Electronic copy available at: https://ssrn.

com/abstract=3721216
 WP/20/191

What are the Economic Effects of Pandemic Containment Policies?

Evidence from Sweden

by Jana Bricco, Florian Misch, and Alexandra Solovyeva

IMF Working Papers describe research in progress by the author(s) and are published
to elicit comments and to encourage debate. The views expressed in IMF Working Papers
are those of the author(s) and do not necessarily represent the views of the IMF, its
Executive Board, or IMF management.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
© 2020 International Monetary Fund WP/20/191

IMF Working Paper

European Department

What are the Economic Effects of Pandemic Containment Policies? Evidence from Sweden

Prepared by Jana Bricco, Florian Misch and Alexandra Solovyeva 1

Authorized for distribution by Khaled Sakr

September 2020

IMF Working Papers describe research in progress by the author(s) and are published to
elicit comments and to encourage debate. The views expressed in IMF Working Papers are
those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board,
or IMF management.

Abstract

This paper examines the economic effects of policies to contain Covid-19, by extracting
lessons from Sweden’s experience during the ‘Great Lockdown’. Sweden’s approach was
less stringent and based more on social responsibility than legal obligations compared to
European peers. First, we provide an account of Sweden’s strategy and the health outcomes.
Second, drawing on a range of data sources and empirical findings, our analysis of the first
Covid-19 wave indicates that a less stringent strategy can soften the economic impact
initially. These benefits could be eroded subsequently, due to potentially higher infection
rates and a prolonged pandemic, but in Sweden’s case, the evidence remains mixed in this
regard, and it is premature to judge the outcome of Sweden’s containment strategy. In
addition, the economic effects of the containment strategy also depend on social behavior,
demographics and structural features of the economy, such as the degree of export
orientation, reliance on global supply chains, and malleability to remote working.
JEL Classification Numbers: E0, I12, I18
Keywords: Containment, Sweden, Covid-19
Author’s E-Mail Address: JBricco@imf.org, FMisch@imf.org, and ASolovyeva@imf.org

1
We thank Khaled Sakr for detailed guidance, and Mahmood Pradhan, Ravi Balakrishnan, Nazim Belhocine,
Ashok Bhatia, Sascha Buetzer, Jorge Canales Kriljenko, Paul Elger, Ziji Huang, Daniel Leigh, Srobona Mitra,
Carlos Mulas Granados, Catherine Pattillo, Laura Papi, Emil Stavrev, Philippe Wingender, and Silvia Zucchini,
staff of the Swedish Ministry of Finance and Ministry of Health and Social Affairs and the Riksbank for
comments. Tamari Mitaishvili and Chun Jin provided excellent research assistance. An earlier version of this
paper underpins an IMF Sweden country focus published in June 2020, available here:
https://www.imf.org/en/News/Articles/2020/06/01/na060120-sweden-will-covid-19-economics-be-different

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 Disclaimer

The current Covid-19 pandemic and the associated economic crisis are rapidly changing. The
analysis included in this paper is based on data available as of mid-July 2020, with a few
exceptions where we used more recent data as indicated. Some of the conclusions of the
paper may be updated in subsequent studies. The IMF Covid-19 page is periodically updated
on the global outbreak, including country-specific information.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 4

TABLE OF CONTENTS

ABSTRACT _______________________________________________________________2

I. INTRODUCTION ________________________________________________________6

II. SWEDEN’S CONTAINMENT STRATEGY IN INTERNATIONAL

PERSPECTIVE____________________________________________________________8

III. SWEDEN’S HEALTH OUTCOMES SO FAR _______________________________9

IV. CONTAINMENT POLICIES AND ECONOMIC OUTCOMES _______________14

A. AGGREGATE ECONOMIC ACTIVITY ___________________________________________14
B. SECTORAL ECONOMIC ACTIVITY _____________________________________________16
C. JOB SEEKERS AND BENEFICIARIES OF SHORT-TERM WORK SCHEMES_________________18
D. MOBILITY INDICATORS ____________________________________________________19

V. CONCLUSION AND POLICY IMPLICATIONS ______________________________________23

REFERENCES ___________________________________________________________26

FIGURES
1. STRICTNESS OF SOCIAL DISTANCING MEASURES __________________________________8
2. NUMBER OF REPORTED COVID-19 CASES (AUGUST 12, 2020)_______________________10
3. DAILY INFECTIONS ________________________________________________________10
4. TOTAL NUMBER OF REPORTED COVID-19 DEATHS (AUGUST 13, 2020) _______________10
5. DAILY COVID-19 DEATHS __________________________________________________11
6. EXCESS COVID-19 MORTALITY RATES ________________________________________11
7. DISTRIBUTION OF SINGLE PERSON HOUSEHOLDS _________________________________11
8. ELDERLY POPULATION _____________________________________________________11
9. SWEDEN: DEATH BY AGE GROUPS ____________________________________________12
10. TIMELINE ON RESTRICTION TO VISIT THE ELDERLY ______________________________12
11. COVID-19 DEATHS AND POPULATION WITH FOREIGN BACKGROUNDS ________________13
12. COVID-19 DEATHS AND HOUSEHOLD SIZE_____________________________________13
13. GDP QOQ GROWTH, 2020:Q1 ______________________________________________14
14. PRIVATE CONSUMPTION AND INVESTMENT, 2020:Q1 ____________________________14
15. STRINGENCY OF CONTAINMENT AND GDP QOQ GROWTH, 2020:Q1 ________________15
16. ECONOMIC ACTIVITY INDICATORS ___________________________________________15
17. GDP QOQ GROWTH (PRELIMINARY), 2020:Q2 _________________________________16
18. GDP HOH GROWTH, 2020:H1 ______________________________________________16
19. PURCHASING MANAGERS INDEX: MANUFACTURING _____________________________16
20. SWEDEN: SHORT-TERM LAYOFFS AND EMPLOYMENT SHARE BY SECTOR _____________16
21. PURCHASING MANAGERS INDEX: SERVICES ___________________________________17
22. EUROPE: HOTEL REVENUES IN MARCH _______________________________________17
23. TRADE OPENNESS AND HOSPITALITY INDUSTRY ________________________________17
24. JOBSEEKERS ____________________________________________________________18

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 5

25. JOBSEEKERS AND BENEFICIARIES OF SHORT-TERM WORK SCHEMES _________________18

26. NORDICS: CHANGES IN MOBILITY ___________________________________________20
27. NORDICS: WORKPLACE MOBILITY AND CARD TRANSACTIONS FOR SERVICES _________20
28. SWEDEN: CHANGES IN VISITS TO WORKPLACES ________________________________22

TABLES
1. CONTAINMENT MEASURES IN THE NORDICS _____________________________________9
2. EFFECTS OF CONTAINMENT ON VISITS OF WORKPLACES ___________________________22
3. ROBUSTNESS CHECKS _____________________________________________________23

APPENDICES
I. CORRELATION BETWEEN 2020:Q1 GROWTH AND CONTAINMENT POLICIES _____________29
II. DESCRIPTION OF VARIABLES USED FOR THE ECONOMETRIC ANALYSIS _______________29
III. ANNOUNCED FISCAL, MONETARY AND MACRO-FINANCIAL POLICY RESPONSES AS OF
JUNE 2020 _______________________________________________________________30

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 6

I. INTRODUCTION

Many countries have imposed aggressive containment and social distancing measures in an
effort to slow the spread of Covid-19. While the pandemic and containment measures
undoubtedly lead to significant economic fallouts, the effects of the latter are still debated.
Sweden’s strategy to contain the Covid-19 pandemic has been less stringent than in other
advanced economies during the ‘Great Lockdown’ in Europe in March and April 2020. This
makes Sweden an outlier in terms of its containment policies, even though the containment
strategies implemented by Sweden and Nordic peers also shared many similarities. The
objective of this paper is to examine the effects of containment policies on macroeconomic
and health outcomes during the first wave of the pandemic, using various data sources and
drawing on evidence from Sweden and its international peers.

Sweden’s containment strategy has been based more on recommendations and social
responsibility rather than legal obligations. Sweden has refrained from universal school
closures, stay-at-home orders, and mandatory business closures. The death toll has been higher
than in regional peers and has been concentrated along geographical and demographic lines.

By contrast, the economic ramifications of a less stringent containment strategy remain

uncertain. As of September 2020, IMF staff, the Swedish authorities, the National Institute of
Economic Research (NIER), and commercial banks all project a severe recession in Sweden,
ranging from around -5 to -3 percent. Although the range of the projected contraction is less
steep than the average forecast for advanced economies, it is not much different from some
other Nordic countries.

There is a rapidly growing number of recent papers that examine the effects of containment
polies on economic outcomes. Building on the canonical Susceptible Infected Recovered
(SIR) model by Kermack and McKendrick (1927), recent theoretical studies argue that
targeted containment measures can generally achieve the same health outcomes at a lower
economic cost. 2

One strand of the empirical literature looks at the effects of containment measures on
employment. Aum et al. (2020) present causal evidence on the effects of the spread of the
pandemic on labor markets and shows that the number of infections, regardless of lockdown
policies, result in job losses, which underlines the role of voluntary social distancing. Cajner

2
See Avery (2020) for a survey of the literature on SIR models. For example, Eichenbaum et al. (2020) develop
a single-group SIR model where people’s decision to reduce consumption and work moderates the severity of
the epidemic but exacerbates the recession. They show that test-based quarantines have larger social benefits
than broad-based containment policies like lockdowns. Ellison (2020) argues that allowing for heterogeneity in
contact rates is important because calibrating the classic SIR models to data generated by heterogeneous models
can lead to biased forecasts and understatement of the forecast uncertainty. Acemoglu et al. (2020) find that in a
model with infection/fatality rates varying across age, policies differentially targeting age groups significantly
outperform uniform policies with most of the gains realized by having stricter lockdown policies on the elderly.
Favero et al. (2020) show that policies of gradual reopening that take into account age-specific and sector-
specific risks may save many lives with limited economic costs.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 7

et al. (2020) document that the decline in employment has been much larger at the bottom of
the wage distribution in the U.S. Chen et al. (2020) do not find significant effects of
containment policies on state-level unemployment claims in the U.S.

Other empirical papers examine the effects on private consumption using financial
transaction data. Andersen et al. (2020) estimate that the decline in aggregate spending in
Sweden was 26 percentage points, only 4 percentage points less than in Denmark despite the
stark differences in the stringency of containment policies, potentially highlighting the
importance of voluntary changes in behavior. By contrast, Baker et al. (2020) show that
consumer spending in the U.S. changed more strongly in states with shelter-in-place orders.

Another strand of the empirical literature examines the effects on non-traditional indicators
and mobility. 3 Deb et al. (2020a) find the dynamic cumulative effects of containment to have
a significant impact on Nitrogen Dioxide emissions, mobility, and other indicators, although
they show that fiscal and monetary stimuli can soften the adverse economic effects. Maloney
and Taskin (2020) examine the effects of closing nonessential businesses, sheltering in place,
and school closures mostly on presence at work. They show that these interventions are
effective, but that voluntary actions also matter. Fang et al. (2020) find that the lockdown of
Wuhan reduced mobility within, into, and out of Wuhan significantly. Born et al. (2020) find
that the decline in mobility in Sweden was only slightly lower than in a synthetic control
group comprising countries that had a more stringent lockdown. Chen et al. (2020) also find
that trends in electricity usage between March and April 2020 was broadly similar across
Nordic economies, including Sweden.

We contribute to the debate on the effects of containment policies by presenting an array of

evidence and data using Sweden’s role as an outlier. First, we examine a range of economic
outcomes and present novel stylized facts. Second, we expand earlier regression analysis
methodologically by exploiting within-week country variation in containment policies that
are measured using a continuous index.

To preview our results, we show that based on Sweden’s experience, milder containment
measures can soften the economic impact of the pandemic initially as suggested by evidence
on aggregate growth in the first quarter of 2020 (which was stronger than in most other
advanced economies), mobility indicators, and indicators measuring economic activity in
non-tradeable sectors. At the same time, some of these economic benefits could be eroded
subsequently, possibly because of higher infection rates that had varying effects across
sectors. In Sweden’s case, the evidence on erosion remains mixed, not least because GDP
growth in 2020:Q2 was worse than in peers, while the cumulative performance in 2020:H1
remains relatively favorable. We also argue that the economic impact of the containment
strategy depends on social behavior, demographics and the structure of the economy, and it is

3
Mobility indicators are typically based on cell phone data.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 8

still premature to judge the outcome of Sweden’s containment strategy as it will depend on
developments in the next quarters.

The remainder of the paper is organized as follows. First, we discuss Sweden’s containment
strategy in an international perspective. Second, we compare health outcomes. Third, we
examine the economic impact by analyzing several types of conventional and non-
conventional indicators. Finally, we discuss policy implications.

II. SWEDEN’S CONTAINMENT STRATEGY IN INTERNATIONAL PERSPECTIVE

Similarly to other countries, the Swedish government’s overarching goal from the onset of
the Covid-19 pandemic has been to safeguard people’s lives and health and to maintain
health care capacity, by flattening the curve of infections and limiting the number of people
that are infected at the same time. 4 While Sweden has put in place a range of social
distancing and other containment measures, starting from early March—around the same
time when most other European countries started ramping up containment measures—its
overall strategy remained less restrictive than by its peers. In addition, Sweden never entered
a period of a stringent lockdown with compulsory stay-at-home orders, school-closures and
mandatory business closures. Data from the Oxford Covid-19 Government Response
Tracker, which seeks to quantify the stringency of containment measures, suggest that there
was a substantial difference between Sweden and the other Nordics as well as the average of
the European Economic Area (EEA) and the UK in terms of stringency of social distancing
measures during the Great Lockdown (Figure 1). 5

Figure 1. Strictness of Social Distancing Measures

(Index)
90
80
SWE
70
60
EEA
50
40 Nordics

30
20
10
0

Sources: Oxford COVID-19 Government Response Tracker; IMF Staff calculations. Notes: 'EEA' is the average
of all EEA countries including the UK and excluding Greece, Latvia, Lichtenstein, Lithuania, and Greece. 'Nordic'
is the average of Sweden, Denmark, Finland and Norway.

4
https://www.government.se/articles/2020/04/strategy-in-response-to-the-covid-19-pandemic/

5
The Government Response Stringency Index published daily by Oxford’s Covid-19 is a composite measure
based on nine response indicators including school closures, workplace closures, and travel bans, rescaled to a
value from 0 to 100 (100 = strictest response); see Hale et al. (2020) for details.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 9

This clear difference in the stringency index during the period of the lockdown can be mainly
ascribed to the fact that the containment strategy has been based to a larger extent on
recommendations rather than legal requirements, appealing to common sense and social
responsibility. Social responsibility has been a corner stone of Sweden’s strategy, which
called on the population to exercise more caution based on their risk assessment regardless of
the de jure regulations. This is believed to have complemented the legal restrictions and,
thus, moderated the public health impact.

While Sweden’s containment strategy has attracted much attention, it is sometimes not
characterized completely accurately. 6 For instance, several articles and reports portray that
Sweden has not put in place any restrictions; see for instance Krueger et al. (2020). In
addition, while it has often been reported that restaurants remained opened in Sweden, it is
rarely mentioned that they actually operated under several restrictions, including mandatory
social distancing rules. Retailers, sports clubs, and hair salons were also subject to a set of
guidelines. 7 Finally, there are also similarities in the containment measures across the Nordic
countries which are sometimes ignored. For instance, no Nordic country mandated closures
of manufacturing plants, and secondary schools were closed throughout the region (Table 1).

Table 1. Containment Measures in the Nordics

Containment Visiting Closed Assem- Secondary Restaurants Primary Retailers Sports Clubs Hair Salons Construct.,
measures elderly Borders blies school & Schools & Manufact.
higher earlier
Sweden Open with Open with Open with Open with
Non-EU >50
restrictions restrictions restrictions restrictions
Denmark Take out Retailers
>10
(lockdown) only partly open
Finland Take out Grade 0-3 Open with
>10
(lockdown) only open restrictions
Norway Open with Retailers
>5
(lockdown) restrictions partly open

Notes: Color coding refers to strictness of measures implemented in each category. The darker the red the
more stringent is the containment measures. Lockdown refers to the period of mid-March/beginning April.

III. SWEDEN’S HEALTH OUTCOMES SO FAR

As of mid-August, Sweden’s reported Covid-19 cases per capita have surpassed those of
Nordic peers and countries heavily impacted by Covid-19 such as Italy, Spain, and the UK
but not the U.S. (Figure 2). There is robust cross-country evidence that the stringency of
containment strategies affects the number of Covid-19 cases, especially in countries like the

6
For more on the strategy and its rationale see public statements by Sweden’s health ministry and interviews by
the country’s chief epidemiologist.

7
https://www.folkhalsomyndigheten.se/globalassets/publicerat-material/foreskrifter/konsoliderade/hslf-
fs_2020_12.pdf

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 10

Nordics (Deb et al., 2020b). 8 The number of recorded daily infections has peaked in end-
June, much later than in other Nordic countries (Figure 3). However, the recorded number of
cases is affected by the scale of testing which has increased significantly in Sweden,
suggesting that the actual peak of the first wave has occurred much earlier.

Figure 2. Number of Reported Covid-19 Cases (August 12, 2020)

While Sweden’s death toll per capita remains below that of the UK, Spain, and Italy, it is much
higher than in other Nordics (Figure 4). The number of daily deaths has been falling since its
peak in April (Figure 5) which occurred later than in other Nordic countries. Errors in the
attribution of deaths in Sweden are unlikely as reported Covid-19 deaths explain a large share
of currently observed excess mortality, suggesting that official statistics are rather accurate and
do not underreport Covid-19 deaths contrary to statistics in some other countries (Figure 6).

Figure 3. Daily Infections Figure 4. Total Number of Reported

Covid-19 Deaths (August 13, 2020)

8
In the case of Sweden, other factors may have affected the spread of the pandemic that are not related to
containment policies, such as a higher number of imported Covid-19 cases at the onset of the crisis from people
returning from skiing holidays; see Public Health Agency of Sweden (2020c) for evidence that a large share of
infections in February and March can be linked to travelling to Austria and Italy.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 11

Figure 5. Daily Covid-19 Deaths Figure 6. Excess Covid-19 Mortality Rates

(Persons) (COVID-19 deaths divided by total excess deaths 1/)
120
7-day moving average 14-day moving average
240
100 180
120
80
60
0
60

40

20 Sources: The Economist "Tracking covid-19 excess deaths across countries"

1/ Cumulative since the week a given country recorded more than 50 deaths.
0 Weeks (w) mentioned next to country name. 2/ Norway's indicator is actually
Mar-20 Apr-20 May-20 Jun-20 Jul-20 negative, i.e. less people died during these weeks than expected. 3/ Denmark,
Belgium, Germany, France and Sweden's indicators are above 100 percent as more
Sources: Bloomberg and Johns Hopkins University. COVID-19 deaths were recorded than expected.

Favorable demographic characteristics may have helped contain the number of deaths compared
to Italy, the UK and Spain. Sweden has a very high share of single-person households
(Figure 7), and the share of elderly in the total population is lower than in some of its European
peers (Figure 8). This implies that other countries with less favorable demographics that apply
the Swedish containment strategy could have even worse health outcomes.

Figure 7. Distribution of Single Person Figure 8. Elderly Population

Households
(2018, percent)
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
Denmark

EU

UK
Norway

France
Finland

Italy
Sweden

Belgium

Spain
Germany

Source: Eurostat EU-SILC survey

The Covid-19 deaths appear to be concentrated among certain segments of the population.
While those aged 65 or older represent close to 20 percent of the population, they account for
more than 88 percent of Covid-19 deaths in Sweden, with the remaining deaths primarily
concentrated in the 50 and older age group (Figure 9). In other Nordic countries including
Denmark, Finland, and Norway, the elderly account for an almost identical share of deaths. It
is also important to note that about half of Sweden’s deaths were in elderly nursing facilities.
Both Denmark and Sweden issued a recommendation not to visit nursing homes in early
March, but, compared to Finland and Norway, a legally binding ban was only implemented
later in Sweden (see Figure 10). A government inquiry on how Sweden handled the

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 12

pandemic may bring more clarity about the underlying causes of the high death rates in
elderly nursing facilities, including whether the timing of the ban played any role. 9

Figure 9. Sweden: Death by Age Groups

(in percent of total Covid-19 deaths)
100.0
89.2
90.0

80.0

70.0

60.0

50.0

40.0

30.0

20.0
9.6
10.0
1.2
0.0
Age 0-49 Age 50-69 Age 70-90 and above
Sources: Swedish Institute of Health

Figure 10. Timeline on Restrictions to Visit the Elderly

Denmark: Recommendation to avoid visiting nursing homes

Nordics

Norway: Ban on visiting vulnerable groups

Finland: Ban on visiting nursing homes

11 12 16 17 March 30 April
Sweden

Ban on visiting nursing homes (effective April 1)

Recommendation to avoid visiting nursing homes

There is also a heavy concentration of deaths among the residents of the Stockholm region.
The Stockholm region accounts for half of Sweden’s Covid-19 deaths but only for 23 percent
of the country’s population. The number of deaths per capita in Stockholm remains below
Madrid and New York City as of mid-August (Figure 4). Within the Stockholm region, the
death toll was also unevenly distributed, possibly reflecting economic inequities and social
patterns. There appears to be a correlation between the number of reported deaths and the
share of the residents that were either born abroad or are children of parents born abroad

9
See press release on inquiry for more information: https://www.government.se/press-releases/2020/06/mats-
melin-to-chair-covid-19-inquiry-in-sweden/

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 13

(Figure 11), in line with the findings of the Public Health Agency. 10 Likewise, there appears
to be a correlation between the number of reported deaths and household size, indicating that
demographic characteristics matter (Figure 12).

Figure 11. Covid-19 Deaths and Population Figure 12. Covid-19 Deaths and Household
with Foreign Background Size

18
Number of cumulative reported COVID-19

18

Number of cumulative reported COVID-19

deaths per 10,000 inhabitants (persons)

deaths per 10,000 inhabitants (persons)

16
16 Stockholm city (excluded from the regression line)
14 Stockholm City (excluded from the regression line)
14
12 12
10 10
8 8
6
6
4
4
2
0 2
20 40 0
60 80 100 0
Foreign background (percent of total population) 2.1 1.5 2.41.8 2.7 3
Notes: Foreign background population includes foreign-born citizens and those who Average household size (persons)
were born in Sweden with two foreign parents. Data points include 26 municipalities Notes: Data points include 26 municipalities in Stockholm region (including
in Stockholm region (including Stockholm city) and 14 districts in Stockholm city. Stockholm city) and 14 districts in Stockholm city. Data as of May 31st 2020.
Death data as of May 31st 2020. Sources: Statistics Sweden, Region Stockholm and Stockholms Stad.
Sources: Statistics Sweden, Region Stockholm and Stockholms Stad.

Although the authorities did not include herd immunity as a goal of their containment
strategy, they reasoned that it could be achieved as a by-product of the strategy. The
authorities’ SEIR model predicted that 26 percent of the population in the Stockholm region
will have been infected with Covid-19 by May 1, but it remains unclear whether herd
immunity is being established. First, while empirical estimates vary, antibody rates are likely
low. 11 A study on antibodies in blood donors that was published by the Swedish Public
Health Authority in mid-June suggests that 6.3 percent of the blood donors had antibodies to
the virus by end-May, with the highest share, 10 percent, found in the Stockholm region. 12 It
is unclear to what extent these results can be generalized to the entire population. Other
countries which had stricter containment measures seem to have achieved antibody rates that
are not far off from those reported in the Swedish study. For example, a recent antibodies
study in Spain found that 5 percent of the population tested positive. 13 A subsequent survey
that was conducted by the Swedish authorities end-June and published in September found
that in a district of Stockholm 18 percent of the population developed antibodies. 14 However,
some studies have shown that a form of immunity can exist without creating antibodies

10
According to the Public Health Agency of Sweden (2020a), the incidence of confirmed cases was generally
higher among people born outside Europe. Similar results have been reported for Denmark; see
https://files.ssi.dk/COVID19-epi-trendogfokus-07052020-4eu7
11
Public Health Agency of Sweden (2020b)
12
https://sverigesradio.se/sida/artikel.aspx?programid=2054&artikel=7498860
13
https://www.reuters.com/article/us-health-coronavirus-spain-idUSKBN22P1IK
14
https://www.folkhalsomyndigheten.se/nyheter-och-press/nyhetsarkiv/2020/september/nya-resultat-om-
antikroppar-mot-covid-19-i-olika-grupper-i-befolkningen/

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 14

(Sekine et al, 2020). Second, whether infection with the virus generates sufficient protective
immunity is still under investigation (Randolph et. al, 2020).

Sweden’s experience highlights the challenge of protecting the vulnerable in the absence of a
more stringent lockdown, but the exact effects on the number of deaths are unclear, and the
relatively high number of infections and deaths could also be attributed at least partially to
other factors. For example, Conyon et al. (2020) compare Denmark’s and Norway’s
lockdown measures in a difference-in-difference model to Sweden’s approach and find that
stricter lockdown policies in Sweden would have been associated with fewer Covid-19
deaths. By contrast, Born et al. (2020) use a synthetic counterfactual for Sweden which is a
weighted average of other European countries to conclude that Covid-19 infections and
deaths would not have been significantly different in Sweden under a lockdown.

IV. CONTAINMENT POLICIES AND ECONOMIC OUTCOMES

A. Aggregate Economic Activity

In this subsection, we consider the effects of containment policies on aggregate economic
activity. The growth outcome in the first quarter of 2020 provides some initial clues about the
effects of containment policies on economic activity. By August 2020, all advanced economies
have released first-quarter GDP data, although some of the estimates could be substantially
revised. The simple average of these estimates is a QoQ GDP decline of 2.5 percent.

Sweden did not record an economic contraction in the first quarter of 2020 contrary to almost
all other advanced economies. 15 2020:Q1 data show positive GDP growth of 0.2 percent
QoQ. While this was mainly driven by exports (Figure 13), private consumption and
investment (which can be assumed to be more affected by containment policies) also
performed better than in most advanced economies, although they experienced negative
growth (Figure 14). Obviously, the first quarter coincides with the onset of the crisis, which
started to intensify only in March.

Figure 13. GDP QoQ Growth, 2020:Q1 Figure 14. Private Consumption and
Investment, 2020:Q1
(Percent, seasonally adjusted, non-annualized) (Contributions to QoQ growth; percent, seasonally adjusted, non-annualized)
2 4
3
0
2
-2 1
0
-4
-1

-6 -2
-3
-8
-4

-10 -5 Private Consumption Gross Fixed Capital Formation

-6
-12
JPN

SWE
CYP

ITA
FRA
GBR
FIN

NLD
USA

CAN

ISL

ISR
NZL

SVN
DNK
GRC
PRT
AUS

DEU

AUT

KOR

BEL
ESP
NOR*
NLD
AUS
LTU

SGP

USA
KOR

DNK

AUT
TWN

CYP

GRC

HKG
SWE

ITA
ISR

CZE
JPN

BEL
NOR*

ISL
FRA
IRL

NZL

FIN

DEU
CAN
GBR

LVA

EST

SVN
SVK
ESP

CHN
CHE

LUX

PRT

Note: *Mainland.
Note: *Mainland.
Sources: National authorities; IMF Staff calculations.
Sources: National authorities.

15
The other exception was Ireland.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 15

2020:Q1 growth outcomes across advance economies appear to be correlated with the length
and intensity of containment measures. More stringent containment measures are associated
with larger output losses: The correlation coefficient between the average level of the Oxford
Stringency Index (which reflects both the number of days containment measures were in
place and their intensity) and output growth in the first quarter of 2020 is about
-0.6 (Figure 15). Using alternative samples and measures of GDP growth as robustness
checks in the Appendix I, we find statistically significant correlation coefficients ranging
from -0.4 to -0.6, implying that the finding is robust. However, there remain several caveats,
and it is too early to draw definitive conclusions. There is a large variation in growth
outcomes across advanced economies that cannot be explained by differences in containment
strategies alone, and a range of other factors, including infection rates, supply chain
disruptions or declining exports, may matter as well. The sample size is small, and hence it is
not possible to control for confounding unobserved effects. 16

Figure 15. Stringency of Containment and Figure 16. Economic Activity Indicators
GDP QoQ Growth, 2020:Q1
(Index; percent, seasonally adjusted) (Index Jan 2020=100, seasonally adjusted)
2 105

SWE
0 100

-2 95

-4 90
GDP growth

-6
85

-8 Correlation -0.6
80

-10
75

-12
SWE NOR GBR FIN
0 10 20 30 40 50 60 70
Jan-Mar 2020 Average Stringency Index Jan-20 Feb-20 Mar-20 Apr-20 May-20
Sources: National authorities, Oxford COVID-19 Government Response Tracker; IMF Staff calculations. Sources: National authorities; IMF Staff calculations.

It remains unclear whether any economic benefits from avoiding a lockdown have been
subsequently eroded by relatively weaker economic activity. Sweden’s monthly activity
indicator suggests that the economic contraction accelerated in April and continued (albeit at
a slower pace) in May, which stands in contrast to Norway, the UK, and (to a lesser extent)
Finland, where activity appears to have rebounded in May (Figure 16). These patterns would
be consistent with the view that Sweden’s high infection rates in May relative to other
countries have depressed economic activity.

The GDP contraction in 2020:Q2 was significantly larger in Sweden than in other Nordics,
supporting this view (Figure 17). However, this difference was largely driven by a decline in
exports rather than in private consumption and investment. In addition, the cumulative fall of
GDP in the first six months of 2020 has been smaller in Sweden compared to other Nordic
countries (Figure 18). These data could still be revised, and it is important to note that other
factors may have been at play.

16
It may also be worthwhile to repeat this exercise once data for 2020:Q2 are available, given that many
advanced economies implemented containment measures only at the end of 2020:Q1.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 16

Figure 17. GDP QoQ Growth, 2020:Q2 Figure 18. GDP HoH Growth, 2020:H1
(Percent, seasonally adjusted, non-annualized) (Percent, seasonally adjusted, non-annualized)

0 0

-1
-1
-2

-3 -2

-4
-3
-5

-6 -4
-7
-5
-8

-9
-6
FIN NOR* DNK SWE
SWE FIN DNK NOR*
Note: *Mainland. Note: *Mainland.
Sources: National authorities. Sources: National authorities; IMF Staff calculations.

B. Sectoral Economic Activity

The impact of containment policies appears to have been heterogeneous across sectors. The
latest manufacturing data show similar trends across the Nordics and in the Euro Area as a
whole, mainly reflecting a drop in external demand and disruptions in supply chains.
Sweden’s manufacturing sector, which has not been directly constrained by domestic social
distancing and containment measures, has been impacted by the external environment. In
April, the PMI manufacturing index sunk by 17 points compared to February, broadly in line
with Germany and the Euro Area average (Figure 19). As of July 15, the manufacturing
sector accounted for 30 percent of all temporary layoffs, although it only provides 15 percent
of total employment (Figure 20).

Figure 19. Purchasing Managers Index: Figure 20. Sweden: Short-term Layoffs and
Manufacturing Employment Share by Sector
(Seasonally adjusted)
(Percent of total layoffs; percent of total employment)
55
40 16
Short-term layoffs 14
50 30 Employment share (RHS) 12
10
45 20 8
6
40 10 4
2
0 0
35
& support services

& communication

Education

Finance & insurance

Professional, scientific

Transportation
Accommodation
Wholesale & retail trade;

Human health

Real estate activities

Agriculture, forestry
Manufacturing

& mining, quarrying

Construction
& social work
& food services

& technical activities

Supply of energy
Administrative
& storage

Information
repair of vehicles

& fishing

30
SWE NOR DNK Euro Area DEU FRA ITA

25
Jan-20 Feb-20 Mar-20 Apr-20 May-20 Jun-20 Notes: Layoffs data as of July 15, 2020. This chart only includes the largest 14 sectors by 2018
employment.
Sources: IHS Markit, Swedbank and Haver Analytics,
Sources: Swedish Agency for Economic and Regional Growth and Eurostat.

By contrast, at the peak of the lockdown, the services sector has been less severely affected
in Sweden compared to international peers, likely reflecting both the differences in
mitigation strategies and the sectoral composition of the services sector. Figure 21 shows that
the services sector’s PMI fell by 17 points between February and April, while the decline in
Germany, France and Italy was much larger (there is no data for the other Nordic countries).

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 17

Hotel revenue declined less than in other countries, further confirming that the non-tradable
sectors are more directly affected by the severity of containment policies (Figure 22).
Figure 21. Purchasing Managers Index: Figure 22. Europe: Hotel Revenues in
Services March
(Seasonally adjusted) (Percentage changes between 2020 and 2019)
60 0

-10
50
-20

40 -30

-40
30
-50
20 -60

-70
10
SWE Euro Area DEU FRA ITA -80
0 FRA CHE ESP AUT DEU BEL PRT IRL GBR SWE*
Jan-20 Feb-20 Mar-20 Apr-20 May-20 Jun-20 Notes: Data are not perfectly comparable. Source of Sweden's data is different from
that of the other countries.
Sources: IHS Markit, Swedbank and Haver Analytics. Sources: IMF STR Hotel Data and Swedish Growth Agency.

As Figure 23 shows, the structure of Sweden’s economy makes it less vulnerable to the
pandemic and containment measures. Its share of the hospitality and recreation sectors
(which is likely to be most affected by the pandemic) is relatively small, similar to Nordic
peers, but in contrast to other European countries. 17 In addition, the country’s reliance on
exports is almost as strong as Germany making it more vulnerable to fluctuations in external
demand and global supply chain disruptions.

Figure 23. Trade Openness and Hospitality Industry

17
Other factors matter for vulnerability as well. The international travel restrictions were imposed on Sweden
for a relatively long period and may have limited the number of foreign visitors, although increased domestic
tourism may have partially compensated for that if people in Sweden travelled less abroad. In addition, Sweden
is among countries with the highest share of jobs that can be performed from home, thereby increasing
resilience to containment measures. Dingel and Neiman (2020) estimate that in Sweden 44 percent of jobs can
be done from home, the third-highest rate in the world after Luxembourg and Switzerland.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 18

C. Job Seekers and Beneficiaries of Short-Term Work Schemes

Newly available weekly data on the number of registered job seekers in Nordic and other
countries provide valuable insights into the impact of the crisis and containment policies.
Such data are typically constructed from applications for social assistance and may therefore
deviate from Labor Force Survey (LFS) estimates of the unemployment rate which are
normally released at a monthly frequency and may not capture short-term or temporary
spikes in the number of job seekers. These data come with caveats, including that they are
subject to revisions, not seasonally adjusted and possibly inaccurate: In times of a severe
shock, government agencies may be slow in processing applications, thereby understating the
true extent of the number of job seekers. The latter may also be overstated if some
applications are invalid, for instance if some individuals submit multiple applications. That
said, weekly register-based figures of job seekers are one of the few high frequency
indicators that directly measures a key macroeconomic outcome and has attracted much
interest, including in the U.S. and other countries.

The increase in the number of jobseekers in Denmark, Norway, and Sweden (for which
comparable weekly data are available) has followed different trends (Figure 24). In Sweden,
the increase in job seekers has been somewhat slower, but contrary to Denmark and Norway,
where the peak appears to have been reached, Sweden remained on a steady upward trend as
of mid-July. In Sweden, the increase in jobseekers between early March and June has been
almost 2 percentage points of the labor force in contrast with Denmark and especially
Norway, where it has been much lower. This figure excludes employees that were given
dismissal notices

Figure 24. Jobseekers Figure 25. Beneficiaries of Short-term Work

Schemes
(Percent of 2019 labor force; register-based; weekly frequency)
10
SWE DNK NOR

0
15-Mar
22-Mar
29-Mar

10-May
17-May
24-May
31-May
7-Jun

5-Jul
14-Jun
21-Jun
28-Jun
5-Apr
1-Mar
8-Mar

12-Apr
19-Apr
26-Apr
3-May

Notes: Partially based on estimates.

Sources: National authorities; World Bank; IMF staff calculations and estimates.

However, ignoring workers that receive wage subsidies understates the adverse effects of the
crisis on the labor market. Norway, Denmark, and Sweden have implemented wage subsidy
schemes whereby firms can reduce the wage of employees that they pay, and the government

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 19

pays a subsidy that covers part of the reduction of the wage. 18 While the implementation
differs in administrative terms (e.g., in Sweden, the subsidy is paid to the employer, whereas
in Norway employees who are laid off temporarily on a part or full-time basis can apply for
partial or full unemployment assistance), the economics of the schemes are similar. 19 They
share common objectives: In addition to providing social protection by financing part of the
wages of employees who work less or not at all, they avert permanent job losses, and enable
firms to quickly start operating again as the crisis abates. Denmark, Norway, and Sweden
have periodically published detailed information on the number of beneficiaries of these
schemes. There are some gaps in the dataset, so some missing observations had to be
estimated, and comparability is not perfect given differences in the schemes. For example,
given that the reduction in the hours worked is different across countries, the fall in
employment cannot be compared across countries and be inferred from comparing the
number of beneficiaries.

The increase in the number of all beneficiaries of short-term work (for which rules differ
across countries) as a share of the labor force amounts to around 10 percentage points in
Sweden (Figure 25). 20 This figure still excludes applications for short-term work not yet
processed. The initial increase in the number of job seekers and beneficiaries of short-term
work schemes has been somewhat slower in Sweden than in Norway, possibly because a new
short-term work scheme had only been implemented by early April, somewhat later than in
Denmark and Norway. 21 Of course, at least some of the increase in the number of
beneficiaries of short-term work schemes is temporary as the latter are generally expected to
eventually return to work.

D. Mobility Indicators

Mobility indicators, which are based on cell phone data, are increasingly being used to track
the effects of the pandemic. They measure the percentage change in mobility relative to a
pre-determined baseline and are provided by the Apple Mobility Trends Reports and Google
Covid-19 Community Mobility Reports. The indicators we use for the purpose of this paper
include percentage changes in visits of workplaces, retail and recreation outlets, and transit
stations provided by Google relative to the median value of the January 3–February 6 period

18
For Denmark, we have only considered the wage compensation scheme which has by far most beneficiaries.

19
LFS unemployment statistics typically do not or not fully capture beneficiaries of short-term work schemes.
See Riksbank (2020a) for a discussion.
20
Another factor that potentially affects employment statistics by reducing the labor force and
employment/unemployment numbers could be early retirement as pointed out by IMF (2020). The authors find
that economic crises are typically associated with significant acceleration in the numbers opting for early
retirement as well as disability claims. However, monthly data on early retirement applications for Denmark in
March and April 2020 were in line with the same months in 2019 and did not suggest a surge in applications.

21
Other countries show broadly similar patterns. New Zealand, which had at least temporarily one of the
strictest lockdowns among all advanced economies, also publishes weekly unemployment. While the number of
jobseekers increased only slowly and remained constant from around mid-May, the number of beneficiaries of
wage subsidies increased much more rapidly.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 20

in 2020. We also use the changes in requests for driving and transit directions relative to
January 13, 2020 provided by Apple.

Compared to peers, the decline in mobility in Sweden was milder during the most severe
lockdown period between mid-March and mid-April (Figure 26). Also, presence at
workplaces and at retail and recreation outlets declined by much less in Sweden during this
period. Transaction-level data from Swish, Sweden’s leading mobile payment provider,
which is a popular means of payment for commuter services show a significant decline in
payments for travel services. 22 Mobility indicators are certainly not a perfect measure of
broad-based economic activity as at least some jobs continue to be performed irrespective of
containment measures (for instance in critical areas, or those jobs that are suitable for
telework). There is actually little correlation between the weekly number of job seekers and
overall weekly employment. However, there is a strong correlation for example between
weekly credit card purchases of services and visits to workplaces, indicating that mobility
indicators at least partially measure economic activity (Figure 27).

Figure 26. Nordics: Changes in Mobility Figure 27. Nordics: Workplace Mobility
and Card Transactions for Services
(Percentage change relative to baseline, March 13 to April 12)
20
0
Card Transactions for services (% y-y changes)

-10
0
-20
-20
-30

-40 -40
-50
-60
-60
Nordics SWE
-70 -80
work retail stations driving transit Swish
Notes: 'Nordics' refers to the average of DNK, FIN and NOR. 'Swish' reflects the
change in the number of payments for travel-related services using the Swedish -100
instant payment system 'Swish'. The baseline generally reflects data from early -80 -40 -20 -60 0 20
2002 but differs by indicator. Sources: Apple, Google, Swish and IMF Staff Workplace mobility (% changes)
calculations. Sources: Nordea, Google Mobility Trend Report and IMF staff calculations.

We run regressions at the country-day level to examine the correlation between containment
policy and cell phone-based mobility:

𝑀𝑀𝑖𝑖,𝑡𝑡 = 𝛽𝛽𝐶𝐶𝑖𝑖,𝑡𝑡 + 𝛿𝛿𝑍𝑍𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡

where 𝑀𝑀𝑖𝑖,𝑡𝑡 , 𝐶𝐶𝑖𝑖,𝑡𝑡 and 𝑍𝑍𝑖𝑖,𝑡𝑡 denote the percentage change in cell phone-based visits to
workplace mobility, the stringency of containment policy as measured by the Oxford Covid-
19 Government Response Tracker and a range of control variables, including the number of
infections and deaths, whether there is income support and other relief from governments,
and whether there is comprehensive testing and contact tracing available (see Appendix II for
details), respectively, and where indices i and t refer to the country and day. All regressions

22
Cell phone-based mobility data come with caveats. They are not necessarily representative or comparable
across countries, so they are only indicative.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 21

control for unobserved county-week and country-day-of-week effects (to capture effects of
country-specific holidays for instance). Our baseline sample includes 24 EEA countries for
which data are available and the UK.

The results show that the stringency of containment policies has significant effects. In our
baseline specification (first column in Table 2), an increase from 0 (no containment) to
100 (maximum containment) results in a decline of more than 50 percentage points in the
presence of people at workplaces. Of course, this is a hypothetical scenario as only
5 countries within the EEA and no country in the Nordic region scored values above 90 in
the stringency of containment at any point in time. The coefficient estimate is robust to
different samples. In specifications 2 and 3, we only use Nordic countries and advanced
economies, respectively, whereas in specification 4, we use all countries for which data are
available.

Interestingly, when only including Nordic countries, the predicted decline in mobility as a
result of moving from a scenario with no containment to one with the maximum level of
containment increases (specification 2). This is plausible as in Nordic countries, it is likely
that a relatively large share of jobs can be performed from home so that people are more
willing to follow containment measures, and there is plausibly a stronger tradition of abiding
by government rules and recommendations (specification 2).

The magnitude and the signs of coefficients of the control variables that are significant are
plausible. The number of new infections and Covid-19-related deaths both dampen mobility.
The availability of income support enables individuals to stay at home, implying that the
negative sign of the coefficient is plausible as well. The coefficients on the availability of
financial relief, testing and tracing are not significant (and the sign is not robust across
specifications).

The results also imply that presence at workplaces decreased in Sweden by more than what
would be expected from the regressions. When considering the approximate period of the most
severe lockdown in the Nordic region as a whole, from March 13 until April 12, the average
decrease in presence at retail and recreation outlets was somewhat larger in Sweden (by up to
8 percentage points which amounts to a third of the actual decline depending on the
specification) than what would be expected from the regression (Figure 28). This suggests that
individuals decided to voluntarily scale back certain activities above and beyond official
recommendations and requirements that are reflected by the Oxford Covid-19 Government
Response Tracker and implied by the other explanatory variables.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 22

Table 2. Effects of Containment on Visits of Workplaces

(1) (2) (3) (4)
VARIABLES workplaces workplaces workplaces workplaces

Containment policy (index) -0.510*** -0.612** -0.510*** -0.519***

(0.0543) (0.181) (0.0511) (0.0277)
New infections (7-day ave.) -1.076*** -5.413*** -1.000*** -0.420***
(0.254) (0.771) (0.233) (0.143)
New deaths (7-day ave.) -5.628*** 2.570 -5.659*** -6.228***
(1.609) (17.71) (1.628) (1.550)
Income support -5.875* 9.521 -5.309* -2.000
(3.125) (4.850) (2.909) (1.265)
Financial relief -0.159 -3.168 -1.326 -1.894
(2.751) (5.686) (2.673) (1.151)
Comprehensive testing 0.332 4.439 -0.309 -0.440
(1.534) (4.691) (1.966) (1.152)
Comprehensive tracing 3.029 -1.132 2.724* 0.0147
(1.835) (2.068) (1.554) (0.905)

Observations 4,089 580 4,396 17,841

R-squared 0.860 0.796 0.873 0.903
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
The dependent var is the percentage change in the cell-phone based visits of workplaces.
All specs include country-week and country-day-of-the-week FEs
Errors in brackets clustered at the country level.
Data as of July 16.
*** p<0.01, ** p<0.05, * p<0.1

Figure 28. Sweden: Changes in Visits to Workplaces

(Percentage change rel. to baseline; actual vs. regresion-based prediction)

-5

-10

-15

-20

-25

-30
Actual Prediction (1) Prediction (2) Prediction (3) Prediction (4)
Notes: 'Actual' refers to the actually observed change in visits to workplaces;
whereas the predictions are regression-based and refer to specifications (1) to (4).
Baseline refers to data from early 2020. Sources: Google and IMF Staff calculations.

Our baseline coefficient estimates are robust to using alternative mobility indicators. In Table 3,
specification (1), we use visits to transit stations; in specification (2), we use visits to retail and
recreation outlets; in specification (3), we use the Apple driving indicator; and in specification
(4), we use the Apple transit indicator (which is available for fewer countries). The coefficient
estimates are remarkably robust and even increase relative to our baseline estimate.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 23

Table 3. Robustness Checks

(1) (2) (3) (4)
VARIABLES stations retail driving transit

Containment policy (index) -0.561*** -0.742*** -0.654*** -0.754***

(0.0417) (0.0697) (0.0453) (0.0946)
New infections (7-day ave.) -0.836*** -0.621*** -0.529*** -0.479***
(0.150) (0.178) (0.128) (0.126)
New deaths (7-day ave.) -1.682 -3.173* -1.678 0.249
(1.468) (1.780) (1.221) (1.034)
Income support -5.479** -4.563 -3.922 -8.391*
(2.502) (2.865) (2.904) (4.289)
Financial relief -2.466 -2.544 -3.129 -6.808*
(1.797) (2.349) (1.927) (3.665)
Comprehensive testing 0.144 1.222 -2.291* 0.0214
(0.806) (1.516) (1.182) (1.962)
Comprehensive tracing 2.983* 1.819 1.464 6.153*
(1.574) (1.723) (2.073) (3.407)

Observations 4,089 4,089 4,019 2,453

R-squared 0.958 0.946 0.977 0.979
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
The dependent var is the percentage change in cell-phone based mobility indicators.
All specs include country-week and country-day-of-the-week FEs
Errors in brackets clustered at the country level.
Data as of July 16.
*** p<0.01, ** p<0.05, * p<0.1

V. CONCLUSION AND POLICY IMPLICATIONS

Our preliminary findings indicate that at the time when other countries imposed a strict
lockdown in March and April, economic activity appears to have declined less in Sweden.
Sweden’s private consumption and investment in 2020:Q1 fell only slightly, in contrast to
more substantial declines in most other advanced economies. Sweden also experienced a
smaller decline in mobility indicators and in the service sector compared to other advanced
economies. Our analysis shows that these effects can be attributed to differences in
containment strategies between Sweden and its peers.

However, at the time when other countries started opening up their economies in 2020:Q2,
observed infection and death rates continued to be relatively high in Sweden, with potentially
adverse economic effects. In 2020:Q2, Sweden’s GDP fell by more than in Nordic peers, the
number of jobseekers and beneficiaries of short-term work schemes remained high and the
monthly economic activity indicator declined for longer than in international peers. However,
the cumulative fall of GDP in the first six months of 2020 has been smaller in Sweden
compared to other Nordic countries, and some of the decline in 2020:Q2 was driven by a fall
in exports.

In addition, based on evidence from Sweden and Nordic peers, our analysis also suggests that
other factors were important. Mobility in Sweden may have decreased more than what would

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 24

be expected from the relaxed de jure containment measures, suggesting that there was
voluntary practice of social responsibility and distancing. In addition, external demand from
trading partners and supply chain disruptions impact output regardless of the severity of the
containment policies. This is reflected in the sharp contraction in manufacturing activity and
the high share of its employees on short-term work. Finally, other structural features of the
economy such the relative importance of the hospitality industry and the share of jobs that
can be performed from home determine the economic effects of the containment strategy as
well.

Looking ahead, it remains unclear how Sweden’s containment strategy and its avoidance of a
strict lockdown in particular will play out over the medium term. 23 Despite the relatively
benign 2020:Q1 GDP outcome, almost all observers project a decline for the year that is
similar to other Nordic countries. Any final verdict will also depend on whether, as a by-
product of its approach, Sweden will be closer to achieving herd immunity, thereby
increasing its resilience in the event of another wave of infection. While medical knowledge
about Covid-19 is still accumulating, immunity gains—although on the rise—have been
lower than initially projected.

The large number of deaths, particularly among elderly and in areas with a higher share of
persons that are foreign-born or have foreign-born parents highlights the challenge of
protecting the vulnerable in the absence of a more stringent lockdown, even in a country with
favorable socio-demographic characteristics, including a very high share of single-person
households. The outcome could have been worse in countries that may replicate Sweden’s
strategy but have different demographics, resources, or history of abiding by social contracts.
Swift decisive macroeconomic policy action remains critical to avert more dire economic
outcomes. Sweden’s policy response to combat the economic impact of the pandemic has
been prompt, large, and well-designed. Given its comfortable fiscal space, Sweden was in a
good position to provide timely and substantial support to companies and households through
various compensation schemes, guarantees, and tax deferrals. 24

Future work could examine the effects of containment strategies in greater detail, including
any differential effects of particular measures. The distributional effects of social distancing
in Sweden and other Nordics could also be further explored. The evidence so far implies
highly unequal effects. Recent research in this area include Furceri et al. (2020), who show
that past epidemics led to increases in inequality; and Alstadsæter et al. (2020), who use

23
Riksbank (2020b) notes that the relationship between GDP growth in Sweden and that of its major trading
partners may deviate from the pre-Covid patterns even in the longer run—in ways either more or less beneficial
from a Swedish perspective.
24
The IMF website provides a comprehensive list of fiscal, monetary and macro-financial measures that were
implemented across the globe. See https://www.imf.org/en/Publications/Policy-
Papers/Issues/2018/06/15/pp041118 assessing-fiscal-space for a discussion on the concept of fiscal space.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 25

register-based data from Norway, show that temporary layoffs affected financially vulnerable
parts of the population to a relatively large extent.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 26

REFERENCES

Acemoglu, D., Chernozhukov, V., Werning, I., & Whinston, M. D. (2020). A multi-risk SIR
model with optimally targeted lockdown (NBER Working Paper No. 27102).

Alstadsæter, A., Bratsberg, B., Eielsen, G., Kopczuk, W., Markussen, S., Raaum, O., &
Røed, K. (2020). The first weeks of the coronavirus crisis: Who got hit, when and
why? Evidence from Norway (NBER Working Paper No. 27131).

Andersen, A. L., Hansen, E. T., Johannesen, N., & Sheridan, A. (2020). Pandemic, shutdown
and consumer spending: Lessons from Scandinavian policy responses to COVID-19.
arXiv preprint arXiv:2005.04630.

Aum, S., Lee, S. Y. T., & Shin, Y. (2020). COVID-19 doesn't need lockdowns to destroy
jobs: The effect of local outbreaks in Korea (NBER Working Paper No. 27264).

Avery, C., Bossert, W., Clark, A., Ellison, G., & Ellison, S. F. (2020). Policy implications of
models of the spread of coronavirus: Perspectives and opportunities for economists
(NBER Working Paper No. 27007).

Baker, S. R., Farrokhnia, R. A., Meyer, S., Pagel, M., & Yannelis, C. (2020). How does
household spending respond to an epidemic? consumption during the 2020 covid-19
pandemic (NBER Working Paper No. 26949).

Born, B., Dietrich, A., & Müller, G. J. (2020). Do lockdowns work? A counterfactual for
Sweden. Covid Economics, 16, 1-22.

Cajner, T., Crane, L. D., Decker, R. A., Grigsby, J., Hamins-Puertolas, A., Hurst, E., Kurz,
C., & Yildirmaz, A. (2020). The US labor market during the beginning of the
pandemic recession (NBER Working Paper No. 27159).

Chen, S., Igan, D., Pierri, N., & Presbitero, A. F. (2020). Tracking the economic impact of
COVID-19 and mitigation policies in Europe and the United States (IMF Working
Paper 20/125).

Conyon, M. J., He, L., & Thomsen, S. (2020). Lockdowns and COVID-19 deaths in
Scandinavia. Covid Economics, 26, 17-42.

Deb, P., Furceri, D., Ostry, J. D., & Tawk, N. (2020a). The economic effect of containment
measures. Covid Economics, 24, 32-75.

Deb, P., Furceri, D., Ostry, J. D., & Tawk, N. (2020b). The effect of containment measures
on the COVID-19 pandemic. (IMF Working Paper 20/159).

Dingel, J., & Neiman, B. (2020). How Many Jobs Can be Done at Home? (NBER Working
Paper No. 26948).

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 27

Eichenbaum, M., Rebelo, S., & Trabandt, M. (2020). The macroeconomics of testing during
epidemics (NBER Working Paper No. 27104).

Ellison, G. (2020). Implications of heterogeneous SIR models for analyses of Covid-19

(NBER Working Paper No. 27373).

Fang, H., Wang, L., & Yang, Y. (2020). Human mobility restrictions and the spread of the
novel coronavirus (2019-ncov) in China (NBER Working Paper No. 26906).

Favero, C. A., Ichino, A., & Rustichini, A. (2020). Restarting the economy while saving lives
under Covid-19 (CEPR Discussion Paper No. 14664).

Furceri, D., Loungani, P., Ostry, J. D., & Pizzuto, P. (2020). Will Covid-19 affect inequality?
Evidence from past pandemics. Covid Economics, 12, 138-157.

Hale, T., Webster, S., Petherick, A., Phillips, T., & Kira., B (2020). Data use policy: Creative
Commons Attribution CC BY standard. Oxford COVID-19 Government Response
Tracker, Blavatnik School of Government.

IMF (2020). Pension schemes in the COVID-19 crisis: Impacts and policy considerations.
Special Series on COVID-19, July.

Kermack, W. O., & McKendrick, A. G. (1927). A contribution to the mathematical theory of

epidemics. Proceedings of the royal society of London. Series A, Containing papers
of a mathematical and physical character, 115(772), 700-721.

Krueger, D., Uhlig, H., & Xie, T. (2020). Macroeconomic dynamics and reallocation in an
epidemic. Covid Economics, 4, 21-55.

Maloney, W. F., & Taskin, T. (2020, May 15). Voluntary vs mandated social distancing and
economic activity during COVID-19. VoxEU. https://voxeu.org/article/covid-social-
distancing-driven-mostly-voluntary-demobilisation

Public Health Agency of Sweden (2020a). Demographic description of confirmed COVID-19

cases in Sweden March 13 – May 7, 2020.

Public Health Agency of Sweden (2020b). Estimates of the peak-day and the number of
infected individuals during the COVID-19 outbreak in the Stockholm region, Sweden.

Public Health Agency of Sweden (2020c). Helgenomsekvensering av svenska SARS-CoV-2

som orsakar covid-19.

Randolph, H. E., & Barreiro, L.B. (2020). Herd Immunity: Understanding COVID-
19. Immunity 52, no. 5, 737-741.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 28

Riksbank (2020a). Monetary Policy Report, April 2020.

Riksbank (2020b). GDP growth in Sweden relative to other countries in the wake of Covid-
19. Riksbank Economic Commentaries, June 12.

Sekine, T., Perez-Potti, A., Rivera-Ballesteros, O., Strålin, K., Gorin, J., Olsson, A., &
Llewellyn-Lacey., S.(2020). Robust T cell immunity in convalescent individuals with
asymptomatic or mild COVID-19. Cell, forthcoming.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 29

APPENDIX I. CORRELATION BETWEEN 2020:Q1 GROWTH AND CONTAINMENT POLICIES

Figure A.1. Stringency of Containment and Figure A.2. Stringency of Containment and
GDP QoQ Growth, 2020:Q1 GDP YoY Growth, 2020:Q1
(Index; percent) (Index; percent)
2
4
SWE
0
2
SWE
-2
0
Correlation -0.4 (China excluded)
-4
GDP growth

GDP growth
-2

Correlation -0.5 (China excluded)

-6
-4

-8 Correlation -0.6
-6 Correlation -0.6

-10 -8

-12 -10
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Jan-Mar 2020 Average Stringency Index Jan-Mar 2020 Average Stringency Index

Sources: National authorities, Oxford COVID-19 Government Response Tracker; IMF Staff calculations. Sources: National authorities, Oxford COVID-19 Government Response Tracker; IMF Staff calculations.

In this Appendix, we present a few alternative scatter plots and the associated correlation
coefficients of containment measures and GDP growth in 2020:Q1, using different sample
and YoY GDP growth. Excluding China from the sample weakens the inverse relationship
between growth and the stringency of containment, and the correlation coefficient between
QoQ growth and the stringency of containment increases to -0.4 but remains statistically
significant. This possibly reflects the strict enforcement of containment measures in China.
The YoY GDP growth and the stringency of containment measures seem to be more robust
to outliers, with the correlation coefficient increasing from -0.6 to -0.5 when China is omitted
from the sample.

APPENDIX II. DESCRIPTION OF VARIABLES USED FOR THE ECONOMETRIC ANALYSIS

Variable Description Source

Containment policy (index) Index measuring the stringency of Hale et al. (2020)
containment on scale from 0 to 100

New infections (7-day ave.) 7-day rolling average of new daily Johns Hopkins University
infections per 100,000 inhabitants

New deaths (7-day ave.) 7-day rolling average of new Covid-19- Johns Hopkins University
related deaths per 100,000 inhabitants

Income support Dummy; indicates whether government Hale et al. (2020)

provides any type of income support to
individuals
Financial relief Dummy; indicates whether the Hale et al. (2020)
government provides any type of relief
from financial obligations

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216
 30

Comprehensive testing Dummy; indicates whether there is at Hale et al. (2020)

least testing for anyone showing Covid-
19 symptoms
Comprehensive tracing Dummy; indicates whether there is Hale et al. (2020)
comprehensive contact tracing for all
identified cases
workplaces Percentage change relative to baseline in Google
visits to workplaces

stations Percentage change relative to baseline in Google

visits of transit stations
retail Percentage change relative to baseline in Google
visits of retail and recreation outlets

driving Changes in requests for driving Apple

directions relative to baseline

transit Changes in requests for transit Apple

directions relative to baseline

APPENDIX III. ANNOUNCED FISCAL, MONETARY AND MACRO-FINANCIAL POLICY

RESPONSES AS OF JUNE 2020
Fiscal Policies (% 2019 GDP) Monetary Policies Macro-Financial Policies
Release Easing Easing Advice agains
Discretionary Government Policy Rate Liquidity Quantitative Swap CCyB/Systemi Liquidity Amortization Dividend
Measures Guarantees Deduction Provision Easing Lines c Risk Buffer Buffers Requirement Payouts
Sweden 5.6 4.6 - x x x 2.5% x x x
Norway 5.2 4.3 125 x x x 1.5% x x x
Denmark 1/ 5.7 7.7 * x x 1.0% x x
Finland 2/ 2.9 5.4 ** x x x 1.0% x x x
Notes:
1/ Denmark is pegged to the Euro. * Danmarks Nationalbank (DNB) is selling/buying foreign currency when the exchange rate is deviating from its bands
In March the DNB increased interest rates by 15 bsp to defend the fixed exchange rate regime.
** Finland is part of the Euro Area with the ECB in charge of monetary policy.

©International Monetary
Electronic copy available at: Fund. Not for Redistribution
https://ssrn.com/abstract=3721216