Earnings Losses And The Role Of The Welfare State During The Scp Covid Scp 19 Pandemic Evidence From Sweden
Summary
A research article titled Earnings Losses and the Role of the Welfare State During the COVID-19 Pandemic: Evidence From Sweden, published in the Review of Income and Wealth, Series 70, Number 4, December 2024, by researchers affiliated with IFAU, Uppsala University and IFN. Using a difference-in-differences approach and population-wide monthly data on earnings and government transfers for 5.7 million individuals, it studies how existing transfers and new pandemic measures offset income losses in Sweden. The article reports that labor earnings dropped by 2.7 percent in 2020 and that existing transfers and new pandemic measures reduced earnings losses to 1.5 percent. It reports that transfers replaced 43 percent of earnings losses in 2020, with the largest losses at the lower end of the income distribution. It closes with references and a supporting information note.
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
Cite this document
“Earnings Losses And The Role Of The Welfare State During The Scp Covid Scp 19 Pandemic Evidence From Sweden.”
Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/earnings-losses-and-the-role-of-the-welfare-state-during-the-scp-covid-e366319b69/
, SHA-256 34d3c9087ad234684726fbf56cf479649495d0bad06914db5363fa53f66ed01d.Full text
bs_bs_banner
Review of Income and Wealth
Series 70, Number 4, December 2024
DOI: 10.1111/roiw.12670
EARNINGS LOSSES AND THE ROLE OF THE WELFARE STATE
DURING THE COVID-19 PANDEMIC: EVIDENCE FROM SWEDEN
BY ADRIAN ADERMON
Uppsala Center for Labour Studies (UCLS), Uppsala Center for Fiscal Studies (UCFS), Institute for
Evaluation of Labour Market and Education Policy (IFAU), Uppsala University
LISA LAUN
IFAU and UCLS
PATRIK LIND
IFAU
MARTIN OLSSON
Research Institute of Industrial Economics (IFN) and IFAU
JAN SAUERMANN*
IFAU, UCLS, Institute of Labor Economics (IZA)
AND
ANNA SJÖGREN
IFAU and UCLS
Many governments introduced temporary adjustments to counter the economic and health conse-
quences of the COVID-19 pandemic. We study the importance of already-existing government transfers
and new pandemic measures to mitigate individual income losses during the onset of the pandemic in
Sweden using a difference-in-differences approach and population-wide data on monthly earnings and
government transfer payments. We find that labor earnings dropped by 2.7 percent in 2020. Existing
transfers and new pandemic measures reduced earnings losses to 1.5 percent. These average effects mask
considerable differences in earnings losses, which were, by and large, evened out by existing transfers
and new pandemic measures.
JEL Codes: C23, D31, E24, H12, H20
Keywords: COVID-19, income inequality, government transfers, short-time work
Note: We thank Adam Altmejd, Evelina Björkegren, Maria Hemström Hemmingsson, Lena
Hensvik, Oskar Nordström Skans, and two anonymous reviewers for comments and feedback.
This research was conducted within SWECOV—the Swedish Register-Based Research Program on
COVID-19—hosted by Stockholm University and financed by Riksbankens Jubileumsfond. We grate-
fully acknowledge financial support from the Swedish Research Council (grant number 2022-06315).
Martin Olsson also gratefully acknowledges financial support from the Marianne and Marcus
Wallenberg Foundation (2020-0049).
*Correspondence to: Jan Sauermann, Institute for Evaluation of Labour Market and Education
Policy (IFAU), Box 513, 751 20, Uppsala, Sweden (jan.sauermann@ifau.uu.se).
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-
NoDerivs License, which permits use and distribution in any medium, provided the original work is
properly cited, the use is non-commercial and no modifications or adaptations are made.
981
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
1. INTRODUCTION
In addition to being the most acute public health crisis in recent decades,
the COVID-19 pandemic perturbed global supply chains (Bonadio et al., 2021)
and drastically reduced economic activity. This led to rapid and substantial GDP
losses in the second quarter of 2020: up to 9.1 percent in the USA, 7.6 percent in
Sweden, and 11.4 percent in the European Union (EU) overall, followed by a fast
recovery in the third quarter of 2021 (Milesi-Ferretti, 2021; OECD, 2020). Judging
that existing transfer systems, which are designed to mitigate income risks from
adverse events such as unemployment (UI) and illness, would be inadequate in this
pandemic, many governments saw a need to dampen the expected negative impact
for individuals and businesses.
Similar to other countries, Sweden introduced new and substantial firm sup-
port measures, such as short-time work (STW) allowance, to preserve employment
relationships and save jobs (Giupponi & Landais, 2023), in addition to increasing
the generosity and reach of government transfers to individuals. Understanding
to what extent incomes were protected by the already-existing transfer system
during the pandemic and the role played by these new pandemic measures is of key
interest for future policy development and, in particular, in building resilience to
future economic and other crises. Earlier studies have highlighted the consequences
of the pandemic itself for economic inequality (Angelov & Waldenström, 2023a;
Clark et al., 2021; Stantcheva, 2022) and studied the role of existing transfers
and new pandemic measures through cross-country simulation models (Almeida
et al., 2021; Cantó et al., 2022). There is, however, far less empirical evidence on
the relative importance of already-existing transfers and new pandemic measures
and their ability to compensate earnings losses across the distribution and reduce
the unequal impact of the pandemic on individual incomes.
The purpose of this paper is to assess how the COVID-19 pandemic affected
labor earnings across the income distribution and for various groups on the labor
market and to analyze the degree to which already-existing government transfer
systems and new pandemic measures compensated for these earnings losses. To
this end, we use monthly administrative data on labor earnings, social insurance
benefits, and means-tested income transfers for the Swedish working-age pop-
ulation, which includes 5.7 million individuals. We also use monthly individual
employee-level data on STW allowance transfers directed to firms. We focus pri-
marily on the first year of the pandemic, that is, 2020, when economic uncertainty
was acute and the government acted decisively to alleviate the crisis. To estimate
the causal effects of the pandemic, we use a difference-in-differences (DiD) model
that compares the monthly changes in various incomes in 2020 to those in previ-
ous years. This estimation framework results in comparable treatment and control
groups (see, e.g., Hensvik et al., 2021) and controls for seasonal (monthly) variation
in outcome variables. To shed light on the economic recovery from the pandemic,
we also provide some suggestive results for 2021.
Our main findings can be summarized as follows: first, the pandemic decreased
average labor earnings by 2.7 percent during 2020. These earnings losses, however,
already include the STW allowance, which essentially kept many individuals in
employment with only minor earnings reductions. Our estimates of the increase in
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
982
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
STW allowance suggest that net earnings losses of STW allowance amounted to
4.5 percent. Because all STW allowance recipients are unlikely to have lost their
jobs without the STW allowance, this represents an upper bound of earnings losses
in the absence of STW allowance.
Second, the already-existing transfer system and the new pandemic measures
together replaced 43 percent of earnings losses in 2020, primarily through UI and
sickness benefits. We further show that the new pandemic measures were almost
as important for replacing earnings losses as the already-existing transfer system.
Disaggregating the different types of transfers, we find that, among pandemic
measures, changes to UI benefits and sickness benefits affected incomes in 2020 the
most.
Third, we analyze which groups suffered the most from earnings losses in
2020. We find that labor earnings at the lower end of the pre-pandemic income
distribution were most heavily affected by the pandemic and that losses were par-
ticularly large for young individuals, those with low education, and foreign-born
individuals and, most notably, for those working in the hospitality sector before
the pandemic. While STW allowance limited earnings losses in a similar fashion
across the disposable income distribution, government transfers to individuals
compensated those with the largest relative earnings losses the most and were
largely able to even out differences across the income distribution.
While our main analysis focuses on the effects of the pandemic during 2020,
we also provide some evidence for 2021. This analysis suggests that earnings losses
in 2021, compared to pre-pandemic levels, were substantially smaller than those
in 2020. Transfer payments, however, remained on a high level throughout 2021,
leading to higher overall total income compared to pre-pandemic levels.
This paper contributes to the literature on the effects of the COVID-19
pandemic on inequality. Most closely, our paper complements Angelov & Walden-
ström’s (2023a), henceforth AW, who use monthly tax data from Sweden to study
how labor earnings and earnings inequality were affected by the pandemic. In
addition to documenting an increase in inequality in labor earnings, they estimate a
DiD model and find that the pandemic decreased pretax earnings by, on average, 3.8
percent. They also study two COVID-19 policies that were directed at firms—STW
allowance and reorientation support—and use simulations that suggest that overall
inequality would have increased two to three times more absent these policies.
While there is some overlap between their paper and ours, there are also several
important differences. First, the focus is different. AW are mainly interested in the
impact of the first year of the pandemic on earnings inequality among workers.
While we also study the effect of the pandemic on labor earnings, our main con-
tribution is to document the effect of the pandemic on a broader income measure
for the entire working-age population. By including government transfers, we can
show which parts of the economic safety net were most important for different
groups. We thus set out to document the full set of pandemic policy changes to the
public safety net, including STW allowance. In our study, we also provide evidence
on the effects of the pandemic for 2021.
Second, our data sources differ. For the monthly data, AW are restricted to
taxable earnings for individuals who had at least 1 month of nonzero earnings
between January 2019 and March 2021. While AW define zero earnings as an
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
983
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
indirect measure of UI, we observe the full population, including the unemployed
and those out of the labor force. This likely explains why we find a slightly smaller
average effect on labor earnings, 2.7 percent compared to AW’s 3.8 percent. We
further observe a whole range of additional income sources, including different
types of UI benefits, sickness benefits, parental benefits, and income support on a
monthly basis. The great detail of our data allows us to decompose government
transfers into existing transfers and new pandemic measures. In the annual income
data, we observe both taxable and nontaxable transfers, while AW observe only
taxable transfers.
Third, our analysis adds to the understanding of the economic consequences
of the pandemic for different groups by providing results for a wider range of
subgroups and across the full income distribution, including the very poorest
groups. In terms of background characteristics, AW use age, gender, and working
in the public or private sector. We provide evidence by education, industry, family
structure (children in a household, being married or cohabiting), being foreign
born, and features of the occupation relevant to the pandemic (ability to work from
home, working in a contact profession). There is also a small literature studying
inequalities along other work-related dimensions using Swedish microdata. Campa
et al. (2021) use individual-level data from the Swedish Public Employment Service
(PES) to show that the pandemic initially led to increased UI among young and
foreign-born workers. Eliason (2021) finds similar patterns using aggregate data.
Sjögren et al. (2021) use aggregate Labor Force Survey data to show that mothers
with small children were harder hit than other parents.
Internationally, a number of studies have analyzed the effects of the pandemic
on incomes and income inequality (see Stantcheva, 2022, for a recent overview).
Using longitudinal survey data for Australia, Li et al. (2021) find that additional
wage subsidies offset the negative effects of increasing UI for income inequality.
For the USA, Han et al. (2020) show that policy measures were effective in reduc-
ing poverty during the start of the pandemic. In part, this is driven by generous
UI insurance benefits often exceeding lost wages (Ganong et al., 2020; Matias &
Forsythe, 2023) and effectively reducing earnings losses for those in UI (Larrimore
et al., 2022). Bargain et al. (2023), using social security data from Ecuador, find that
while earnings fell for everyone in 2020, they recovered in 2021 for everyone except
the top 1 percent of earners.
A large literature has used microsimulation methods to quantify the impact
of the pandemic on individual incomes and to analyze to what extent government
policies manage to cushion the fall. These papers typically combine models of
national tax and transfer systems with representative survey data from before
the pandemic, allowing researchers to simulate the effects of macroeconomic
changes and policy changes on the distribution of incomes. This approach has
the advantage that it can be deployed very quickly since it does not require the
availability of up-to-date microdata. However, a drawback is that it relies heavily on
modeling assumptions and macroeconomic forecasts. Furthermore, these models
are typically static and do not account for behavioral changes. This contrasts with
our paper, which uses detailed administrative data to capture actual outcomes
before and during the pandemic. Furthermore, while microsimulation studies tend
to cover existing tax-and-transfer systems well (including existing STW schemes),
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
984
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
there are some limitations in their coverage of new COVID-19 policies (see, e.g.,
Almeida et al., 2021; Cantó et al., 2022).
In general, these microsimulation studies have found that existing government
transfers and new pandemic measures could be expected to reduce income losses by
half or more in the EU and the UK. For the EU as a whole, Almeida et al. (2021)
estimate that disposable income would have fallen by over 9 percent absent dis-
cretionary fiscal policy changes but that pandemic measures reduced this fall by
half. Christl et al. (2021) estimate large drops in market incomes, hitting poorer
households particularly hard. However, almost three-fourths of this drop were
absorbed by tax-and-transfer systems and around half came from discretionary
policy measures. Cantó et al. (2022) estimate that household incomes would have
dropped as much as 15 to 25 percent absent policy changes in Belgium, Italy,
Spain, and the UK but that fiscal policy dampened most of this, resulting in actual
drops of 4 to 8 percent. In Germany and Ireland, COVID-19 policies even resulted
in slightly increased incomes for people at the bottom of the income distribution
(Bruckmeier et al., 2021; O’Donoghue et al., 2023). In Finland, the pandemic
resulted in a 4.5 percent drop in market incomes, while disposable incomes fell
only by 1.8 percent (Kyyrä et al., 2021). In contrast to these studies, we find that
earnings losses are smaller (2.7 percent) but also that already-existing government
transfers and pandemic measures helped dampen these losses by 43 percent.
Other microsimulation studies for European countries that reach similar con-
clusions include Brewer & Tasseva (2021) for the UK; O’Donoghue et al. (2020) for
Ireland; Brunori et al. (2021), Figari & Fiorio (2020), and Carta & Philippis (2021)
for Italy; Christl et al. (2023) for Germany; and Christl et al. (2022) for Austria. In
Latin America, existing government transfer systems tended to protect the poorest,
while discretionary COVID-19 policies tended to cushion the income drops higher
up in the income distribution. This resulted in middle-income households experi-
encing the largest income losses (Avellaneda et al., 2021; Lustig et al., 2021). In
Africa, where many work in the informal sector, the pandemic led to increases in
both poverty and income inequality. Neither existing government transfer systems
nor discretionary policies did much to counteract this (Lastunen et al., 2021).
A third strand of the literature uses real-time surveys that were conducted
during the pandemic to analyze the consequences of the pandemic. Unlike our
study, these papers do not estimate the causal effects of the pandemic but rather
provide valuable descriptive evidence on the economic consequences. Using survey
data from France, Germany, Italy, Spain, and Sweden, Menta (2021) finds that
poverty increased sharply during the spring of 2020 and then decreased during the
summer, with Italy being the most affected and France the least. Young individuals
and women were the most affected. Using the same data, Clark et al. (2021) show
a similar time pattern of household disposable income inequality. Adams-Prassl
et al. (2020) find that around one-fifth of workers in the UK and USA lost their
jobs during spring 2020, compared to only around 5% in Germany. Braband
et al. (2022) highlight that, in Germany, part-time workers, low-educated workers,
and low-income workers were particularly affected in terms of UI and income.
In the following section, we present a detailed account of the most important
transfer types and pandemic measures that affected the working-age population
during the pandemic. In Section 3, we provide information on the data sources, the
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
985
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
estimation framework, and descriptive statistics. Section 4 presents the estimation
results. In Section 5, we extend our analysis up until 2021. Section 6 concludes.
2. BACKGROUND
Sweden has an extensive government transfer system with universal sick-
ness and parental benefits at basic and income-related levels. UI insurance is
income-related for individuals who fulfill two conditions—a sufficiently long
voluntary membership in an UI fund and a work requirement—and has a basic
level for those who only fulfill a work requirement (Landais et al., 2021). Social
assistance and housing allowance are means-tested basic income support programs
to guarantee a minimum level of living and housing standard for households with
insufficient means—earnings, benefits, or assets—to support themselves.
As the COVID-19 pandemic hit Sweden in March 2020, the government intro-
duced several pandemic measures—in the form of both changes to already-existing
government transfers and the introduction of entirely new transfers. Early in the
pandemic, an STW allowance scheme was introduced to protect jobs and main-
tain employer–employee links. To increase insurance coverage, membership and
work requirements in the UI insurance were eased, and to incentivize sick leave,
the waiting-day deduction in the sickness insurance was reimbursed by the govern-
ment. Moreover, in July 2020, the housing allowance was raised to strengthen poor
households and prevent evictions of families.
In the following subsections, we first describe how COVID-19 affected both
health outcomes and the Swedish economy. Thereafter, we present in detail both
existing transfers and pandemic measures analyzed in this study: the introduction of
the STW allowance scheme, UI insurance, sickness insurance, parental benefits, and
basic income support. Table A1 in the Online Appendix lists details on all pandemic
measures introduced during the pandemic.
2.1. How Was Sweden Affected During the First Year of the Pandemic?
In many dimensions, the Swedish experience of the COVID-19 pandemic
represents what happened in an average European country. The first wave of
SARS-CoV-2 hit Sweden in March 2020. As a reaction to this first wave, there
was a rapid decrease in individual mobility and economic activity. In contrast to
many European countries, Sweden did not implement strict lockdowns. Instead,
the Swedish Public Health Agency gave general advice and recommendations that
workers should work from home if possible, remote learning should be introduced
in high schools and universities, and people should avoid traveling and having
social contacts outside the family. In addition, public gatherings and events were
limited, and visitors were not allowed in hospitals and residential care facilities for
older people. Distance learning was also introduced in upper secondary schools
and higher education while primary and lower secondary schools mostly remained
open during 2020.
Even though Sweden did not introduce severe pandemic restrictions, the pan-
demic affected the labor market. The employment rate decreased in 2020 for the
first time since the financial crisis and did not fully return to pre-pandemic levels
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
986
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
until the end of 2020. Similarly to its Scandinavian neighbors, the Swedish UI rate
substantially increased in spring 2020 (Juranek et al., 2021) and started decreasing
in June 2021. During the first pandemic wave in the spring of 2020, job vacancies
dropped by 40 percent (Hensvik et al., 2021).
2.2. Government Transfers Before and During the Pandemic
In the following subsections, we present the regulations regarding different
groups of government transfers before the pandemic and, if applicable, its changes
during the pandemic. Table A1 in the Online Appendix provides a detailed overview
of all changes.
Short-Time Work
As it was clear that otherwise healthy and profitable businesses would be
severely affected by the pandemic, an STW allowance scheme was rapidly launched
in response to the massive increase in advance notices of layoffs in March 2020.
The aim was to protect jobs and maintain links between employers and employees.
The STW allowance scheme implied that employers could temporarily reduce
employees’ working hours by 20, 40, 60, or 80 percent. Of the corresponding earn-
ings losses, 75 percent were covered by the government, 10 percent were covered by
the employer, and 15 percent were borne by the worker. This scheme implied that
firms could reduce their wage costs by up to 72 percent while still maintaining the
employment of their workers. Workers on STW allowance worked shorter hours
but retained much of their pay—for example, a fulltime worker with working hours
reduced by 80 percent lost only 12 percent of their pretax earnings.
Unemployment Insurance
Unemployed workers in Sweden are entitled to basic UI benefits while search-
ing for a new job, conditional on fulfilling a work condition of having worked during
6 of the past 12 months and being registered as unemployed with the Swedish PES.
The pre-pandemic basic UI benefit was 365 SEK per day, corresponding to
24 percent of the median full-time equivalent (FTE) daily wage in 2019.1 This
amount was reduced proportionally for former part-time workers. Unemployed
job seekers who have been members of a UI fund (a-kassan) for at least a year also
fulfill the membership condition that qualifies them for income-related benefits.
Fund membership coverage varies across industries and is generally low for not
only young workers but also for workers in service jobs and industries with low UI
risk or large shares of temporary workers. Before the pandemic, the income-related
benefit replaced 80 percent of the previous labor earnings up to a ceiling of 910
SEK per day during the first 100 days of UI. From day 101, the ceiling is reduced
1 In February 2020, 10 SEK correspond to 1.04 USD and 0.94 EUR, respectively. 1 USD (1 EUR)
converted to 9.66 SEK (10.54 SEK). In 2019, the median full-time equivalent monthly wage was 31,700
SEK and the average number of working days per month was 21 (www.arbetstimmarpermanad.se and
https://www.statistikdatabasen.scb.se, accessed 2022-09-08). Given these numbers, the median full-time
equivalent daily wage was 1,510 SEK.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
987
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
to 760 SEK, and from day 201 onwards the replacement rate is lowered to 70
percent of previous earnings. Parents with children under 18 receive up to 450 days.
Overall, about 10 percent of the total UI claims are basic ones.
Several reforms to the UI insurance system were implemented in the first half
of 2020 to limit the adverse consequences of job loss for income and consumption.
In April, the basic benefit was raised from 365 to 510 SEK per day (from 24 to 34
percent of the median FTE daily wage), and a minimum compensation of 255 SEK
per day was introduced (17 percent of the median FTE wage). For income-related
UI insurance, the ceiling during the first 100 days was raised from 910 to 1,200 SEK
per day (from 60 to 80 percent of the median wage). In late June, the ceiling after
the first 100 days in the income-related benefit was increased from 760 to 1,000 SEK
per day (from 50 to 66 percent of the median wage).
In addition to increased benefit levels, the requirements for fulfilling the
membership condition and the work condition were eased. From March 2020
onward, unemployed individuals could receive income-related UI benefits after
already being a member of a UI fund for 3 (instead of 12) months. Before the pan-
demic, qualifying for income-related benefits was conditional on having worked
either (1) for 80 hours per calendar month for 6 of the last 12 months or (2) for
480 hours during a continuous 6-month period during the last year and for at least
50 hours during each calendar month. In late March 2020, the work requirements
were relaxed to 60 hours per month in the first case and 420 hours in total and
40 hours per month in the second case. In addition, the waiting period of 6 days for
unemployed workers to become eligible for UI benefits was abolished.
Unemployed job seekers who participate in an active labor market policy
(ALMP) program can receive a benefit called activity support. When participat-
ing in ALMP programs, individuals receive benefits of the same kind as those
unemployed job seekers without ALMP participation.
Sickness Insurance
Sweden has a universal, publicly administered sickness insurance. The central
part of the insurance is sickness benefits compensating earnings losses, with an 80
percent replacement rate up to a ceiling, for employees whose work capacity is tem-
porarily reduced due to sickness. A 1-day waiting period implies that the worker is
not compensated for earnings losses on the first day of the sickness spell. On days
2–14 of the sickness spell, the employer period, the worker is compensated by the
employer in the form of sick pay. Thereafter, the Social Insurance Agency (SIA) pays
a sickness benefit. A doctor’s certificate is required for sick spells longer than 7 days.
In March 2020, the 1-day waiting period in the sickness insurance became
reimbursable by the SIA to encourage workers to stay home from work at the
slightest symptom of illness. Initially, the maximum payment for the first sick day
was set to 700 SEK, but it increased to 804 SEK in April 2020 (from 46 to 53
percent of the median wage). To further limit the spread of COVID-19 and ease
the burden on the healthcare system, the requirement to have a doctor’s certificate
was postponed from day 8 to day 21 and later abolished altogether (see Table A1 in
the Online Appendix for details). Moreover, during the pandemic, employers were
compensated for their sick pay costs by the government. The full cost was covered
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
988
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
from April to July 2020, after which compensation was paid for above-normal
costs according to a fixed schedule.
Because COVID-19 was classified as a public health hazard in February 2020,
workers were also eligible for disease carrier’s benefits, with an 80 percent replace-
ment rate for earnings losses up to 804 SEK per day (equivalent to 53 percent of the
median wage) if diagnosed with COVID-19 or suspected of COVID-19. In the sum-
mer of 2020, to protect vulnerable groups, the disease carrier’s benefit was extended
to individuals working closely with and to relatives of persons diagnosed with med-
ical conditions, deeming them vulnerable to severe COVID-19. Furthermore, a risk
group compensation, targeting workers with these same medical conditions, who
were unable to work from home, was introduced, with the same replacement rate as
the disease carrier’s benefits.
Parental Benefits
Parents are eligible for 480 days of paid parental leave per child, which need
to be used before the child turns 12. Of those, 390 days are income related with a
replacement rate of 80 percent up to a ceiling and 90 days have a low flat rate. For
working parents of children below age 12, there is also a temporary parental benefit,
which compensates for earnings losses up to a ceiling for a maximum of 60 days per
year when caring for a sick child. Before the pandemic, temporary parental leave
when caring for a sick child required a doctor’s certificate from day 8. This was
increased to day 22 during March–October of 2020. From April 2020 on, it was also
possible for parents to take temporary parental leave during preschool or school
closures, even if their own child was not sick. Note, however, that childcare facilities
and compulsory schools (aged 6–16) generally remained open, unless hit by severe
COVID-19 outbreaks. Otherwise, there were no changes to parental benefits during
the pandemic.
Income Support
Individuals who lack the means to financially support themselves are eligible
for means-tested basic income support in the form of social assistance (SA) from
their municipality of residence. Means-testing is at the household level and requires
the depletion of any savings or other assets (owned housing, cars, etc.) before sup-
port is granted. Young adults and families with children may also be eligible for a
housing allowance from the SIA. To further support poor families and prevent the
eviction of children, an additional housing allowance was introduced in the summer
of 2020. It was targeted toward the families with children already receiving regular
housing allowance and increased the total allowance by 25 percent, summing to a
maximum of 1,335 SEK per month.
3. DATA AND EMPIRICAL STRATEGY
3.1. Data Sources and Outcome Variables
The analysis is based on several administrative data sources collected within
Stockholm University’s Swedish Register-Based Research Program on COVID-19
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
989
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
(SWECOV) program, containing information on the entire Swedish population
from 2015 onward. We analyze the impact of the COVID-19 pandemic on labor
earnings and investigate to what extent different types of government transfers cush-
ioned the potentially adverse effects on individual incomes in 2020. Although our
focus is the income replacement offered by government transfers directed to indi-
viduals, we also study the role of the STW allowance scheme introduced during the
pandemic. This is motivated by the STW allowance’s direct impact on the earn-
ings of the individual worker, who is the final recipient of STW pay. In addition to
demographic and socioeconomic information, the data include labor earnings and
all payments from the major government transfer systems directed to individuals in
Sweden, as well as STW pay.
In the main analysis, we analyze the impact of the pandemic on the following
income sources:
Labor earnings: Payments of gross labor earnings from employers to employees,
including employers’ sick pay during the 14-day employer period and STW pay
to the individual for reducing hours worked during the pandemic (see Section 2
for details).
Labor earnings excluding STW allowance: Labor earnings, as defined earlier,
excluding STW pay.
Transfers: The sum of government transfers from the major transfer sys-
tems directed to individuals: UI benefits, sickness benefits, parental benefits,
and means-tested income support. These include benefits in place before the
pandemic and benefits introduced or changed during the pandemic.
Total income: The sum of labor earnings and the aforementioned government
transfers.
Data on labor earnings are provided by the Swedish Tax Authority from the
monthly payroll tax register during 2019–2021 and the annual personal tax-return
register during 2015–2018.2 All government transfer payments are available on a
monthly basis from 2015 onward. Table A2 in the Online Appendix provides a
detailed overview of all transfers and their data sources.
We do not observe receipts of study allowance, that is, student grants and loans,
at a monthly level and therefore do not include these in the main analysis. New
enrollment in higher education increased by 13 percent in the fall of 2020 and by 28
percent among recent upper secondary school graduates.3 We also disregard some
2 The monthly earnings data do not include earnings for sole proprietors or business owners, but
results in Table A4 in the Online Appendix show that our main result holds when including these income
sources at an annual level. We also do not observe capital incomes on a monthly basis, further limiting
our ability to study sole proprietors and business owners. In Table A6 in the Online Appendix, we can,
however, show that there were limited effects overall on capital income in 2020 from an annual model
(see Section 5).
3 Swedish Higher Education Authority, https://www.uka.se/publikationer--beslut/publikationer
--beslut/statistiska-meddelanden/statistiska-meddelanden/2020-11-18-16-000-fler-sokande-utan
-tidigare-hogskolestudier.html, accessed 2022-09-08.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
990
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
transfers that mainly respond to slow-moving demographic changes and thus are
unlikely to have been impacted by the pandemic in 2020.4
3.2. Decomposing Existing Transfers and Pandemic Measures
In our definition of pandemic measures, we include both the introduction of
new types of benefits, such as STW allowance and the increased generosity of exist-
ing transfers, such as the increased ceiling of UI insurance. To identify the contribu-
tion of the transfers already existing at the onset of the pandemic separately from the
contribution of the pandemic measures introduced in 2020, we compute the amount
of a given transfer type that an individual would have received had pandemic mea-
sures not been introduced, taking the individual’s behavior as given. Under the
assumption that behaviors, for example, the risk of becoming unemployed or sick,
did not change in response to the pandemic measures themselves, the high level of
detail of our data also allows us to disentangle the additional benefit an individual
received due to the introduction of pandemic measures.5 This assumption is more
likely to hold in the early stages of the pandemic.
3.3. Estimation Framework
The strength of our analysis is the access to monthly data on labor earnings,
government transfers, and STW pay. Because the COVID-19 pandemic was a sud-
den and unforeseen change in the economic environment, we can exploit the onset
of the pandemic to study the effects of the pandemic on individual incomes during
2020. We do this by applying a DiD framework, in which we compare the change in
monthly income from different sources before and after the onset of the pandemic
in March 2020 with the corresponding change during previous years. This type of
method has previously been employed when studying the impact of the pandemic
(see, e.g., Hensvik et al., 2021; Angelov & Waldenström, 2023a, 2023b).
For our main estimation results, we define cohorts k = 2016, … , 2019 con-
sisting of all individuals aged 19–63 in December of year k. To allow for several
unaffected months for the pandemic cohort sampled in 2019, we follow the cohorts
from July in the sampling year k to December in year k + 1; that is, our DiD estima-
tion consists of a total of 18 months during which we compare outcome variables.
Our main model can be written as:
(1) Yi,k,m,t = 𝛼 + 𝛽 × Pandemict + 𝜆m + 𝜆k + 𝜀i,k,m,t
4 These include disability benefits, pensions, additional housing allowance for pensioners,
work-related injury compensation, child allowance, and pregnancy benefits. Results from an annual
model (see Section 5) in Table A6 in the Online Appendix suggests that study allowance increased by
an average of 31 SEK per month during 2020 and that pension benefits increased by about 2 SEK per
month during 2020.
5 The only transfer we are not able to precisely decompose into existing transfers and pandemic
measures are activity support benefits to unemployed workers participating in an active labor marker
program. These are paid out based on the same criteria following the basic and income-related unem-
ployment insurance. Because of missing information on unemployment insurance membership for these
recipients, we are not able to incorporate all changes introduced during the pandemic. Missing infor-
mation for activity support benefits should, if anything, result in an underestimation of the effect of
UI-related pandemic measures.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
991
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
where i denotes an individual of cohort k, m the running month for each cohort
(1 = Julyk , … , 18 = Decemberk + 1 ), and t the calendar month (July 2016, … ,
December 2020). The vectors 𝜆m and 𝜆k are month- and cohort-fixed effects. The
treatment dummy Pandemic takes the value 1 from March 2020 onward and zero
otherwise, and the parameter of interest 𝛽 captures the average monthly effect
of the COVID-19 pandemic on individual incomes from March to December
2020. The variable Pandemict corresponds to an interaction between an indicator
for belonging to the treatment group (k = 2019) and being in the post-treatment
period (March–December in year k + 1). This effect is the overall effect of the
pandemic, including underlying economic effects (such as layoffs), any behavioral
effects due to the pandemic, and also any behavioral effects that are related to
the changes in the government transfer system during the pandemic. While this
might seem like a strong assumption, Ganong et al. (2020) provide evidence for
the USA that the strong increase in UI supplements had only minor effects on
re-employment rates.
We are also interested in the dynamic effects of the COVID-19 pandemic in
2020. To study this, we augment equation (1) and estimate month-specific treatment
effects 𝛽s in an event-study framework:
Deck+1
∑
(2) Yi,k,m,t = 𝛼 + 𝛽s × Ts × Treati + 𝜆m + 𝜆k + 𝜀i,k,m,t
s=Julyk
where Ts is the monthly indicator, Treati = 1{k = 2019} is an indicator for belong-
ing to the 2019 cohort that was exposed to the pandemic, and 𝛽s estimates the
month-specific effects. January is used as the reference month (i.e., TJanuaryk+1 is
left out). Although no government pandemic responses were implemented until
March, we cannot rule out some behavioral responses already in February, as the
pandemic started reaching Europe. We use January as the reference point to allow
for such “anticipatory” effects. Because individuals can enter the analysis in dif-
ferent cohorts, standard errors are clustered at the individual level throughout the
analysis.
Our key identifying assumption is that, absent the COVID-19 pandemic, all
cohorts would have had similar trends in labor earnings and benefit levels. While
there is no formal test of this assumption, we assess its plausibility by comparing
pre-trends from July in year k to February in year k + 1 in the following subsection
and Section 4.3.
3.4. Descriptive Evidence
Figure 1 shows the development of our main outcome variables, that is, labor
earnings (Figure 1a), the four transfer types (Figure 1b–e), and total income
(Figure 1f) for each of the analysis cohorts during the 18-month follow-up period
between July of year k and December of year k + 1. Because labor earnings are
available at the monthly level only from January 2019, panels a and f show the
corresponding information from January to December of year k + 1. Because STW
allowance did not exist before the pandemic, we do not show it here.
Figure 1a shows the development of average labor earnings for the analysis
cohorts 2018 and 2019. Although monthly data on labor earnings are restricted
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
992
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
(a) Labor earnings (b) UI benefits
Analysis cohort: 2019 2018 Analysis cohort: 2019 2018 2017 2016
29,000 800
28,000 700
Amount (SEK) Amount (SEK)
27,000 600
26,000 500
25,000 400
24,000 300
July Oct Jan April July Oct July Oct Jan April July Oct
Sample year Sample year+1 Sample year Sample year+1
(c) Sickness benefits (d) Parental benefits
Analysis cohort: 2019 2018 2017 2016 Analysis cohort: 2019 2018 2017 2016
650 900
600 800
Amount (SEK) Amount (SEK)
550 700
500 600
450 500
400 400
July Oct Jan April July Oct July Oct Jan April July Oct
Sample year Sample year+1 Sample year Sample year+1
(e) Income support (f) Total income
Analysis cohort: 2019 2018 2017 2016 Analysis cohort: 2019 2018
260 31,000
240 30,000
Amount (SEK) Amount (SEK)
220 29,000
200 28,000
180 27,000
160 26,000
July Oct Jan April July Oct July Oct Jan April July Oct
Sample year Sample year+1 Sample year Sample year+1
Figure 1. Average Monthly Labor Earnings, Government Transfers, and the total income over time
Notes: This figure displays the average monthly income over time by different cohorts. Data on
monthly labor earnings are available from January 2019 onward. Labor earnings for the 2019 cohort
include payments for short-time work allowance. Unemployment (UI) benefits include UI insurance
and activity support. Sickness benefits include sickness insurance and disease carriers’ benefits. Parental
benefits include parental leave and temporary parental leave benefits, and income support includes social
assistance and housing allowance. Total income is the sum of all income sources.
to 2018 onward, the figure paints a clear picture. Before the pandemic, average labor
earnings were higher for the 2019 cohort in comparison to the same months for the
2018 cohort. Once the pandemic hits the 2019 cohort, their average labor earnings
drop relative to the average levels of the 2018 cohort 1 year earlier.
Figure 1b-e show the development of different transfers for the cohorts
k = 2016, … , 2019. For these outcomes, an 18-month comparison period is avail-
able. For UI insurance, sickness insurance, and parental benefits, respectively,
the average levels are fairly similar during the pre-period from July in year k to
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
993
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
February in year k + 1 (i.e., data point to the left of the dashed vertical line).
The same holds for the period between March and December in year k + 1 for
cohorts unaffected by the pandemic. During the pandemic (cohort k = 2019
between March and December in k + 1), however, the average benefits are clearly
higher than those of previous cohorts. From a DiD perspective, these patterns in
the data suggest that the parallel trends assumption is met. For income support,
shown in Figure 1e, the figure shows that average benefits were already at a higher
level before the pandemic (solid line, k = 2019). During the pandemic, however,
income support transfers were further increased, relative to the previous cohorts.
This suggests that, despite higher levels before the pandemic, income support was
affected by the pandemic.
Figure 1f shows the corresponding figure for total income, that is, the sum of
labor earnings and the four transfer types. As expected, the lines are similar to those
of labor earnings, but levels are slightly larger.
Table A3 in the Online Appendix shows some descriptive statistics for the 2019
cohort. In the sample, which comprises all the 5,746,478 individuals aged 20–64
registered in Sweden in January 2020, average monthly labor earnings are 25,592
SEK (2,790 USD). Seventy-five percent of all individuals have positive labor earn-
ings. Between 4 and 5 percent of the individuals receive transfers from UI insurance,
sickness insurance, and income support, respectively. Almost 8 percent of the indi-
viduals receive parental benefits.
4. THE IMPACT OF THE PANDEMIC ON INCOMES DURING 2020
We now turn to the average and dynamic effects of the pandemic during 2020
on earnings and incomes from various sources for the working-age population,
based on estimations of equations (1) and (2), respectively. We disentangle the pro-
tective role of the already-existing transfer system from the additional measures
introduced during the pandemic. We further provide evidence on the effect of the
pandemic on different population groups and trace its impact across the income
distribution.
4.1. Labor Earnings, STW Allowance, and Transfers
Table 1 presents the average effects of the pandemic across the entire popula-
tion from March to December 2020, based on equation (1). Panel a presents the
results for average amounts in SEK. Column 1 in panel a shows that the pandemic
reduced monthly labor earnings in 2020 by, on average, 683 SEK (approximately 70
USD). This corresponds to an earnings reduction of 2.7 percent compared to the
pre-pandemic average for the 2019 cohort. However, without the STW allowance
introduced in March 2020, the drop in labor earnings would most likely have been
higher because more workers would have lost their jobs. On average, payments for
STW allowance amounted to 481 SEK per person and month between March and
December 2020 (column 2, panel a). An upper bound on the pandemic’s impact on
labor earnings is thus a reduction of 1,164 SEK per person and month (column 3,
panel a) or 4.5 percent of pre-pandemic labor earnings. This suggests that, without
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
994
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
TABLE 1
THE EFFECT OF THE COVID-19 PANDEMIC ON EARNINGS, STW PAYMENTS, GOVERNMENT TRANSFERS, AND
TOTAL INCOME
(1) (2) (3) (4) (5)
Labor
Earnings
(Excluding
Labor STW STW Total
Earnings Allowance Allowance) Transfers Income
Panel a: Amounts (SEK)
COVID-19 effect −683.2 481.3 −1,164.5 289.3 −410.6
(11.047) (0.831) (11.091) (1.652) (10.797)
Mean dependent 25,665.2 – 25,665.2 1,645.3 27,394.9
variable
Percentage −2.662 – −4.537 17.59 −1.499
change
Panel b: Take-up (Percentage)
COVID-19 effect −0.829 4.589 −0. 858 5.501 −0.0152
(0.013) (0.007) (0.013) (0.012) (0.011)
Mean dependent 75.06 – 75.06 18.47 83.28
variable
Percentage −1.105 – −1.143 29.77 −0.0183
change
Number of 5,900,301 6,195,625 5,900,301 6,195,625 5,900,301
individuals
Number of 137,339,492 408,541,848 137,339,492 408,541,848 137,339,492
observations
Notes: This table shows results from estimating equation (1). The mean of the dependent variable is
defined as the average of the period between July 2019 to February 2020 for the 2019 treatment cohort.
Transfers include unemployment benefits, sickness benefits, parental benefits, and means-tested income
support, and total income includes labor earnings plus transfers. Standard errors in parentheses are
clustered at the individual level. STW, short-time work.
the introduction of STW allowance, labor earnings would have been reduced by
between 2.7 and 4.5 percent.
Government transfers to individuals also played an important role in mitigat-
ing the adverse income effects of the pandemic. On average, the pandemic brought
an increase in transfers by 289 SEK during the period from March to December
2020 (column 4, panel a). The estimated effect includes both existing transfers and
new pandemic measures, such as higher replacement in UI insurance. Total income,
that is, labor earnings plus government transfers, decreased on average by 411 SEK
(column 5, panel a). This implies that increased transfers compensated, on average,
for 42 percent (289/683) of the drop in labor earnings. In Table A6 in the Online
Appendix, we estimate an annual model using individualized disposable household
income (see Section 5 for details). The estimated effect is smaller (−223) than our
estimate for total income but is also more imprecisely estimated (with a standard
error of 78).
Panel b presents the results on the extensive margin, that is, the change in
the percentage of the population with positive labor earnings, STW allowance,
and different types of income. Column 1 in panel b shows that the share of the
population with positive labor earnings decreased by 0.8 percentage points, or
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
995
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
about 1 percent, on average during 2020. Almost 5 percent of the population
received the new STW allowance (column 2, panel b). The average take-up of any
kind of transfers increased substantially, by 5.5 percentage points or more than 18
percent (column 4, panel b), implying that the share receiving income from labor
or transfers, taken together, was unaffected (column 5, panel b).6
4.2. Existing Government Transfers Versus Pandemic Measures
Starting in March 2020, Sweden introduced a range of temporary changes to
the existing government transfer system (see Section 2.2 and Table A1 in the Online
Appendix). Table 2 shows results disaggregated by transfer type and whether the
transfer existed before the pandemic or was introduced as part of the new pan-
demic measures. Panel a displays the overall effects of the transfer system by type,
where the overall impact of 289 SEK is decomposed into the four transfer types: UI
benefits, sickness benefits, parental benefits, and income support. The results show
that, of the total effect of the different transfer types in 2020, UI benefits increased
the most: 67 percent of the total increase in transfers is due to UI benefits (193 SEK
of 289 SEK). Sickness benefits increased by 79 SEK (27 percent of the increase in
transfers), followed by parental benefits and income support with 5 SEK and 13
SEK (2 and 4 percent), respectively.
We present estimates of the effects on the take-up of the various transfers in
panel b. Column 1 shows that the fraction of the working-age population receiv-
ing UI benefits increased by 1 percentage point, corresponding to an increase of
31 percent. The fraction receiving any sickness benefits increased much more: 5.4
percentage points or by 169 percent. Columns 3 and 4 show that the take-up of
parental leave benefits and income support was essentially unchanged.
In the next step, we disentangle the share of transfer payments that comes from
the existing transfer system that was in place before the pandemic and the share that
comes from measures introduced during the pandemic. Even though we are not
able to disentangle potential behavioral responses to changes in the welfare system,
these estimates provide us with an indication of the relative contribution of different
transfer types and that of existing government transfers and pandemic measures. In
that sense, these estimates provide a benchmark to which degree the existing transfer
system protected individual incomes.
6 We present robustness analyzes of the results in Table 1 in the Online Appendix. Since labor earn-
ings are available at a monthly level only from 2019 onward, whereas monthly transfers are available from
2016, the number of observations is lower in columns 1, 3, and 5 in Table 1. The results in Table A5 limit
all estimations to the same population as for labor earnings, including only 2019–2020 and estimating
the model on 12 months (January–December) with two pre-months (January–February) rather than
18 months with 8 pre-months (June–February) as in our main model. The corresponding estimates are
similar for transfer income (272.6, panel a) and take-up of transfers (5.6, panel b). Panels c and d also
include individual characteristics in the model. Because our sample includes the full population, the com-
position of which does not change much from year to year, we do not expect controls to matter. Because
results with and without inclusion of controls are indeed virtually identical, we henceforth present results
without individual controls, unless otherwise stated. Table A4 in the Online Appendix shows correspond-
ing estimates when using annual labor earnings information divided by 12 for all years. Despite some
differences, the results are fairly similar to those shown in Table 1.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
996
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
TABLE 2
EFFECTS OF THE COVID-19 PANDEMIC ON TRANSFER PAYMENTS, OVERALL AND ATTRIBUTED TO THE
EXISTING TRANSFERS AND TO PANDEMIC MEASURES
(1) (2) (3) (4) (5)
UI Sickness Parental Income Total
Benefits Benefits Benefits Support Transfers
Panel a: Overall Effects–Amounts (SEK)
COVID-19 effect 193.1 78.68 5.089 12.52 289.3
(0.936) (1.000) (1.012) (0.316) (1.652)
Mean dependent 371.3 460.5 595.6 217.9 1645.3
variable
Percentage change 52.00 17.09 0.854 5.746 17.59
Panel b: Overall Effects–Take-up (Percentage)
COVID-19 effect 0.0106 0.0535 0.000 0.002 5.501
(0.000) (0.000) (0.000) (0.000) (0.012)
Mean dependent 0.0342 0.0317 0.0892 0.0461 18.47
variable
Percentage change 30.92 168.8 0.500 4.482 29.77
Panel c: Existing Transfers
COVID-19 effect 130.1 18.76 5.089 2.624 157.0
(0.857) (0.992) (1.012) (0.314) (1.609)
Mean dependent 371.3 460.4 595.6 217.9 1645.3
variable
Percentage change 35.16 4.075 0.854 1.204 9.544
Panel d: Pandemic Measures
COVID-19 effect 62.50 59.91 – 9.895 132.3
(0.169) (0.109) – (0.027) (0.202)
Number of 6,195,625 6,195,625 6,195,625 6,195,625 6,195,625
individuals
Number of 408,541,848 408,541,848 408,541,848 408,541,848 408,541,848
observations
Notes: This table shows results from estimating equation (1). Columns 1 to 4 show the estimate
for the respective transfer type; column 5 shows the sum of the four transfer types. Panels a and b show
the effects of the transfer type overall; panel c shows corresponding results for the transfers that existed
before March 2020; and panel d shows the effects of the newly introduced or changed transfers during the
pandemic. The mean of the dependent variable is defined as the average of the period between July 2019
to February 2020 for the treatment cohort. Standard errors in parentheses are clustered at the individual
level. UI, unemployment.
Panel c shows corresponding results for the existing transfer system, that is,
the changes in transfers that would have accrued, given individual behaviors, had no
pandemic measures been introduced. Of the existing transfers, UI benefits are by far
the most important, making up 82 percent of the total increase in transfer payments
(130 SEK of 157 SEK), followed by sickness benefits (12 percent), parental support
(3 percent), and income support (1.7 percent). Under the implicit assumption that
individual behavior is unaffected by the introduction of pandemic measures, the
existing government transfer system would thus have protected individuals’ incomes
by 157 SEK or 23 percent of individual earnings losses due to the pandemic, thereby
contributing a bit over half of the total replacement rate of 42 percent of the transfer
system (see Section 4.1).
Panel d of Table 2 shows the estimated increases in transfer payments
due to new pandemic measures. Changes to UI and sickness insurance had
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
997
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
the most substantial impact, with on average 63 SEK and 60 SEK per month,
respectively, constituting 47 and 45 percent of the effect of pandemic mea-
sures, respectively. For UI, the main changes include faster eligibility for
income-related UI (shortened membership and work requirements) and higher
floors and ceilings in both basic insurance and income-related UI. For sickness
insurance, reimbursement of the waiting-day deduction, affecting all individ-
ual sickness spells, resulted in increased transfers. There were no changes to
parental benefits, and the increase in income support of on average 10 SEK per
month (8 percent) resulted from the additional housing allowance. Overall, the
pandemic measures increased the benefits by 132 SEK per month, which corre-
sponds to 19 percent of the total labor earnings losses of 683 SEK due to the
pandemic.
The estimates in Table 2 help us understand the relative importance of existing
transfers and pandemic measures within each transfer type. For UI, the benefits
from pandemic measures amount to roughly half of those of the existing system.
For sickness insurance, however, the pandemic measures were three times larger
than the existing transfers.
4.3. The Dynamics of the Effects on Earnings, Government Transfers, and Income
Extending our DiD model to a dynamic event study version, as shown
in equation (2), serves two purposes. First, it enables us to trace the dynamic
effects on incomes and transfers that might have been caused by dynamics
in economic activity and by the design of the pandemic measures, such as
changes in the minimum membership requirements for UI payments. Second,
it allows us to assess further the parallel trends assumption underlying our
DiD model.
Figure 2 shows the dynamic effects on earnings and income measures
(Figure 2a) and the different types of transfers (Figure 2b). Figure 2a shows that
average labor earnings fell drastically at the onset of the pandemic. The policy
response to introduce STW allowance was rapid. In line with the estimates shown
in Table 1, the measure of labor earnings excluding STW allowance (dashed-dotted
line) fell more than 2,000 SEK by May 2020, compared to pre-pandemic levels.
As mentioned, this represents an upper bound on the impact of the pandemic on
labor earnings because it is likely that UI would have been higher in the absence
of the STW allowance but unlikely that all workers on STW allowance would
have been dismissed. The provision of STW allowance roughly halved the fall
in labor earnings. The dashed line displays the fall in income when the increase
in transfers from the existing government transfer system is included, and the
solid line displays the fall in income when transfers from pandemic measures and
adjustments are also accounted for. From May 2020 onward, all income types
shown in panel a started to recover and total income even surpassed pre-pandemic
levels at the end of the year. Note that the sharp recovery in December 2020 likely
stems, at least in part, from payments due to agreed wage raises accumulated
for the past few months as union-employer-negotiated wage agreements were
finally signed in late 2020 (Medlingsinstitutet, 2021). Despite this positive trend
toward the end of 2020, the overall pandemic effect on earnings in 2020 shown in
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
998
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
(a) Income measures
LE + ET + PM LE + ET LE LE excl. STW pay
500
0
Effect (SEK)
–500
–1000
–1500
–2000
July Oct Jan April July Oct
2019 2020
(b) Benefit measures
UI benefits Sickness benefits Parental benefits Income support
300
200
Effect (SEK)
100
0
–100
–200
July Oct Jan April July Oct
2019 2020
Figure 2. Dynamic Effects on Earnings, Incomes, and Government Transfers
Notes: This figure shows event-study estimates with 95 percent confidence intervals (standard errors
clustered at the individual level) (a) for earnings and various income measures and (b) for transfers. ET,
existing transfers; LE, labor earnings; PM, pandemic measures. Estimates are taken from equation (2).
Table 1 was a reduction of wage earnings by 2.7 percent and a loss of total income
of 1.5 percent.
Figure 2b shows the estimated monthly changes for the four transfer types. In
line with Table 2, most transfers increased during the pandemic, although by differ-
ent magnitudes. As suggested by the point estimates in Table 2, increases in UI and
sickness benefits have the most significant compensatory effect on the drop in labor
earnings. While sickness and parental benefits increased sharply at the beginning
of the pandemic as many workers were sick or needed to care for sick children, UI
reached a high level by July. This delayed response can be explained by notice peri-
ods in employment contracts, the design of UI, and the changed eligibility criteria,
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
999
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
which allowed individuals to receive income-related UI benefits after 3 months of
membership in their respective UI funds (instead of, previously, 12 months), cre-
ating an additional incentive to become a member in one of Sweden’s UI funds.
Indeed, from May 2020 onward, membership levels were around 7 percent higher
than in 2019. Before March 2020, however, membership levels were essentially the
same as in the previous year.7
Figure 2 also allows us to look at pre-trends. Albeit estimated only for a
short pre-pandemic period, due to the lack of monthly labor earnings data before
2019, all income measures in panel a show pre-pandemic estimates close to zero,
suggesting that the parallel trends assumption holds. Panel b shows that for all
transfer types except UI, the estimates are close to zero before March. UI levels
appear lower before the pandemic compared to the reference month, January, but
display no trend. It seems January is somewhat of an outlier with a high level of
UI. Had February been chosen as the reference month, pre-pandemic estimates
would have been closer to zero. The estimates for UI, nevertheless, show a clear
break in the trend with the onset of the pandemic, suggesting a causal effect of
the pandemic on UI transfers. We argue that the estimated pre-pandemic effects
presented in Figure 2 support a causal interpretation of our estimated effects of
the pandemic on earnings, transfers, and income.8
4.4. The Effect of the Pandemic on Different Groups
So far, the focus has been on the average effects of the pandemic on labor
earnings, transfers, and total income. Still, it is likely that some groups were more
severely affected by the spread of the virus itself, while the rapid contraction of
economic activity hit groups differentially. Figure 3 shows estimates of the effects
on labor earnings and total income, that is, the sum of earnings and transfers, for
several groups defined by demographic, socioeconomic, and work characteristics.
Demographic groups include gender, age, migration background, and family status.
Socioeconomic groups are limited to education categories. Work characteristics
include whether or not an individual was employed in an occupation with a pos-
sibility to work from home, whether or not an individual worked in the hospitality
sector that was severely hit by the pandemic, and whether or not an individual
worked in a “contact profession” with regular contact to other individuals.9 Note
that the groups are not mutually exclusive. Instead, the sample is divided into
7 Data taken from the Swedish Unemployment Insurance Inspectorate (IAF)’s website (https://www
.iaf.se, accessed 2022-09-08).
8 We note that most of the pre-treatment point estimates would be statistically significantly different
from zero in a hypothesis test setting. We choose not to focus on this for two reasons: first, given our
large dataset, almost any estimate will be statistically significant, even when the point estimate is very
small in economic terms, which is the case here. Second, Roth (2022) warns against conditioning event
studies on tests for pre-trends, because this might cause both point estimates and confidence intervals to
be biased in the subsequent analysis.
9 The measure for the possibility of working from home uses occupational information created by
Dingel & Neiman (2020) based on SOC12. Hensvik et al. (2021) harmonize it to the Swedish standard
SSYK2012. The measure for contact profession comes from Centre for Epidemiology and Community
Medicine (CES) and Center for Occupational and Environmental Medicine (CAMM). It approximates
how much physical contact an occupation requires. The measure goes from 0 (no physical contacts)
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1000
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
Total income Labor earnings
Demographics
Man
Woman
Age: 21–29
30–54
55–64
Foreign born
Born in Sweden
Household w/o children
Single parent
Cohab/married w/ child
Educaon
Compulsory
High school
College
Work situaon
Hospitality sector
All other sectors
WFH possibility
No WFH possibility
Contact profession: Low
Average
High
All
–8 –7 –6 –5 –4 –3 –2 –1 0
Effect (%)
Figure 3. Effects on Labor Earnings and Total Income by Group
Notes: This figure shows estimation results for labor earnings (gray dots) and total income (black
dots) with corresponding 95 percent confidence intervals using standard errors clustered at the individual
level. Each estimate is based on a separate regression for the respective subsample. WFH, working from
home.
groups by each characteristic separately. Characteristics are determined in the
sampling year, that is, pre-pandemic for the 2019 treated cohort.
The results show that the labor earnings losses (in percentage) were largest
for young individuals, individuals with low education, foreign-born individuals,
and, by far most severe, for those who worked in the hospitality industry before
to 100 (very close physical contact). We create three groups: low (0–49), medium (50–74) and, high
(75–100) contact.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1001
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
the pandemic.10 Unlike evidence from other countries and earlier studies for
Sweden (Adams-Prassl et al., 2020; Angelov & Waldenström, 2023a; Dang & Viet
Nguyen, 2021), we do not find that the pandemic hit women more severely than
men. On the contrary, earnings losses and total income losses are significantly
larger for men than for women.11
Comparing the losses of labor earnings to the impact on total income, our
analysis shows that the government transfer system functioned in a compensatory
way; that is, those hit hardest were also those who were “compensated” for large
losses by the transfer system. In spite of this, young individuals emerged as a group
that was hit hard, and the transfer types included in this analysis failed to fully
compensate for the earnings losses. This group might, however, have found other
ways to mitigate labor earnings losses, for example, by registering in education and
thus increasing their take-up of study allowance, that is, student grants or student
loans. Unfortunately, our data do not include study allowance on a monthly basis,
but CSN (2021) reports that the number of students at municipal adult education,
vocational colleges, and universities increased by 17, 9, and 29 percent, respectively,
in 2020 compared to the year before. We can, however, estimate the effect of the
pandemic on the uptake of study allowance using annual data. Our estimates
using an annual model (see Section 5) suggest that there was an increase in study
allowance by SEK 31 per month in 2020 (Table A6 in the Online Appendix).
However, adding study allowance to total income does not significantly change
the results presented in Figure 3. The group-level estimates for total income with
and without study allowance are shown in Figure A3 in the Online Appendix,
using annual information on earnings and study allowances. While including study
allowance reduces the income loss for young adults somewhat, most estimates are
unaffected and patterns across groups remain.
To analyze whether pandemic measures targeted specific groups, Figure A2
in the Online Appendix shows the replacement rate for existing transfers and
pandemic measures for different groups. The replacement rate is defined as the
estimated COVID-19 effect on existing transfer or pandemic measures divided by
the estimated COVID-19 effect on labor earnings using equation (1). This analysis
shows that among demographic groups, single parents received the strongest
income protection from transfers during the pandemic in relative terms. They
benefited from the additional housing allowance, increased UI, and sickness
benefits. Workers in jobs characterized by a high level of personal contact also
received a high replacement rate from the transfer system for their loss in labor
earnings. This group is largely made up of health sector employees and teachers,
and they benefited from high protection from sickness benefits and waiting-day
reimbursement. In Figure A4 in the Online Appendix, we present estimates for
10 The percentage effect for group G is defined as the estimated effect for group G relative to the
average pre-pandemic total income for group G in the 2019 cohort. Figure A1 includes corresponding
estimates for all sectors of industry.
11 One hypothesis for this difference is that Angelov & Waldenström (2023a) also include capital
income. Our estimates using an annual model (see Section 5), however, show a very small effect on capital
income (Table A6). A second possibility is that our sample includes individuals without labor earnings.
This group is larger among women.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1002
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
the extent to which the introduction of the STW allowance protected the labor
earnings of these different groups. The amount of STW allowance is put in relation
to the loss in labor earnings. This analysis shows that STW allowance was almost
as large as the loss in earnings for workers with low-contact jobs and workers with
the ability to work from home.
4.5. Income Losses Across the Income Distribution
The previous subsection shows significant heterogeneity in the effect of the
pandemic on labor earnings with less heterogeneity in the effect on total incomes,
suggesting that the government transfer system had a compensatory effect on total
incomes. This section presents an analysis of the effects along the entire distribution
of pre-pandemic disposable income. In Figure 4 we present estimates for the impact
on labor earnings, labor earnings excluding STW allowance, incomes including
labor earnings and transfers from the existing government transfer system, and
total income including pandemic measures. The estimates are presented for each
vingtile of the distribution. Because our previous analysis shows different effects of
the pandemic for men and women, we conduct the analysis separately for men and
women, and overall. Panel a shows that male labor earnings fell by 8 percent for
the lowest vingtile and by 6 percent for the second vingtile. For the remaining part
of the income distribution, the earnings losses range from 4 to 2 percent for the
highest vingtiles. The figure clearly shows that, in the absence of STW allowance,
these earnings losses would have been substantially larger, by up to 2 percentage
points. Existing transfers reduced income losses the most at the lower end of the
income distribution. But including these transfers, the income losses are still the
largest for low-income groups. The pandemic’s effect on total income, which also
includes the transfers due to pandemic measures, is similar across the pre-pandemic
income distribution. Therefore, the pandemic measures provided important income
protection for men with low incomes.
Panel b shows corresponding results for women. The key differences in the
patterns of men are the following: first, losses in labor earnings are smaller than
for men in the lower half of the income distribution. In the upper part of the
income distribution, losses in labor earnings are relatively similar to losses of men.
Second, STW allowance played a smaller role for women: while STW allowance
prevented about 2 percentage points of earnings losses for men, the corresponding
protection for women is between 1 and 1.5 percentage points. Third, total income
losses are also slightly smaller for women, overall. But the figure also shows that
while ordinary transfers did not contribute much to leveling out female income
losses at the very low end and instead had a more protective effect for below
median incomes above the 15th percentile, the pandemic measures, in particular,
the additional housing allowance, had a protective role at the very bottom of the
income distribution of women. Yet, women in the bottom vingtile experienced
slightly larger income losses, relative to their previous economic situation, than the
lowest vingtile among men. Figure A5 in the Online Appendix shows the results in
absolute terms. These figures show that absolute losses were greater at the top of
the distribution and that men experienced larger losses than women.
In panel c, we show the corresponding results for our entire sample. For the
entire population, the larger earnings losses in the lowest vingtiles, which are driven
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1003
Review of Income and Wealth, Series 70, Number 4, December 2024
(a) Men
—LE+ET+PM --+- LE+ET LE ==" LE excl. STW pay
-6- oboe
By aga
5-847;
-107 £
-127
I] I 1 —
5 25 50 75 100
Disposable income distribution
(b) Women
—LE+ET+PM --+- LE+ET LE ==" LE excl. STW pay
c= qege deb EE Ep oie
3 6 c hi
co i
lu -8 44
-107
1 1 — 1 —
5 25 50 75 100
Disposable income distribution
(c) Overall
—LE+ET+PM =** LE+ET LE -"-" LE excl. STW pay
o7
— : Bente age ta +
sf y Spee age EE
= Porc ae
@
ra
Ww
1 1 — 1 —
5 25 50 75 100
Disposable income distribution
Figure 4. The Effect of COVID-19 in 2020 Across the Disposable Income Distribution
Notes: Each figure shows estimates for the respective earnings and income definitions for (a) men, (b)
women, and (c) the overall sample. ET, existing transfers; LE, labor earnings; PM, pandemic measures.
Each point estimate and respective 95 percent confidence interval is based on a separate regression of
the respective vingtile of the 2-year pre-pandemic disposable income distribution.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1004
SUOTUPUOD pure SULIT aI 29g “[STOT/ZO/TT] HO KreIQrT OUTTUD KOTEM “ArIQHT MUsTOatuP vresdde Kq OLOTT M101/1 TT FOT/OpAUOD: Koya Aresquouruo/:sduy Wo papeolUMoc “b “PZOT 166PSLPI
Stu49) 109" KOLA
pl
asuaorT SUOUTUEOD aAtTeAID e[qeor{dde oy £q powsOAo ome SapoTE WO ‘Asn Jo SofnI JO AreIGYT aUTTUD KOUTA UO (sUOT
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
by women, can be clearly seen. From roughly the fourth vingtile, however, the
earnings losses are largely equalized through both existing transfers and pandemic
measures.
5. EFFECTS BEYOND 2020
So far, we have focused on the immediate impact of the COVID-19 pandemic
during the onset of the pandemic in 2020. After the turbulence of 2020, economies
worldwide started to recover in 2021 and economic uncertainty was much reduced.
The analysis strategy we employ exploits the sudden onset of the pandemic in March
2020 and relies on the assumption that monthly variations in previous years are
good counterfactuals for what would have happened during 2020 had the pandemic
not occurred. This assumption becomes less valid the further in time the analysis is
extended, since other things could have happened that would change the develop-
ment compared to previous years. An additional challenge to extend the analysis
beyond 2020 is that monthly data on labor earnings are available only from 2019.
Also, the monthly data on transfers are not available through the entire year of 2021.
Although our main analysis strategy is not well suited to analyze long-term
effects, this section provides some suggestive evidence of the economic recovery
beyond the first year of the pandemic. Because monthly data on labor earnings
are available only from 2019 and because we have only annual information on the
different income sources for 2021, we are unable to use our previous identification
strategy that makes use of monthly data. Instead, we use annual data on the cor-
responding outcomes or the annual measures that are most similar to the monthly
records used in the previous analysis.
We estimate the following equation:
(3) Yi,k,y,t = 𝛼 + 𝛽1 × d2020y + 𝛽2 × d2021y + 𝜆t + 𝜆k + 𝜀i,k,y,t ,
where Y is the outcome of individual i of cohort k in calendar year y (2016, … ,
2021) for the observation year t = {k, k + 1, k + 2}. The vectors 𝜆t and 𝜆k are obser-
vation year and cohort fixed effects. The treatment dummies d2020 and d2021 take
the value 1 during 2020 and 2021, respectively, and 0 otherwise, and the parameters
of interest 𝛽1 and 𝛽2 capture the average annual difference during 2020 and 2021
compared to previous years. For ease of comparison to previous results, estimated
effects are presented as monthly averages.
Note that all annual estimates presented in this paper have the limitation that
the “treated” period covers all of 2020 and thus includes the last two pre-pandemic
months. This introduces a downward bias in the annual estimates for 2020, which
should be kept in mind when interpreting them.
It is reassuring that the estimated effects for 2020 in Table 3, despite different
data sources and a coarser estimation strategy, differ only slightly from the esti-
mates presented in Table 1 using monthly data. Some caution in interpreting the
exact levels of these estimates is nevertheless warranted. Yet, the analysis is use-
ful for comparing the effects of the pandemic on incomes in 2020 and 2021 to the
incomes in 2019. Column 1 of Table 3 shows that the labor earnings losses of SEK
176 in 2021 were substantially smaller, only one quarter, compared to the SEK 665
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1005
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
TABLE 3
THE EFFECT OF THE COVID-19 PANDEMIC ON EARNINGS, STW PAYMENTS, GOVERNMENT TRANSFERS, AND
TOTAL INCOME DURING 2020 AND 2021 RELATIVE TO 2019
(1) (2) (3) (4) (5)
Labor
Earnings
Labor STW (Excluding STW Total
Earnings Allowance Allowance) Transfers Income
COVID-19 effect −664.8 392.1 −1056.9 240.7 −424.0
2020 (6.565) (0.674) (6.607) (1.644) (6.251)
COVID-19 effect −175.6 121.9 −297.5 298.4 122.8
2021 (9.720) (0.441) (9.731) (2.375) (9.266)
Mean dependent 25,540.6 0 25,540.6 1,643.2 27,183.8
variable
Percentage change −2.603 – −4.138 14.65 −1.560
2020
Percentage change −0.687 – −1.165 18.16 0.452
2021
Number of 6,195,625 6,195,625 6,195,625 6,195,625 6,195,625
individuals
Number of 62,365,078 62,365,078 62,365,078 62,365,078 62,365,078
observations
Notes: This table shows results from estimating equation (1). All outcomes are defined as the
annual outcome divided by 12. Standard errors in parentheses are clustered at the individual level. Labor
earnings also include income of sole proprietors and business owners. Transfers include unemployment
benefits, sickness benefits, parental benefits, and means-tested income support, and total income includes
labor earnings plus transfers. STW, short-time work.
noted for 2020. Income from STW was also lower in 2021, SEK 122 compared to
SEK 392 in 2020. The increase in transfers was, however, larger in 2021 compared
to 2020, SEK 298 as opposed to SEK 241. As a result, the average individual in the
working-age population actually gained SEK 123 in total income in 2021 as com-
pared to 2019. Transfers during 2021 thus more than compensated for the loss of
labor income earnings. Hence, over the two pandemic years, the average loss of labor
earnings was about 1.6 percent, while the average loss of total income was some
−0.6 percent.
Table 4 reveals that increases in both UI and sickness benefits compared to
the pre-pandemic period were larger in 2021 than in the first pandemic year. More-
over, parental benefit receipts, which were reduced the first pandemic year, declined
further in 2021 and the increase in income support compared to 2019 was only
slightly lower in 2021 than in 2020. This implies that increased UI benefits con-
tributed 72 percent of the increase in total transfers in 2021 compared to 68 percent
in 2020, while sickness benefits constituted 31 percent in 2021 and 26 percent in
2020. These results show that both UI and sickness benefits remained important
sources of income in 2021.
6. CONCLUSIONS
During the COVID-19 pandemic, many governments introduced temporary
adjustments to the government transfer system to counter the economic and public
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1006
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
TABLE 4
THE EFFECT OF THE COVID-19 PANDEMIC ON GOVERNMENT TRANSFERS DURING 2020 AND 2021
(1) (2) (3) (4) (5)
UI Sickness Parental Income Total
Benefits Benefits Benefits Support Transfers
COVID-19 effect 2020 164.8 61.93 −5.444 19.48 240.7
(0.935) (0.991) (0.993) (0.341) (1.644)
COVID-19 effect 2021 214.5 92.73 −24.10 15.27 298.4
(1.351) (1.470) (1.371) (0.507) (2.375)
Mean dependent variable 397.7 468.6 578.7 198.2 1643.2
Percentage change 2020 41.43 13.22 −0.941 9.832 14.65
Percentage change 2021 53.93 19.79 −4.165 7.705 18.16
Number of individuals 6,195,625 6,195,625 6,195,625 6,195,625 6,195,625
Number of observations 62,365,078 62,365,078 62,365,078 62,365,078 62,365,078
Notes: This table shows results from estimating equation (3). All outcomes are defined as the annual
outcome divided by 12. Standard errors in parentheses are clustered at the individual level. Columns 1 to
4 show the estimate for the respective transfer type; Column 5 shows the sum of the four transfer types.
health consequences of the COVID-19 pandemic. We study the effects of the
COVID-19 pandemic on labor earnings and government transfers using a DiD
approach and population-wide Swedish administrative data on monthly earnings
and monthly government transfer payments.
We find that, on average, labor earnings declined by 2.7 percent during the first
10 months of the pandemic. Almost half of this decline, however, was replaced by
government transfers, primarily UI and sickness benefits. Moreover, we find that the
generous STW allowance, introduced to protect jobs, constituted a large share of
individual labor earnings during 2020 and is thus likely to have significantly reduced
the economic impact of the pandemic on earnings by an order of magnitude similar
to the government transfer system. Our analysis of 2021 points to a recovery of
labor earnings, while transfers remained high.
Our detailed data allow for disaggregation of government transfers into pay-
ments due to already-existing transfers that were in place before the pandemic and
payments due to pandemic measures that were introduced in response to the pan-
demic. The analysis shows that the pandemic measures were almost as important for
replacing earnings losses as the existing transfers in 2020. The most important trans-
fer systems were UI and sickness insurance. Important pandemic measures included
the raised ceilings in income-related UI and the waiting-day reimbursement in the
sickness insurance system.
Our results further show that labor earnings at the lower end of the
pre-pandemic income distribution were most heavily affected by the pandemic and
that losses were particularly large for young individuals, those with low-education,
foreign-born individuals and, most notably, for those working in the hospitality
sector before the pandemic. Yet, both the existing transfers and the pandemic mea-
sures were compensatory and were largely able to even out differences in income
losses, at least in 2020. The STW scheme, on the contrary, was non-compensatory,
replacing a larger share of earnings losses higher up in the distribution in 2020.
The analysis for 2021 shows that earnings losses were substantially smaller
during 2021 than in 2020 compared to pre-pandemic levels but that transfer
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1007
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
payments remained high. This resulted in increasing total income compared to
pre-pandemic levels.
Our findings show that swift discretionary policy measures can play an impor-
tant role in mitigating the economic consequences of unanticipated crises. While
existing transfer systems can mitigate some of these consequences, using them to
target particularly vulnerable groups with additional measures and introducing
measures adjusted to the nature and type of crisis can be equally important.
A challenge in designing temporary measures lies in targeting the groups most
affected by upcoming crises and being able to remove temporary measures when
they are not needed any longer.
While these findings hold first and foremost for the pandemic in Sweden, we
can compare estimates with earlier crises in Sweden and other countries during the
pandemic. Regarding earlier crises, our estimate for the loss in labor earnings is
remarkably similar to recent findings for the crisis in the 1990s and the financial cri-
sis in Sweden, which led to 12 percent lower labor incomes 2 years after the onset of
the crisis in the 1990s and to a 1 percent increase for the financial crisis (Engdahl &
Nybom, 2021). With regard to the comparability to other countries, it is important
to highlight that the main policy responses to the pandemic were similar to other
countries. STW allowance, for example, has been used in many countries, including
the United Kingdom, Germany, and France. It would be valuable to have studies
similar to ours for other countries to corroborate our findings.
REFERENCES
Adams-Prassl, A., Boneva, T., Golin, M., & Rauh, C. (2020). Inequality in the impact of the coronavirus
shock: Evidence from real time surveys. Journal of Public Economics, 189, 104245.
Almeida, V., Barrios, S., Christl, M., De Poli, S., Tumino, A., & van der Wielen, W. (2021). The impact
of COVID-19 on households’ income in the EU. Journal of Economic Inequality, 19, 413–31.
Angelov, N., & Waldenström, D. (2023a). COVID-19 and income inequality: Evidence from monthly
population registers. Journal of Economic Inequality, 21, 351–79.
Angelov, N., & Waldenström, D. (2023b). The impact of COVID-19 on economic activity: Evidence
from administrative tax registers. International Tax and Public Finance 30, 1718–1746.
Anna, S., Engdahl, M., Hall, C., Holmlund, H., Lundin, M., Mühlrad, H., & Öckertal, B. (2021). Barn
och unga under coronapandemin (p. 2). IFAU.
Avellaneda, A., Chang, R., Collado, D., Jara, H. X., Mideros, A., Montesdeoca, L., Rodríguez, D.,
Torres, J., & Vanegas, O. (2021). Assessing the cushioning effect of tax-benefit policies in the Andean
region during the COVID-19 pandemic. In CeMPA Working Paper 8/21, Institute for Social and
Economic Research. University of Essex.
Bargain, O., Carrillo-Maldonado, P., & Jara, H. X. (2023). Top earners and earnings inequality dur-
ing the COVID-19 pandemic: Evidence from Ecuadorian administrative data. In WIDER working
paper series wp-2023-4, World Institute for Development Economic Research (UNU-WIDER). World
institute.
Bonadio, B., Zhen, H., Levchenko, A. A., & Pandalai-Nayar, N. (2021). Global supply chains in the
pandemic. Journal of International Economics, 133, 103534.
Braband, C., Consiglio, V. S., Grabka, M., Hainbach, N., & Königs, S. (2022). Disparities in labour
market and income trends during the first year of the COVID-19 crisis – Evidence from Germany.
In IZA Discussion Papers 15475, Institute of Labor Economics (IZA). IZA.
Brewer, M., & Tasseva, I. V. (2021). Did the UK policy response to Covid-19 protect household incomes?
Journal of Economic Inequality, 19, 433–58.
Bruckmeier, K., Peichl, A., Popp, M., Wiemers, J., & Wollmershäuser, T. (2021). Distributional effects
of macroeconomic shocks in real-time. Journal of Economic Inequality, 19, 459–87.
Brunori, P., Maitino, M. L., Ravagli, L., & Sciclone, N. (2021). Distant and different? Lockdown and
inequalities in Italy. Economia Pubblica, 2021, 39–54.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1008
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
Cantó, O., Figari, F., Fiorio, C. V., Kuypers, S., Marchal, S., Romaguera-de-la-Cruz, M., Tasseva, I. V., &
Verbist, G. (2022). Welfare resilience at the onset of the COVID-19 pandemic in a selection of Euro-
pean countries: Impact on public finance and household incomes. Review of Income and Wealth, 68,
293–322.
Carta, F., & Philippis, M. D. (2021). The impact of the COVID-19 shock on labour income inequal-
ity: Evidence from Italy. In 606, Questioni Di Economia e Finanza (Occasional Papers). Economic
Research and International Relations Area.
Campa, P., Roine, J., & Strömberg, S. (2021). Unemployment inequality in the pandemic: Evidence from
Sweden (p. 16330). CEPR. Discussion Paper.
Christl, M., De Poli, S., Figari, F., Hufkens, T., Leventi, C., Papini, A., & Tumino, A. (2021). The
cushioning effect of fiscal policy in the EU during the COVID-19 pandemic. In JRC work-
ing papers on taxation and structural reforms 02/2021. European Commission, Joint Research
Centre.
Christl, M., De Poli, S., Hufkens, T., Peichl, A., & Ricci, M. (2023). The role of short-time work and
discretionary policy measures in mitigating the effects of the COVID-19 crisis in Germany. Inter-
national Tax and Public Finance, 30, 1107–36.
Christl, M., De Poli, S., Kucsera, D., & Lorenz, H. (2022). COVID-19 and (gender) inequality in income:
The impact of discretionary policy measures in Austria. Swiss Journal of Economics and Statistics,
158, 4.
Clark, A. E., D’Ambrosio, C., & Lepinteur, A. (2021). The fall in income inequality during COVID-19
in four European countries. Journal of Economic Inequality, 19, 489–507.
CSN. (2021). Studiestödet 2020. Rapport 2021:2. CSN.
Dang, H.-A. H., & Viet Nguyen, C. (2021). Gender inequality during the COVID-19 pandemic: Income,
expenditure, savings, and job loss. World Development, 140, 105296.
Dingel, J. I., & Neiman, B. (2020). How many jobs can be done at home? Journal of Public Economics,
189, 104235.
Eliason, M. (2021). The unequal (?) burden of unemployment in Sweden during the first wave of the
COVID-19 pandemic (p. 14). IFAU. Working paper.
Engdahl, M., & Nybom, M. (2021). Arbetsmarknadseffekter av konjunkturnedgångar. Rapport, 2021,
8.
Figari, F., & Fiorio, C. V. (2020). Welfare resilience in the immediate aftermath of the COVID-19 out-
break in Italy, EUROMOD Working Paper, No. EM6/20, University of Essex, Institute for Social
and Economic Research (ISER).
Ganong, P., Noel, P., & Vavra, J. (2020). US unemployment insurance replacement rates during the
pandemic. Journal of Public Economics, 191, 104273.
Giupponi, G., & Landais, C. (2023). Subsidizing labor hoarding in recessions: The employment and
welfare effects of short time work. Review of Economic Studies, 90, 1963–2005.
Han, J., Meyer, B. D., & Sullivan, J. X. (2020). Income and poverty in the COVID-19 pandemic. National
Bureau of Economic Research. Working Paper, 27729.
Hensvik, L., Barbanchon, T. L., & Rathelot, R. (2021). Job search during the COVID-19 crisis. Journal
of Public Economics, 194, 104349.
Juranek, S., Paetzold, J., Winner, H., & Zoutman, F. (2021). Labor market effects of COVID-19 in Swe-
den and its neighbors: Evidence from administrative data. Kyklos, 74, 512–26.
Kyyrä, T., Pirttilä, J., & Ravaska, T. (2021). The Corona crisis and household income: The case of a
generous welfare state. In Mimeo 61, VATT Institute for Economics Research. VATT.
Landais, C., Nekoei, A., Nilsson, P., Seim, D., & Spinnewijn, J. (2021). Risk-based selection
in unemployment insurance: Evidence and implications. American Economic Review, 111,
1315–55.
Larrimore, J., Mortenson, J., & Splinter, D. (2022). Earnings shocks and stabilization during COVID-19.
Journal of Public Economics, 206, 104597.
Lastunen, J., Rattenhuber, P., Adu-Ababio, K., Gasior, K., Jara, H. X., Jouste, M., McLennan, D.,
Nichelatti, E., Oliveira, R. C., Pirttilä, J., Richiardi, M., & Wright, G. (2021). The mitigating role of
tax and benefit rescue packages for poverty and inequality in Africa amid the COVID-19 pandemic.
In Working paper 2021/148. Helsinki.
Li, J., Vidyattama, Y., La, H. A., Miranti, R., & Sologon, D. M. (2022). Estimating the impact of
COVID-19 and policy responses on Australian income distribution using incomplete data. Social
Indicators Research, 162, 1–31.
Lustig, N., Martinez Pabon, V., Sanz, F., & Younger, S. D. (2021). The impact of COVID-19 and
expanded social assistance on inequality and poverty in Argentina, Brazil, Colombia and Mexico.
In Commitment to equity (CEQ) working paper series 92. Tulane University, Department of
Economics.
Matias, C., & Forsythe, E. (2023). Distributional impacts of the COVID-19 pandemic and the CARES
act. Journal of Economic Inequality, 21, 325–49.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1009
14754991, 2024, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/roiw.12670 by Uppsala University Library, Wiley Online Library on [12/02/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Review of Income and Wealth, Series 70, Number 4, December 2024
Medlingsinstitutet. (2021). Avtalsrörelsen och lönebildningen 2020. Årsrapport.
Menta, G. (2021). Poverty in the COVID-19 era: Real-time data analysis on five European countries. In
S. Bandyopadhyay (Ed.), Research on economic inequality: Poverty, inequality and shocks. Emerald
Publishing Limited.
Milesi-Ferretti, G. M. (2021). A most unusual recovery: How the US rebound from COVID differs from
rest of G7. Brookings Institution, 8, 2021.
O’Donoghue, C., Sologon, D. M., & Kyzyma, I. (2023). Novel welfare state responses in times of crises:
COVID-19 crisis vs. the great recession. Socio-Economic Review, 21, 501–31.
O’Donoghue, C., Sologon, D. M., Kyzyma, I., & McHale, J. (2020). Modelling the distributional impact
of the COVID-19 crisis. Fiscal Studies, 41, 321–36.
OECD (2020). G20 GDP growth-second quarter of 2020. OECD.
Roth, J. (2022). Pretest with caution: Event-study estimates after testing for parallel trends. American
Economic Review, 4, 305–22.
Stantcheva, S. (2022). Inequalities in the times of a pandemic. Economic Policy, 37, 5–41.
SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this
article at the publisher’s web site:
Data S1. Supporting Information.
Appendix S1. Supplementary Appendix.
© 2023 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of
International Association for Research in Income and Wealth.
1010
File and source
- File
- earnings-losses-and-the-role-of-the-welfare-state-during-the-scp-covid-scp-19-pandemic-evidence-from-sweden.pdf
- Size
- 654,501 bytes
- SHA-256
- 34d3c9087ad234684726fbf56cf479649495d0bad06914db5363fa53f66ed01d
- Original
- doi.org