemployment
near the trough of the Great Recession, 29 percent of those with a high
school degree or less were receiving UI compared to 47 percent of college

22. COVID Impact Survey: Version 1, National Opinion Research Center, University of
Chicago, https://www.norc.org/Research/Projects/Pages/covid-impact-survey.aspx.
134 Brookings Papers on Economic Activity, Summer 2020

Table 1. Program Receipt among the Unemployed
Any UI Any SNAP Both UI and Neither
(%) (%) SNAP (%) (%)
(1) (2) (3) (4)
Panel A: Furloughed individuals, June 2020
≤ High school 42 11 6 52
Some college 55 24 18 38
Bachelor degree or higher 52 9 6 46
Panel B: Short-term unemployed individuals, 2008
≤ High school 29 29 6 48
Some college 37 21 5 47
Bachelor degree or higher 47 6 3 50
Source: Authors’ tabulations of the COVID Impact Survey (panel A) and the 2008 SIPP Panel (panel B).
Note: We tabulate data on receipt of UI and SNAP, where the survey asks “In the past 7 days, have you
either received, applied for, or tried to apply for any of the following forms of income assistance, or not?”
The sample consists of those reporting they are unemployed due to furlough at the time of the survey.
Panel B includes individuals age 20–59 who were unemployed and looking for work for at least a week
in the first month of wave 6 of the 2008 SIPP (January–April 2010) and had been unemployed for fewer
than four months. Receipt of UI and SNAP is measured for the first month of wave 6. UI refers to own
receipt and SNAP refers to receipt within the household. All statistics are weighted to be representative
of the adult population.

graduates. It is also important to note that this table suggests that there is
only partial overlap between UI and SNAP receipt among the unemployed/
furloughed, and a substantial share obtain SNAP but not UI. Around half
of furloughed (during COVID-19) or short-term unemployed (during the
Great Recession) report receiving neither UI nor SNAP.
To explore why UI does not reach all unemployed workers, now and
in previous recessions, we use the 2019 CPS Annual Social and Economic
Supplement (which covers the 2018 calendar year) and the 2020 UI
calculator in Ganong, Noel, and Vavra (2020) to simulate the share of
individuals age 20–59 with positive earnings who would be eligible for UI
(under normal UI rules, i.e., without federal expansions) if they became
unemployed.23 There are sharp disparities in eligibility, with much lower

23. The code for the Ganong, Noel, and Vavra (2020) calculator is available at https://
github.com/ganong-noel/ui_calculator. Ganong, Noel, and Vavra (2020) also present eligibility
estimates using their calculator; their approach differs slightly from ours. They focus on
all workers who are US citizens, have hourly wage and salary earnings above the federal
minimum wage, and who are eligible for UI based on their earnings history. Our sample
differs in that we restrict the sample to workers age 20–59 and expand it to include all workers
regardless of immigration status and with any positive earnings, not just those with wage
and salary earnings above the federal minimum wage. When estimating potential eligibility
should they be laid off and average weekly benefits, we treat workers who are likely unauthor-
ized immigrants as ineligible for UI benefits. We also ignore self-employment income in
determining UI eligibility and benefits.
BITLER, HOYNES, and SCHANZENBACH 135

eligibility rates for those in lower-income families (see online appendix
figure 7). For workers in families with income below 100 percent of
poverty, only 63 percent are eligible for UI compared to 87 percent among
all workers. Among those with income below poverty, 14 percent of the
ineligible are unauthorized (not eligible to work legally), another 7 percent
are ineligible due to being self-employed, and 17 percent are authorized and
have wage and salary earnings, but do not meet the work history require-
ments.24 Importantly, the new PUA provisions in the CARES Act have
attempted to fill the gap in eligibility for the self-employed and those with
insufficient work history so it is possible that more of these 7 + 17 percent
now have UI eligibility; changes have not altered ineligibility rates for
unauthorized workers. Thus, as many as 14 percent of those under the
poverty level may still be ineligible under the best-case scenario. In addi-
tion, there is widespread variation in the share of those unemployed who
obtain UI conditional on being eligible. O’Leary and Wandner (2020)
report that in 2018, the share of the eligible unemployed receiving UI
ranged from 10.5 percent in North Carolina to 95 percent in Rhode Island.
Murray and Olivares (2020) report that states with higher rates of pre-
COVID-19 UI utilization among those eligible are paying out more claims
in the COVID-19 era, suggesting a role for administrative burdens.
Next, we turn to real-time administrative data to assess how the UI
system responded to this unprecedented increase in unemployment. Ideally
we would present, weekly and by state, the number of persons receiving
regular UI, PUA, and the $600 supplement, along with the dates of ini-
tiation for the new programs. While we (and others) have made valiant
attempts to assemble this, as of this writing there is no systematic data
source available to identify this information. One approach is to use Depart-
ment of Labor reports of weekly continuing claims. However, many con-
cerns have been raised about the use of continuing claims to capture the
number of recipients, particularly for PUA. First, the count of continuing
claims is the number of weeks times people, not the number of people; this
is particularly problematic when there are delays in processing and back
payments are issued with first payments.25 Second, continuing claims can
include claimants who are still pending a determination and denials can
occur after this stage (Hedin, Schnorr, and von Wachter 2020). Additionally,
PUA continuing claims appear to be inconsistently reported during the

24. We follow Passel (2007) to identify survey respondents as unauthorized immigrants.
25. For example, if it takes four weeks to process the claim when the first payment is
made, it will count “4” in continuing claims that week due to the back pay.
136 Brookings Papers on Economic Activity, Summer 2020

COVID-19 crisis.26 Another approach is to use Department of Labor reports
of weekly initial claims, yet these also have weaknesses, including sub
sequent denials, double counting due to returning to UI after brief return
to work, and, particularly for PUA, capturing possible fraud.27
Despite these data challenges, the available evidence clearly points to
significant delays, especially in the rollout of PUA across states. This is not
surprising, as states had to design entirely new methods to ensure eligibility
for PUA, and states varied widely in their administrative capacity and the
need for social distancing in the early months of the pandemic. Addition-
ally, for a state to receive federal reimbursements for PUA, its recipients
must be ineligible for state UI. In practice, in some states PUA applicants
must apply to and be rejected from the regular UI program before they
could separately apply for PUA, leading to further delays. Using informa
tion from state press releases, we can document significant delays and wide
differences in when PUA was first paid out, ranging from as early as March
in New Hampshire (which had passed a program expanding UI to the
self-employed even before the CARES Act), to April 30 in California, to
May 11 in West Virginia, and May 26 in Kansas.28 States also varied in the
timing of their payment of the federal supplemental $600 weekly payment
(FPUC) which was meant to go to all UI recipients.
Using less granular monthly data, we can also calculate for the United
States a more reliable measure of the UI utilization rate by taking the
ratio of “first UI payments” available monthly for regular and PUA UI
from the Department of Labor (currently through May 2020 for all states

26. Take Florida for example: the first initial claim reported to Department of Labor
for PUA was for the week ending June 27 despite an April 25 press release announcing
people could start applying for PUA. Additionally, there have been no continuing claims
reported as data were downloaded August 9, yet the state data dashboard reports they have
paid out $453 million of PUA as of August 9. Using data shared by Murray and Olivares
(2020) and Cajner and others (2020), we document similar discrepancies between the timing
of when the first week first claims were reported to the Department of Labor and when states
reported that they started accepting PUA claims, with at least twenty-three states accept-
ing PUA applications at least seven days before the first week of initial claims was reported
to Department of Labor, and with the average difference being twenty-nine days. We thank
them for generously sharing their data.
27. For example, the state of Ohio froze 270,000 claims as of August 7 in order to
investigate fraud at a time when about 500,000 PUA claims had been paid, and the US Labor
Department inspector general raised concerns about fraud in a May 26 Alert Memorandum.
28. Many states also started by sending PUA applicants the minimum payment (plus,
where relevant the additional $600 federal payment), and then later determined actual
payment eligibility amounts and sent back payments where appropriate.
BITLER, HOYNES, and SCHANZENBACH 137

reporting PUA first payments) to the total number of unemployed.29 First
payments get around the problem of subsequent denials as well as being
an unduplicated count of recipients. Combining regular state UI and PUA
first payments, we find that 6.4 percent of the unemployed had received a
first payment in March, rising to 53.9 percent in April and 84.9 percent in
May (see online appendix table 6). If we limit to payments for PUA, we
find 1.6 percent of the unemployed received a first payment by April
2020 rising to 11 percent in May.
In summary, the combination of real-time survey and administrative
data, the historical patterns, and policy changes during COVID-19 suggest
that while UI is serving the majority of the unemployed, it is far from
universal. During COVID-19, UI has been slow to reach the unemployed
and there is a sizeable share—disproportionately those with low levels of
education—who are not receiving benefits. This is consistent with available
pre-COVID-19 evidence.
Coverage was incomplete for the economic impact payments as well.
According to the daily Treasury statements (shown in online appendix
figure 6), cumulative payments for the onetime economic impact pay-
ments ($1,200 per adult and $500 per child under 17) through the end of
July 2020 are around $215 billion. However, despite the apparent uni-
versality of the payment for those with income below the high-income
phase-out level, the design of the payment scheme has left out the most
disadvantaged Americans. First, the law excludes immigrant families
who are deemed ineligible if any adult or spouse lacks a Social Security
number (unless the family included a member of the military).30 Second, the
payments were sent automatically, with no additional action, for tax filers
(in 2018 or 2019) and those receiving benefits from the Social Security
Administration or Veterans Affairs. Marr and others (2020) estimate that
12 million nonfilers are eligible for the relief payment but did not auto
matically receive it. Instead, to receive these payments individuals are
required to apply for the payment through a new IRS nonfiler tool. This
nonfiler population is a disadvantaged group with low incomes, and an

29. Regular and PUA UI first payments come from the 902P and 5159 forms from the
Department of Labor, respectively. Pandemic Emergency Unemployment Compensation first
payments are very small, so we exclude them from the graphs. For the denominator we use
CPS monthly estimates of those unemployed (adjusted for changes in those with a job but
not at work and not in the labor force over the previous year).
30. Also ineligible are adult dependents, 17-year-olds, and college students whom their
parents can claim as dependents.
138 Brookings Papers on Economic Activity, Summer 2020

Figure 5. Timing of Pandemic Assistance Payments for P-EBT

Share of children participating in National School Lunch Program
living in states disbursing P-EBT (percent)

P-EBT to SNAP
families
50
P-EBT to non-SNAP
families
25 Families
First Act CARES Act

Mar 9 Mar 23 Apr 6 Apr 20 May 4 May 18 Jun 1 Jun 15 Jun 29 Jul 13 Jul 27

Source: Bauer and others (2020), and authors’ calculations from state departments of health services.
Note: The solid (dashed) line displays the share of children who participate in the National School
Lunch Program who live in states that have disbursed P-EBT payments to families receiving free or
reduced-price meals who also participate in SNAP (do not participate in SNAP).

estimated three-quarters of them are eligible for SNAP or Medicaid. Based
on the Urban Institute Coronavirus Tracking Survey, wave 1—fielded
between May 14 and May 27—41 percent of adults with income below
poverty reported that they had not received their economic impact payment
compared with 27 percent among those with income between 100 and
250 percent of poverty and 14 percent among those with income between
250 and 400 percent of poverty (Holtzblatt and Karpman 2020).
Another source of delay in benefits reaching needy families came
from having to create a new program in the midst of the pandemic. When
schools across the United States closed in mid-March, 30 million students
lost daily access to free or reduced-price school meals. To offset this loss,
Congress authorized the new P-EBT program to provide food benefits
to families who lost subsidized school meals. In order to participate,
though, states had to set up and receive approval from the USDA for this
completely new program. Payments came out slowly, as shown in fig-
ure 5. Two months after the Families First Act authorized the program,
very few states had made payments; about 15 percent of eligible families
lived in states where P-EBT benefits began being dispersed to those on
BITLER, HOYNES, and SCHANZENBACH 139

SNAP (where preexisting debit cards could be used), and fewer than
10 percent lived in states where non-SNAP recipients eligible for school
meals programs were dispersing P-EBT benefits. Many states did not
make retroactive payments until June or July.
In sum, this discussion illustrates the delays and incomplete coverage
in the policy response. Also, among those eligible we have incomplete
take-up of these programs. Why? This is a direct result of the “application-
based” policy environment. Across the different relief provisions, some
payments were made automatically (recovery rebate for previous tax filers,
increase in SNAP benefit for existing participants) while others required
application (UI, recovery rebate for some nonfilers, P-EBT for those not
on SNAP in some states). Decades of research show that take-up rates are
incomplete when an application is required. Individuals need to know
about the programs to access them (Currie 2006). Administrative hassles
are built into many programs and contribute to the less-than-complete
take-up (Herd and Moynihan 2019). In addition, as the COVID-19 crisis
has highlighted, states have made policy choices that result in differential
capacity to quickly enroll newly unemployed individuals.

IV. Putting the Policy Response in the Context
of the Broader Social Safety Net
The COVID-19 crisis has been met with an extraordinary economic policy
response. It is important to understand, though, that the US social safety
net—the foundation beneath this policy response—has been redesigned
in recent decades in ways that have made it less responsive to economic
downturns. In the years following the Great Recession, many states have
reduced the generosity of their UI programs. Median replacement rates to
low-income workers are below 50 percent in many states (online appendix
figure 8), providing very limited earnings replacement. More generally, our
social safety net has shifted toward a work-conditioned social safety net,
using earnings subsidies to increase incomes among workers but offering
relatively little out-of-work assistance (to those not elderly or disabled).
These changes have been ushered in through the 1996 welfare reform law,
expansions to the earned income tax credit (EITC), and, for some popu-
lations, work requirements for SNAP. More recently, work requirements
have been adopted in some states for Medicaid and regulations imple-
mented to expand SNAP work requirements. The result is a social safety
net with an emphasis on promoting and rewarding work, a system that
140 Brookings Papers on Economic Activity, Summer 2020

may be adequate during times of low unemployment but provides too
little insurance against job loss and economic shocks.31
Recent work by Bitler, Hoynes, and Iselin (2020) and Bitler and Hoynes
(2016) summarizes how participation in SNAP, UI, the EITC, and cash
welfare varies with the unemployment rate at the state level, and how that
has changed over time. In the period since 2007, only UI shows a robust
countercyclical response, with a 1 percentage point increase in unemploy-
ment leading to an 18 percent increase in UI spending (Bitler, Hoynes,
and Iselin 2020). SNAP has a weaker response, with a 1 percentage point
increase in the unemployment rate leading to a 7 percent increase in SNAP
spending. Neither the work-conditioned EITC nor cash welfare system-
atically change in response to the economy. In other words, despite its
important role in reducing poverty, the EITC is poorly suited to insure
consumption against job loss.32
Overall, the literature shows that on the eve of the COVID-19 crisis,
the safety net was providing uneven and incomplete protection. While UI
is strongly countercyclical overall, not all unemployed workers receive
benefits, including undocumented immigrants and those with inconsistent
work histories. Cash welfare does not respond to aggregate economic need,
and the EITC is not designed to provide insurance against job loss. SNAP
does have the capacity to expand during economic downturns, but benefits
are modest, and since its benefits are food vouchers they are only partially
fungible. In addition, recent policy changes risk further dampening the
protective effects of SNAP by imposing stricter work requirements among
nondisabled adults without dependents and reducing participation among
immigrants and families with mixed immigration status.33

31. For reference, antipoverty effects of existing programs in 2018 for children, adults
with and without children, and the elderly are presented in online appendix figure 9. The
EITC has the largest antipoverty impact for children and adults who live with them, followed
by SNAP, housing assistance, and school meals. Among the elderly and childless adults,
Social Security overwhelmingly has the largest antipoverty effect.
32. Bitler, Hoynes, and Kuka (2017) show that lack of cyclicality of the EITC masks
two opposing responses: a procyclical effect for single filer EITC recipients (whose EITC
payment falls or is lost altogether with economic shocks) and a countercyclical effect for
married filers (or more generally those with higher predicted earnings) for whom a labor
market shock can bring them down into EITC eligibility.
33. When labor market conditions are poor, states can waive SNAP time limits when
particular economic conditions (based on employment statistics in the state or local area) are
met, so that food assistance is not conditional on employment during bad economic times.
The Trump Administration issued a new rule effective April 1, 2020, making it more difficult
to obtain time-limit waivers. Importantly, the new rule requires that states have elevated
unemployment rates for at least the previous twelve months, slowing the ability of the program
to respond to immediate need at the onset of an economic downturn.
BITLER, HOYNES, and SCHANZENBACH 141

V. Needed Policies Moving Forward

Our analysis leads us to two sets of recommendations. The first set of
recommendations relates to changes that need to occur in the short-term
to address the current recession. The increased payments authorized by
Congress for UI, SNAP, and for missed school meals have been crucial if
incomplete responses, but all are in danger of not being continued as cases
continue to surge at the time of this writing. For example, the $600/week
UI supplement was allowed to expire at the end of July, and PUA (cover-
ing the self-employed) is scheduled to expire at the end of December.
The temporary increase in SNAP payments is not tied to the state of the
economy, but instead is only authorized through the duration of national
and state health emergencies. P-EBT has not yet been extended into the
2020 school year for students who are engaged in remote learning. This
potential rollback in support is occurring despite an unemployment rate
that still exceeds the maximum rates experienced in the Great Recession. It
is too soon to phase down increased payments that provide crucial relief to
families experiencing hardships. The current policy response, in particular
those applying to UI and SNAP, should remain in place and be phased out
only as the economic emergency recedes.
As a general matter, we have designed a safety net that needs an addi-
tional boost during recessions. Usual state UI systems generally provide
low payments (as a share of wages) for a short duration. SNAP benefits
are modest and are intended to supplement other food resources. The EITC
tops up low earnings but is not countercyclical. Because these limita-
tions are known, and since there is a high cost both to policy uncertainty
and to delays in relief payments, we think it is wise to build automatic
expansions into key safety net programs during recessions, as proposed
in Boushey, Nunn, and Shambaugh (2019). For example, following the
successful policies of the 2009 stimulus, maximum SNAP benefits should
be increased by 15 percent (thereby reaching those most disadvantaged
recipients who did not gain from the current SNAP expansions). In order
to support a work-based safety net, the UI system should be redesigned
to provide more insurance and to reach a larger share of disadvantaged
unemployed workers during recessions, for example by making permanent
the pandemic expansions to UI that extended coverage to self-employed
and gig workers and to those with limited work histories, although this
may require rethinking the UI tax system for these groups. We need to
build a harmonized federal and state data system to facilitate automated
relief payments to all eligible Americans. For example, information from
142 Brookings Papers on Economic Activity, Summer 2020

state-administered SNAP and Medicaid data systems should have been
available to the Treasury to facilitate EIPs for this group. Finally, this crisis
has made clear the need for states to increase their administrative capacity
for their programs, particularly UI.

VI. Conclusions
The COVID-19 recession is unlike previous recessions due to its depth
and speed of onset. In response to this shock, Congress enacted a number
of smart short-term fixes to the safety net that have improved its ability to
insure low-income families during this recession, including increasing
UI payments and extending eligibility, increasing SNAP payments to some
participants, sending cash relief payments (EIP), and introducing a new
program to replace missed school meals (P-EBT). Without question, these
policies have improved the responsiveness of the safety net to this crisis
and have reduced suffering that would have occurred without these actions.
Even with these valuable policy responses, there is still tremendous
unmet need. Food insecurity has sharply increased, as has the share of
families relying on emergency food pantries. Some excess suffering
occurred because much of the policy response was slow to roll out and
reach needy families. The available yet incomplete data suggest a sizeable
subset who experienced shocks and have not received safety net payments;
for example, some workers who lost their jobs are not receiving benefits
from UI or SNAP. In addition, there remain great economic risks if addi-
tional policy responses are removed too quickly, because the underlying
US safety net does not provide adequate protection during recessions.

ACKNOWLEDGMENTS We thank Seth Murray and Edward Olivares
for sharing data on the timing of PUA claims, and Tomaz Cajner, Andrew
Figura, Brendan Price, David Ratner, and Alison Weingarden for sharing
their code and data on this. Raheem Chaudhry, Danea Horn, Abigail Pitts, and
Natalie Tomeh provided excellent research assistance. We thank Lisa Barrow,
Stacy Dean, Robert Moffitt, Zach Parolin, Brendan Price, Dottie Rosenbaum,
Jesse Rothstein, Geoff Schnorr, Jay Shambaugh, Louise Sheiner, Tim Smeeding,
Ernie Tedeschi, Till von Wachter, Justin Wolfers, and Abigail Wozniak for
helpful comments.
BITLER, HOYNES, and SCHANZENBACH 143

References
Aaronson, Stephanie R., Mary C. Daly, William L. Wascher, and David W. Wilcox.
2019. “Okun Revisited: Who Benefits Most from a Strong Economy?” Brookings
Papers on Economic Activity, Spring, 333–404.
Ananat, Elizabeth O., and Anna Gassman-Pines. 2020. “Snapshot of the COVID
Crisis Impact on Working Families.” Econofact, March 30. https://econofact.org/
snapshot-of-the-covid-crisis-impact-on-working-families.
Bauer, Lauren, Abigail Pitts, Krista Ruffini, and Diane Whitmore Schanzenbach.
2020. The Effect of Pandemic EBT on Measures of Food Hardship. Washington:
The Hamilton Project, Brookings Institution. https://www.hamiltonproject.org/
assets/files/P-EBT_LO_7.30.pdf.
Bitler, Marianne, and Hilary Hoynes. 2016. “The More Things Change, the More
They Stay the Same? The Safety Net and Poverty in the Great Recession.”
Journal of Labor Economics 34, no. S1: 403–4. https://www.journals.uchicago.
edu/doi/abs/10.1086/683096.
Bitler, Marianne, Hilary Hoynes, and John Iselin. 2020. “The Cyclicality of
Cash and Near Cash Transfer Programs.” National Tax Journal 73, no. 3:
759–80.
Bitler, Marianne, Hilary Hoynes, and Elira Kuka. 2017. “Do In-Work Tax Credits
Serve as a Safety Net?” Journal of Human Resources 36, no. 2: 358–89.
Blau, Francine, Josefine Koebe, and Pamela Meyerhofer. 2020. “Essential and
Frontline Workers in the COVID-19 Crisis.” Econofact, April 30. https://
econofact.org/essential-and-frontline-workers-in-the-covid-19-crisis.
Boushey, Heather Ryan Nunn, and Jay Shambaugh. 2019. Recession Ready: Fiscal
Policies to Stabilize the American Economy. Washington: The Hamilton Project,
Brookings Institution. https://www.hamiltonproject.org/papers/recession_ready_
fiscal_policies_to_stabilize_the_american_economy.
Cajner, Tomaz, Andrew Figura, Brendan Price, David Ratner, and Alison
Weingarden. 2020. “Reconciling Unemployment Claims with Job Losses in
the First Months of the COVID-19 Crisis.” Finance and Economics Discus-
sion Series 2020–55. Washington: Board of Governors of the Federal Reserve.
https://www.federalreserve.gov/econres/feds/reconciling-unemployment-claims-
with-job-losses-in-the-first-months-of-the-covid-19-crisis.htm.
Currie, Janet. 2006. “The Take Up of Social Benefits.” In Poverty, the Distribution
of Income, and Public Policy, edited by Alan Auerbach, David Card, and John
Quigley. New York: Russell Sage Foundation.
Ganong, Peter, Pascal Noel, and Joseph Vavra. 2020. “US Unemployment Insurance
Replacement Rates during the Pandemic.” Working Paper 27216. Cambridge,
Mass.: National Bureau of Economic Research. https://www.nber.org/papers/
w27216.
Han, Jeehoon, Bruce D. Meyer, and James X. Sullivan. 2020. “Income and Poverty
in the COVID-19 Pandemic.” In the present volume of Brookings Papers on
Economic Activity.
144 Brookings Papers on Economic Activity, Summer 2020

Hedin, Thomas J., Geoffrey Schnorr, and Till von Wachter. 2020. “An Analysis
of Unemployment Insurance Claims in California during the COVID-19
Pandemic.” Policy Brief. Los Angeles: California Policy Lab. https://www.
capolicylab.org/publications/june-11th-analysis-of-unemployment-insurance-
claims-in-california-during-the-covid-19-pandemic/.
Herd, Pamela, and Donald Moynihan. 2019. Administrative Burden: Policymaking
by Other Means. New York: Russell Sage Foundation.
Holtzblatt, Janet, and Michael Karpman. 2020. “Who Did Not Get the Economic
Impact Payments by Mid-to-Late May, and Why?” Washington: Urban Institute.
https://www.urban.org/research/publication/who-did-not-get-economic-impact-
payments-mid-late-may-and-why.
Hoynes, Hilary, Doug Miller, and Jessamyn Schaller. 2012. “Who Suffers During
Recessions?” Journal of Economic Perspectives 26, no. 3: 27–48. https://www.
aeaweb.org/articles?id=10.1257/jep.26.3.27.
Hoynes, Hilary, and Diane Whitmore Schanzenbach. 2018. “Safety Net Investments
in Children.” Brookings Papers on Economic Activity, Spring, 89–132.
Karpman, Michael, Stephen Zuckerman, and Dulce Gonzalez. 2018. The Well-Being
and Basic Needs Survey: A New Data Source for Monitoring the Health and
Well-Being of Individuals and Families. Washington: Urban Institute. https://
www.urban.org/sites/default/files/publication/98919/the_well-being_and_basic_
needs_survey_1.pdf.
Marr, Chuck, Kris Cox, Kathleen Bryant, Stacy Dean, Roxy Caines, and Arloc
Sherman. 2020. Aggressive State Outreach Can Help Reach the 12 Million
Non-Filers Eligible for Stimulus Payments. Washington: Center on Budget and
Policy Priorities.
Montenovo, Laura, Xuan Jiang, Felipe Lozano Rojas, Ian Schmutte, Kosali Simon,
Bruce Weinberg, and Cody Wing. 2020. “Unequal Employment Impacts of
COVID-19.” Econofact, June 1. https://econofact.org/unequal-employment-
impacts-of-covid-19.
Murray, Seth, and Edward Olivares. 2020. “Job Losses during the Onset of
the COVID-19 Pandemic: Stay-at-Home Orders, Industry Composition, and
Administrative Capacity.” Working Paper. Social Science Research Network,
June 18. http://dx.doi.org/10.2139/ssrn.3633502.
O’Leary, Christopher J., and Stephen A. Wandner. 2020. “An Illustrated Case for
Unemployment Insurance Reform.” Working Paper 19-317. Kalamazoo, Mich.:
Upjohn Institute. https://research.upjohn.org/up_workingpapers/317/.
Parolin, Zachary, Megan Curran, and Christopher Wimer. 2020. “Forecasting
Estimates of Poverty during the COVID-19 Crisis.” Poverty and Social Policy
Brief 4, no. 6. New York: Center on Poverty and Social Policy, Columbia Univer-
sity. https://www.povertycenter.columbia.edu/s/Forecasting-Poverty-Estimates-
COVID19-CPSP-2020.pdf.
Passel, Jeffrey. 2007. “Unauthorized Migrants in the United States: Estimates,
Methods, and Characteristics.” Working Paper 57. Paris: OECD Working Party
on Migration. https://www.oecd-ilibrary.org/social-issues-migration-health/
unauthorized-migrants-in-the-united-states_110780068151.
BITLER, HOYNES, and SCHANZENBACH 145

Rosenbaum, Dottie. 2020. “Boost SNAP to Capitalize on Program’s Effectiveness
and Ability to Respond to Need.” Washington: Center on Budget and Policy
Priorities.
Schanzenbach, Diane Whitmore, and Abigail Pitts. 2020. Estimates of Food
Insecurity during the COVID-19 Crisis: Results from the COVID Impact
Survey, Week 1 (April 20–26, 2020). Evanston, Ill.: Institute for Policy Research,
Northwestern University. https://www.ipr.northwestern.edu/documents/reports/
food-insecurity-covid_week1_report-13-may-2020.pdf.
Tedeschi, Ernie. 2020. “U.S. Fiscal Tracker: Major Disbursements Fall to
$31 Billion.” Evercore ISI Macro Note, June 12.
US Department of Agriculture (USDA). 2020. “Supplemental Nutrition Assistance
Program (SNAP) Fiscal Year (FY) 2021 Maximum Allotments and Deductions.”
https://fns-prod.azureedge.net/sites/default/files/media/file/FY21-Maximum-
Allotments-Deductions.pdf.

File and source

File: excess-savings-during-the-covid-19-pandemic-feds-note.pdf
Size: 241,018 bytes
SHA-256: 2bf268a4a72b67835820c25b72fb6c3a72b612f3ec9c2b70c91a3f13bcb6737f
Our copy: excess-savings-during-the-covid-19-pandemic-feds-note.pdf
Original: www.federalreserve.gov

Excess Savings During The Covid 19 Pandemic Feds Note

Summary

Full text

File and source

Related records