Labor Markets During the COVID-19 Crisis: A Preliminary View
Summary
NBER Working Paper No. 27017, Labor Markets During the COVID-19 Crisis: A Preliminary View, by Olivier Coibion, Yuriy Gorodnichenko and Michael Weber, dated April 2020. The paper uses repeated surveys of households in the Nielsen Homescan panel, a pre-crisis wave with 18,344 responses and a wave run April 2nd-6th, 2020 with 9,445 responses. The authors estimate 20 million lost jobs by April 6th, report the adjusted employment-to-population ratio falling from 60% to 52.2% and the adjusted unemployment rate rising from 4.2% to 6.3%, and describe a decline in labor force participation. The paper compares its measures with Bureau of Labor Statistics figures and reviews related literature. It ends with Table 2, listing survey respondents' reasons for not looking for a job.
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
NBER WORKING PAPER SERIES
LABOR MARKETS DURING THE COVID-19 CRISIS:
A PRELIMINARY VIEW
Olivier Coibion
Yuriy Gorodnichenko
Michael Weber
Working Paper 27017
http://www.nber.org/papers/w27017
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
April 2020
The views expressed herein are those of the authors and do not necessarily reflect the views of the
National Bureau of Economic Research. We thank the National Science Foundation for financial
support in conducting the surveys. We also thank Shannon Hazlett and Victoria Stevens at
Nielsen for their assistance with the collection of the PanelViews Survey. Results in this article
are calculated based on data from The Nielsen Company (US), LLC and marketing databases
provided by the Kilts Center for Marketing Data Center at The University of Chicago Booth
School of Business. Information on availability and access to the data is available at http://
research.chicagobooth.edu/nielsen.
NBER working papers are circulated for discussion and comment purposes. They have not been
peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies
official NBER publications.
© 2020 by Olivier Coibion, Yuriy Gorodnichenko, and Michael Weber. All rights reserved. Short
sections of text, not to exceed two paragraphs, may be quoted without explicit permission
provided that full credit, including © notice, is given to the source.
Labor Markets During the COVID-19 Crisis: A Preliminary View
Olivier Coibion, Yuriy Gorodnichenko, and Michael Weber
NBER Working Paper No. 27017
April 2020
JEL No. C83,D84,E31,J21,J26
ABSTRACT
We use a repeated large-scale survey of households in the Nielsen Homescan panel to
characterize how labor markets are being affected by the covid-19 pandemic. We document
several facts. First, job loss has been significantly larger than implied by new unemployment
claims: we estimate 20 million lost jobs by April 6th, far more than jobs lost over the entire Great
Recession. Second, many of those losing jobs are not actively looking to find new ones. As a
result, we estimate the rise in the unemployment rate over the corresponding period to be
surprisingly small, only about 2 percentage points. Third, participation in the labor force has
declined by 7 percentage points, an unparalleled fall that dwarfs the three percentage point
cumulative decline that occurred from 2008 to 2016.
Olivier Coibion Michael Weber
Department of Economics Booth School of Business
University of Texas at Austin University of Chicago
2225 Speedway 5807 South Woodlawn Avenue
Austin, TX 78712 Chicago, IL 60637
and NBER and NBER
ocoibion@gmail.com michael.weber@chicagobooth.edu
Yuriy Gorodnichenko
Department of Economics
530 Evans Hall #3880
University of California, Berkeley
Berkeley, CA 94720-3880
and IZA
and also NBER
ygorodni@econ.berkeley.edu
The arrival of the covid-19 virus and the policy responses have led to unprecedented numbers of initial claims for
unemployment since early 2020: over 16.5 million by April 4th, 2020, with new claims arriving at a rate of 6-7
million per week. But concerns about state governments’ inability to process so many claims in such a short period,
combined with the fact that many workers are ineligible for unemployment benefits, has led to concerns that total
job losses are being understated by these numbers. Furthermore, because official labor market indicators compiled
by the Bureau of Labor Statistics (BLS) take time to be released, the current state of the U.S. labor market remains
unclear.
Using new ongoing large-scale surveys of U.S. households much like the ones run by the BLS, we provide
some preliminary evidence on the response of labor markets in the U.S. to the current crisis. We focus on three
key variables typically measured by the BLS: the employment-to-population ratio, the unemployment rate, and the
labor force participation rate. Historically, the employment-to-population ratio and the unemployment rate are near
reverse images of one another during recessions as workers move out of employment and into unemployment (or
workers in unemployment find it harder to move into employment). More severe recessions also sometimes lead
to a phenomenon of “discouraged workers,” in which some unemployed workers stop looking for work. This leads
them to be reclassified as “out of the labor force” by the BLS definitions, so the unemployment rate can decline
along with the labor force participation rate while the employment-to-population ratio shows little recovery, not
because the unemployed are finding work but rather because they stop trying to find one. Jointly, these three
metrics therefore provide a succinct and informative summary of the state of labor markets.
Using surveys prior to and at the height of the covid-19 crisis, we provide new estimates of how these
variables have changed in the last two months. Our most recent numbers are from individuals surveyed April 2nd-
6th, 2020, and therefore provide a sneak preview at what the equivalent BLS numbers will likely show when they
are ultimately released. Our findings are striking. First, the employment-to-population ratio (the fraction of the
adult population reporting that they had a paid job) has declined by about 7.5 percentage points. With an adult
(civilian non-institutional) U.S. population of 260 million, this corresponds to nearly 20 million jobs lost as of
April 6th. This estimate is in line with (albeit even higher than) new unemployment claims through this time period
and confirms the widespread job loss.
Twenty million jobs lost relative to the pre-crisis labor force would correspond to an increase in the
unemployment rate of 12.2 percentage points, so to a level of around 16%, were all the newly unemployed looking
for work. This is, however, not what we find. When we construct an unemployment rate in an analogous manner
as the BLS (i.e., define the unemployed as not working but looking for work), we find an increase in the
unemployment rate of only 2 percentage points. This reflects the fact that most of the newly unemployed surveyed
are not looking for new work at this time, so they are defined as out of the labor force rather than unemployed.
Correspondingly, we document an extraordinary decline in the labor force participation rate of nearly 8 percentage
points. In short, we find labor market changes that differ markedly from those of a typical recession.
1
We first mention the related literature, then describe the surveys and how the resulting labor market
statistics before the crisis compare to those of the BLS prior to the crisis and finally turn to what our newer survey
implies for labor markets during the covid-19 crisis.
I Related Literature
We relate to the fast-growing literature studying the economic consequences of the covid 19 pandemic. Binder
(2020) shows that 30% - 40% of American are very concerned about the corona crisis, postponed travel and delayed
purchases of larger ticket items as early as March 2020 but became more optimistic about the unemployment
situation and revised downward their inflation expectations once being told about the cut in the federal funds target
rate on March 3rd. Fetzer et al. (2020) show the arrival of the corona virus in a country leads to a large increase in
internet searches around the world. In a survey experiment on a U.S. population, they find survey participants
vastly overestimate the mortality rate and extent of contagion. Barrios and Hochberg (2020) and Allcott et al (2020)
use internet searches, survey data, and travel data from smartphones to document that political partisanship
determines the perception of risk associated with covid-19 and non-essential travel activity. Dingel and Neiman
(2020) use data from responses to two Occupational Information Network surveys and estimate that about 37% of
jobs can be performed from home, whereas Mongey (2020) documents that employees that are less likely to be
able to work from home are mainly non-white and without a college degree. On the quantitative side, a growing
literature jointly models the dynamics of the pandemic and the economy to quantify the economic costs and
benefits of different economic policies (Atkeson (2020), Barro et al. (2020), Eichenbaum et al. (2020), Guerrieri
et al. (2020), Alvarez et al. (2020), and Dietrich et al. (2020)). Finally, our Nielsen survey builds on previous work
using the Nielsen panelists to study the formation and updating of economic expectations (Coibion et al. (2019,
2020) and D’Acunto et al (2020a, b)).
II Measuring Labor Markets using the Nielsen Survey
We start by describing the Nielsen Homescan survey and how it can be used to construct measures of labor market
outcomes. The pre-crisis wave of the survey was run between January 6th and January 27th 2020 prior to much of
the covid-19 outbreak. Potential participants were those households participating in the Nielsen Homescan, which
is a panel of 80,000-90,000 households who track their purchases on a daily basis. Nielsen allows for surveys to
be emailed to those households and respondents receive points and prizes for participating in Nielsen surveys. The
panel of households used by Nielsen is meant to be representative of the U.S. population in age, size, income, etc.
Possible imbalances are corrected using weights provided by Nielsen. We received 18,344 responses to the first
wave of the survey.
In this and subsequent waves of the survey, we asked respondents several job-related questions. First, they
were asked whether they have a paid job, with answers being either yes or no. Anyone answering yes we define as
being employed. Note that this is slightly different from the BLS, which asks respondents whether they have
2
worked in the survey reference week, and those who “did any work at all for pay or profit” are classified as
employed. This means some respondents who would be classified as employed by the BLS are classified as non-
employed using our question. Consistent with this, we find somewhat lower employment rates (as a share of adult
population) in our pre-crisis data than was the case in corresponding BLS surveys. Table 1, for example, shows
that the BLS was reporting an employment to population ratio of 61.1% in February 2020 while our survey yielded
an employment-to-population ratio of 57.7%.
Respondents who said they did not have a paid job were then asked if they were actively looking for a job,
with possible answers being yes or no. We define those who answer yes as unemployed while those who answer
no are classified as out of the labor force. Again, this is slightly different from the BLS questionnaire, which asks
individuals to select ways in which they had looked for jobs during the prior 4 weeks and only classifies individuals
as unemployed if they select answers which indicate a sufficiently active search such as posting resumes,
contacting potential employers or filling out applications (i.e., not just scanning newspaper ads). Given that we
allow individuals to specify themselves if they are “actively” searching, one might expect that this would also lead
to a higher prevalence of unemployment in our Nielsen survey than in the corresponding BLS survey. Consistent
with this, our estimated aggregate unemployment rate prior to the crisis is 8.6% while the corresponding BLS
estimates for January and February of 2020 were 3.6% and 3.5% respectively. The labor force participation rate,
however, is very similar to that estimated by the BLS: 63.1% vs 63.3% respectively.
While some difference in levels is to be expected given that questions are not identical across the two
surveys, it is important to verify that they are still capturing broadly similar features. We do so by comparing state-
level estimates of all variables from both the pre-crisis BLS and Nielsen surveys. These are illustrated in Figure 1,
which shows state level unemployment rates from the two surveys in Panel A, state level labor force participation
rates in Panel B, and state level employment to population ratios in Panel C. All three figures yield a strong positive
correlation: the two surveys identify the same states as having relatively high or low unemployment, labor force
participation, and employment to population ratios.
We can control for the initial difference in levels of unemployment that stems from the different questions
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁
asked in the surveys in the following way. Let 𝑈𝑈𝐸𝐸𝑖𝑖,𝑡𝑡 be the unemployment rate in state 𝑖𝑖 in the Nielsen survey
𝐵𝐵𝐵𝐵𝐵𝐵
in year 𝑡𝑡 and 𝑈𝑈𝐸𝐸𝑖𝑖,𝑡𝑡 be the unemployment rate in state 𝑖𝑖 in the Current Population Survey (CPS) survey in year 𝑡𝑡.
We run the followings regression
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐵𝐵𝐵𝐵𝐵𝐵
𝑈𝑈𝐸𝐸𝑖𝑖,𝑡𝑡 = 𝑏𝑏0 + 𝑏𝑏1 𝑈𝑈𝐸𝐸𝑖𝑖,𝑡𝑡 + 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 (1)
and then use the estimated coefficients to adjust Nielsen statistic for year 𝑠𝑠 is
� 𝑖𝑖,𝑠𝑠
𝑈𝑈𝑈𝑈 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁
= (𝑈𝑈𝐸𝐸𝑖𝑖,𝑠𝑠 − 𝑏𝑏�0 )/𝑏𝑏�1 . (2)
� 𝑖𝑖,𝑡𝑡
By construction, the average value of 𝑈𝑈𝑈𝑈 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐵𝐵𝐵𝐵𝐵𝐵
is now equal to the average value of 𝑈𝑈𝐸𝐸𝑖𝑖,𝑡𝑡 across states, which
may still differ from the aggregate unemployment rate (which is a population-weighted average across states). We
apply a similar procedure for employment-to-population ratios and labor force participation rates. The resulting
3
adjusted measures of aggregate unemployment, labor force participation, and employment-to-population ratios
pre-crisis from the Nielsen survey are presented in Table 1, column 4. For easier comparison to BLS numbers, we
focus primarily on these adjusted measurements (applied to both waves of our survey) from now on.
III Labor Markets since the Covid-19 Crisis
A second survey was run on households participating in the Nielsen Homescan panel between the afternoon of
April 2nd-6th, 2020. The response rate was 18.6% with 9,445 responses. The same labor market questions were
asked in both waves of the survey, so we can directly compare the two surveys to get a sense of how labor markets
have evolved since the onset of the covid-19 virus and the associated policy responses. Table 1 summarizes results
from the second wave of the survey. Our main results are as follows.
First, the employment-to-population ratio has declined sharply. Using the adjusted metrics described
above, we find that the employment ratio fell from 60% of the population down to 52.2%, a nearly 8% point
decline. As illustrated in Panel B of Figure 2, this decline in employment is enormous by historical standards and
is larger than the entire decline in the employment to population ratio experienced during the Great Recession.
Given that the U.S. civilian non-institutional population is approximately 260 million, this drop in employment to
population ratio is equivalent to 20 million people losing their jobs. This drop is even larger than the 16.5 million
new unemployment claims over this time period.
Second, we find a much smaller increase in the unemployment rate. As documented in Table 1, the
adjusted unemployment rate rose from 4.2% to 6.3%. Panel A of Figure 2 plots this rise relative to previous changes
in unemployment over the last 15 years. While this increase is the single biggest discrete jump in unemployment
over the time period, this change in unemployment corresponds only to about one-third of the increase observed
during the Great Recession. For comparison with the employment to population ratio, if all twenty million newly
unemployed people as measured by the decline in the employment to population ratio were counted in the
unemployment rate, we would have found an increase in the unemployment rate from 4.2% to 16.4%, the highest
level since 1939.
The reason for the discrepancy between the two is that many of the newly non-employed people are
reporting that they are not actively looking for work, so they do not count as unemployed but rather as exiting the
labor force. Consistent with this, we find an extraordinary decline in the adjusted labor force participation rate,
from 64.2% to 56.8%. For comparison, Panel C of Figure 2 plots the historical evolution of the labor force
participation rate over the last 15 years which includes a historically large decline in participation between 2008
and 2016 of 3 percentage points. Even this cumulatively large decline in participation over an eight year period is
dwarfed by the historic decline in participation that we document.
Figure 3 plots state-level changes in unemployment, labor force participation and employment to
population ratios around the time of the covid-19 crisis. The forty-five degree lines indicate no change. When
looking at labor force participation rates and employment-to-population ratios, we see that almost all states fall
4
below the 45 degree line, indicating the declines in each variable are widespread throughout the country and
relatively homogenous in size. The change in unemployment across states is noisier, due in part to higher
measurement error in measuring unemployment rates at local levels, but also indicates geographically dispersed
increases in unemployment. Still, the pattern that comes out from state-level variation is one of broad-based
declines in both employment and labor force participation across the country.
How unusual are these patterns? We have already seen that the size of the changes in each variable is
exceptional, at least for employment and changes in labor force participation. What about their simultaneous
changes? Figure 4 illustrates the historical comovement of the employment to population ratio with unemployment
in Panel A and with labor force participation in Panel B. Panel A shows that, historically, unemployment and
employment are very strongly negatively related. Within short periods, movements in one are reflected almost
perfectly in the other as workers move from employment to unemployment and back. Slow-moving demographics
cause the relationship to gradually change over time, as can be seen by decadal shifts in the curve, but short-run
movements are close to linear. The change that we document since the covid-19 crisis jumps out: we see an
enormous change in the employment-to-population ratio with a much smaller change in unemployment that would
have typically been expected. This pattern is therefore qualitatively different from the historical experience of U.S.
labor markets, even after taking into account the size of the changes.
We find a less unusual pattern relative to historical experience when looking at the change in labor force
participation and employment to population ratios in Panel B. There, we see that the two tend to commove
positively and closely on average: periods when employment growth is strong are also periods during which more
people are entering the labor force. In that sense, the simultaneous decline in employment and labor force
participation is mostly unusual because of the size of the changes. Still, the drop in labor force participation appears
large relative to the historical experience given the size of the decline in employment, which is consistent with the
smaller than normal increase in unemployment.
Why do so many unemployed choose not to look for work? Both surveys included a question asking those
who said they were neither working nor looking for a job to select among possible answers why this was the case.
The results for both waves are presented in Table 2. Prior to the crisis, most respondents out of the labor force
claimed that it was because they were retired, disabled, homemakers, raising children, students, or did not need to
work. Only 1.6% of those out of the labor force were claiming that they could not find a job as one of their reasons
for not searching. At the height of the covid-19 crisis with a much larger number of people now out of the labor
force, we see corresponding declines in the share of homemakers, those raising children and the disabled. However,
we see a large increase in those who claim to be retired, going from 53% to 60%. This makes early retirement a
major force in accounting for the decline in the labor-force participation. Given that the age distribution of the two
surveys is comparable, this suggests that the onset of the covid-19 crisis led to a wave of earlier than planned
retirements. With the high sensitivity of seniors to the covid-19 virus, this may reflect in part a decision to either
leave employment earlier than planned due to higher risks of working or a choice to not look for new employment
5
and retire after losing their work in the crisis. To see this more clearly, we exploit the panel dimension of the survey
and identify respondents who were out of the labor force in April 2020 but in the labor force in January 2020.
Column (3) of Table 2 presents reasons for being out of the labor force reported by those who were employed in
January. Of those, 28% report that they are now out of the labor force because of retirement. Similarly, of those
who were unemployed in January and out of the labor force in April (column 4), 21% report that it is because they
retired. And while 9% of those going from employment to out of the labor force between January and April report
that they are on break from working (as might be the case for some affected by temporary work closures due to
covid-19), the equivalent proportion is 8% for those going from unemployment to out of the labor force, suggesting
that these breaks are voluntary ones, not ones due to temporary work closings. In short, these results point to an
unusual rise in the share of retirements accounting for the exceptional decline in labor force participation during
this time period.
IV Summary
It is still very early on in the covid-19 crisis, but preliminary indicators point toward catastrophic declines in
employment. Our surveys provide additional evidence on this decline in employment, pointing to a 20 million
decline in the number of employed workers. Most strikingly, we find a much less than proportional increase in
unemployment, indicating that most of these newly unemployed workers are not looking for new work. Hopefully
this reflects a transitory characteristic as these individuals face shelters-at-home and few work opportunities. But
the wave of early retirements that we document suggests that more permanent changes may already be taking
place.
References
Allcott, Hunt, Levi Boxell, Jacob Conway, Matthew Gentzkow, Michael Thaler, and David Yang, 2020.
“Polarization and Public Health: Partisan Differences in Social Distancing during the Coronavirus
Pandemic,” Working Paper.
Alvarez, Fernando, David Argente, and Francesco Lippi (2020) “A Simple Planning Problem for
COVID-19 Lockdown” University of Chicago, Becker Friedman Institute for Economic Working
Paper No. 2020-34.
Atkeson, Andrew (2020) “What Will Be the Economic Impact of COVID-19 in the US? Rough Estimates
of Disease Scenarios” NBER Working Paper w26867.
Barrios, John and Yael Hochberg, 2020. “Risk Perception Through the Lens of Politics in the Time of
the COVID-19 Pandemic,” Working Paper.
6
Barro, Robert J., José F. Ursúa, and Joanna Wang (2020) “The Coronavirus and the Great Influenza
Pandemic: Lessons from the ‘Spanish Flu’ for the Coronavirus’s Potential Effects on Mortality
and Economic Activity” NBER Working Paper w26866.
Binder, Carola, 2015. “Coronavirus Fears and Macroeconomic Expectations” Working Paper.
Coibion, Olivier, Dimitris Georgarakos, Yuriy Gorodnichenko and Michael Weber, 2020a. “Forward
Guidance and Household Expectations,” NBER Working Paper w26778.
Coibion, Olivier, Yuriy Gorodnichenko and Michael Weber, 2019. “Monetary Policy Communications
and their Effects on Household Inflation Expectations,” NBER Working Paper w25482.
D’Acunto, Francesco, Ulrike Malmendier, Juan Ospina, and Michael Weber, 2020a. “Exposure to Daily
Price Changes and Inflation Expectations,” NBER Working Paper w26237.
D’Acunto, Francesco, Ulrike Malmendier, Juan Ospina, and Michael Weber, 2020b. “Gender Roles and
the Gender Expectations Gap,” NBER Working Paper w26837.
Dietrich, Alexander M., Keith Kuester, Gernot J. Müller, and Raphael S. Schoenle (2020) “News and
Uncertainty about COVID-19: Survey Evidence and Short-Run Economic Impact,” Becker
Friedman Institute for Economic White Paper.
Mongey, Simon and Alex Weinberg, 2020. “Characteristics of Workers in Low Work-From-Home and
High Personal-Proximity Occupations,” Becker Friedman Institute for Economic White Paper.
Eichenbaum, Martin S., Sergio Rebelo, and Mathias Trabandt (2020) “The Macroeconomics of
Epidemics” NBER Working Paper w26882.
Fetzer, Thiemo, Lukas Hensel, Johannes Hermle, and Christopher Roth, 2020. “Perceptions of
Contagiousness and Mortality of COVID-19 Shape Economic Sentiment,” Working Paper.
Guerrieri, Veronica, Guido Lorenzoni, Ludwig Straub, and Ivan Werning, 2020. “Macroeconomic
Implications of COVID-19: Can Negative Supply Shocks Cause Demand Shortages?” NBER
Working Paper w26918.
7
Figure 1: Pre-crisis employment statistics at the state level.
Panel A. Unemployment rate Panel B. Labor force participation rate
15 WV 90
RI SD
MS
CA
VT
NHFL GA AR CT 80 ND
AL OR PA
10 VA
TX WA
AZ DC CT
IL
MO
NCNJ LA
Nielsen Nielsen
ID ME
IA KSTN MI
NY NM DE
SC NV OH 70 NJ MD DC
UT OK IL
MNSD GA TX VAWI
KSMACO
COMAIN DEKY MS LA NY MI
ORCAPA MN
NH
WI INWY NE
5 MD RI UT
MT NC MO
NE 60 OK ME
AZ
OH IA
WV ARFL
KY NV VT
ND TN WA
WY NM
AL
SC
0 50 MTID
2 3 4 5 6 55 60 65 70 75
BLS BLS
Panel C. Employment to population ratio
SD
80 ND
70 DE
CT MD
Nielsen NJ WY WIMADC
ILIN KS CO NE
MN Linear fit
60 LA NY MIGATX VA
UT
ORPA NH
MS CA
NC
ME
OH MO IA
KY OK
AZ RI
AR NV
FL TN
50 WV NMSC WAVT
AL MT
ID
50 55 60 65 70
BLS
Notes: Bureau of Labor Statistics (BLS) data refer to the January-February 2020 period. Nielsen data are for the pre-crisis wave
of the survey.
8
Figure 2. Time series of key employment statistics.
Panel A: Unemployment rate
percent
BLS
Nielsen
3 4 5 6 7 8 9 10 11
2020M3
2020M2
2006 2008 2010 2012 2014 2016 2018 2020
Panel B: Employment to population ratio
percent 2020M2
52 54 56 58 60 62 64
2020M3
2006 2008 2010 2012 2014 2016 2018 2020
Panel C: Labor force participation rate
66
2020M2
64
percent
60 62
58
2020M3
56
2006 2008 2010 2012 2014 2016 2018 2020
Notes: Each panel plots monthly time series of an employment statistic. The black, solid line shows data from the Bureau of
Labor Statistics (BLS). The red, solid line with circles shows the corresponding values from the Nielsen survey. 2020M2 are the
values from the Nielsen pre-crisis survey. 2020M3 are the values from the Nielsen crisis wave.
9
Figure 3. Adjusted employment statistics by state, pre-crisis vs. crisis levels, Nielsen survey.
Panel A. Unemployment rate (adj.) Panel B. Labor force participation rate (adj.)
20 100
DC WV
15 AR WY
CO
WA 80 SD
AZ NE DC
ND
crisis period crisis period
CAMS
10 NV
DEOKME NH LAILNJ
UTNH
IN VA FL RI MA MDDECT
MN
NY
TX 60 TNV MS
PA WI
CO
TX
MI
ND WIKY NM
IL
NJ
NC
NY
LA
OR VT KY
WVIA IN
OK
OH
CA
ORGA
CT ID NC VA
FL ME KS
5 NEMD SD KS MI
TN GA MO
MNOHIA AL ALWA AZ RI
SC
MT UT PA AR
MO 40 MT VT
0 MA NM
SC
WY ID
-5 20
-5 0 5 10 50 60 70 80 90
pre-crisis period pre-crisis period
Panel C. Employment to population ratio (adj.)
100
WY
80
SD
crisis period
NE ND
UT MA
LAMN CT
60 ID TN IA NH
J
NYIL WI
PA DCMD
DE
MI
GA
TX 45-degree line
NV OH
KY
MOMSOR
CA IN
KS
SCFL OK NC CO
AL ME VA
RI
WV AZ
40 MTWA
VT AR
NM
20
50 60 70 80 90
pre-crisis period
Notes: Nielsen employment statistics are adjusted to match average pre-crisis levels observed in the official data compiled by
the Bureau of Labor Statistics (BLS). The adjustment is described in equations (1)-(2).
10
Figure 4. Historical comovement of key employment statistics.
Panel A: Employment-to-population ratio vs Unemployment rate.
11
BLS: 1970s
BLS: 1980s
10 BLS: 1990s
BLS: 2000s
BLS: 2010
9 Nielsen: COVID19
Unemployment rate
6 7 8
2020M3
5
2020M2
4
3
52 54 56 58 60 62 64
Employment to population ratio
Panel B: Employment-to-population ratio vs Labor force participation rate.
Labor force participation rate
BLS: 1970s
BLS: 1980s
BLS: 1990s
BLS: 2000s
BLS: 2010
Nielsen: COVID19
2020M2
56 57 58 59 60 61 62 63 64 65 66 67 68 2020M3
52 54 56 58 60 62 64
Employment to population ratio
Notes: Each panel shows comovement (by decade) of key official employment statistics compiled by the Bureau of Labor
Statistics (BLS) as well as employment statistics based on Nielsen surveys (red circles with dates). 2020M2 are the values from
the Nielsen pre-crisis survey. 2020M3 are the values from the Nielsen crisis wave.
11
Table 1. Employment statistics at the state and aggregate levels, Nielsen survey.
Nielsen
BLS
Raw moments Adjusted moments
pre-crisis
pre-crisis crisis pre-crisis crisis
(1) (2) (3) (4) (5)
Panel A. State-level data (average)
Unemployment rate 3.52 7.84 10.93 3.50 6.30
(0.84) (2.80) (5.07) (2.55) (4.61)
Labor force participation rate 63.91 62.89 56.94 63.97 57.20
(3.94) (7.19) (8.58) (8.17) (9.75)
Employment to population ratio 61.67 58.00 50.77 61.74 54.11
(4.03) (7.27) (8.63) (7.68) (9.12)
Panel B. Aggregate level
Unemployment rate 8.62 10.95 4.21 6.32
[0.25] [0.43] [0.23] [0.39]
Labor force participation rate 63.4 63.13 56.57 64.23 56.79
[0.33] [0.51] [0.38] [0.58]
Employment to population ratio 61.1 57.68 50.38 59.94 52.22
[0.34] [0.52] [0.36] [0.54]
Notes: Panel A report statistics across states. Averages are in the top row of each subsection and standard deviations are in the bottom rows (in parentheses) of each
subsection. Panel B reports statistics for the aggregate level. Point predictions are reported in the top row of each subsection and standard errors are reported in the bottom
rows (is in square brackets) of each subsection. Column (1) reports pre-crisis data from the Bureau of Labor Statistics (BLS), average values for January-February 2020.
These data are based on the Current Population Survey (CPS). Columns (2) and (3) report moments for the Nielsen survey unadjusted for possible differences in design
between the CPS and Nielsen surveys. Columns (4) and (5) report moments for the Nielsen survey adjusted for possible differences in design between the CPS and Nielsen
surveys. Adjustment is based on state-level regressions and is given by equations (1)-(2).
12
Table 2. Reasons for not looking for a job (for those who do not have a job).
Share of people choosing a reason
All people Employed pre-crisis, Unemployed pre-crisis,
Reason out of labor force in crisis out of labor force in crisis
Pre-crisis Crisis
(1) (2) (3) (4)
Homemaker 0.204 0.155 0.158 0.244
Raising children 0.125 0.082 0.115 0.222
Student 0.023 0.018 0.050 0.019
Retiree 0.527 0.595 0.277 0.210
Disabled, health issues 0.297 0.261 0.142 0.144
Couldn’t find a job 0.016 0.018 0.061 0.133
On break 0.012 0.011 0.092 0.080
No financial need 0.049 0.037 0.039 0.029
Notes: the Nielsen survey question is “Here are a number of possible reasons why people who are not working choose not to look for work. Please select
all that apply to you.” This question is asked for people who do not have a job and are not looking for a job.
13
File and source
- File
- labor-markets-during-the-covid-19-crisis-a-preliminary-view.pdf
- Size
- 255,489 bytes
- SHA-256
- 44952efd3421ff2a81ad1362fd32ca0650fb3f759d0b48c3a8e39947dd897691
- Original
- www.nber.org