Deutsche Bundesbank

Monthly Report
October 2021
43

The global economy during the coronavirus

pandemic

From the beginning of 2020 onwards, the coronavirus pandemic has been shaping economic
developments around the world. These have taken the form of an unprecedented downturn in
advanced and emerging market economies, a raft of measures to prevent the spread of infection,
and extensive monetary and fiscal policy support. Only once effective vaccines had been intro-
duced did a sustained recovery begin to take hold in many places. However, delivery delays and
shortages of key intermediate inputs are preventing this recovery from progressing smoothly.

Despite the global nature of the pandemic, some economies have pulled through the crisis better
than others. In a number of countries, such as the United States and China, economic output has
already returned to – or even significantly exceeded – its pre-​crisis level. Yet many economies,
including the four largest euro area Member States, are still lagging behind.

This heterogeneity is largely down to differences in the pattern of the pandemic and the meas-
ures taken to combat it. This article presents several empirical studies that examine these relation-
ships. Estimates show that workplace closures and stay-​at-​home requirements, for example,
strongly curbed mobility. Although this slowed the spread of the pandemic, it was accompanied
by major economic losses. Euro area countries which were hit particularly hard by the pandemic
and in which restrictions were stricter or in force for longer experienced sharper slumps in activ-
ity. Moreover, countries in which high-​contact services sectors are an economic mainstay proved
particularly vulnerable.

Policymakers did not limit themselves to directly combating the pandemic, but supported the
economy in many and varied ways. In the industrial countries, in particular, monetary and fiscal
policy accommodation significantly cushioned the immediate impact of the crisis. Simulation cal-
culations suggest that the cushioning effects were even greater in the United States than in the
euro area. In many places, specific measures also protected jobs, averted corporate insolvencies
and prevented turmoil in the financial system.

Since the beginning of this year, efforts to curb the pandemic in the long run have been focused
on vaccination campaigns. This could go a long way towards keeping the longer-​term damage
from the pandemic fairly limited in the advanced economies. In many developing and emerging
market economies, on the other hand, the recovery is being held back by slow progress in terms
of vaccination efforts. Over the next few months, the priority will be to push ahead with vaccin-
ation campaigns around the world – and not only for humanitarian reasons. Global economic
interconnectedness means that new waves of the pandemic in developing and emerging market
economies would also damage the advanced economies. Another task will be to phase out the
economic policy assistance measures as the pandemic recedes so as not to hamper the structural
change that the pandemic has made necessary.
 Deutsche Bundesbank
Monthly Report
October 2021
44

Introduction than widely expected.3 Global industrial pro-

duction and world trade had already exceeded
Pandemic as a Almost two years after the outbreak of the their pre-​crisis levels by the end of 2020. High-​
humanitarian
disaster …
coronavirus pandemic, large parts of the world contact services sectors, on the other hand,
remain firmly in its grip. The spread of the virus found it much harder to make up the ground
was first and foremost a humanitarian disaster. they had lost. The sporadic recovery in indus-
So far, over 240 million infections have been tries such as food services, the events industry
recorded around the world, with the real figure and tourism, in particular, was set back several
likely to be much higher. Almost 5 million times by measures to contain new waves of in-
people have lost their lives in connection with fection. Even so, in many countries, the dip in
the virus.1 It is only thanks to the rapid develop- general economic output for 2020 as a whole
ment of effective vaccines that recent waves of was much smaller than had been expected by
infection in the advanced economies have the International Monetary Fund (IMF) in June
been less severe. The situation remains sub- 2020, for example.4
stantially more difficult in many emerging mar-
ket economies and developing countries, No region of the world escaped the pandemic … but with
major regional
where a significant proportion of the popula- and its economic repercussions. Nonetheless, differences
tion will probably not be vaccinated until next some economies appear to have pulled through
year. the crisis better than others. Differences in the
severity of the slump in economic output were
… with enor- As long as vaccination coverage among the already visible in the first half of 2020. The sub-
mous global
economic
population is insufficient, reducing contact re- sequent recovery did not take place equally
repercussions­ mains the most effective way of curbing the swiftly or steadily in all countries, either. In a
spread of the highly infectious and dangerous number of countries, economic activity has al-
virus. Only very few governments have relied ready surpassed its pre-​crisis level once again.
completely on people adjusting their behaviour This includes China, in particular, which has
voluntarily, with most underpinning this with even returned to its original growth path. The
official measures instead. A kind of blueprint rapid ramping-​up of production capacity for
for such measures was provided by China, medical personal protective equipment and the
where the pandemic originated and where the export sector’s focus on certain consumer
rapid spread of the virus brought public life to
a virtual standstill from as early as the end of
January 2020. Just a few months later, coun- 1 Figures are based on data from Johns Hopkins University,
which analyses official national statistics on key data for
tries all over the world were closing businesses the pandemic. The actual spread of the pandemic and the
and schools and imposing stay-​at-​home re- number of victims whose lives it has claimed is probably
significantly underestimated. For instance, the World
quirements and travel restrictions. The upshot Health Organisation assumes that the number of
was a global economic downturn of historic coronavirus-​related deaths worldwide reported in 2020
was around 40% too low. See Dong et al. (2020) and
proportions. In the second quarter of 2020, World Health Organisation (2021).
global real GDP was around 10% below its pre-​ 2 The GDP data are based on an aggregate of 48 econ-
omies using market exchange rates. Between the third
crisis level. Even at the height of the global quarter of 2008 and the first quarter of 2009, the GDP of
financial and economic crisis of 2008-09, losses this group of countries only fell by 4%.
3 For instance, according to the expert dating of business
had not been not nearly as severe.2 cycles, which is widely regarded as an official source, the
recession in the United States came to an end as early as
April 2020. In the history of all business cycles since 1854,
Recovery rapid When the first wave of the pandemic receded this was by far the shortest economic downturn. See Na-
and strong at tional Bureau of Economic Research (2021).
first …
in many countries, restrictions were undone to
4 At the time, the IMF’s forecast had predicted a decline of
a degree over the course of the second quarter 4.9% in global GDP for 2020 as a whole. The latest calcu-
lations revised the decrease downward to 3.1%, mainly
of 2020. This initially caused the economy to owing to the rapid recovery in the second half of 2020. See
rebound strongly and at a more rapid pace International Monetary Fund (2020a).
 Deutsche Bundesbank
Monthly Report
October 2021
45

goods helped give its economy a major boost

Pandemic and economic indicators
(for more information, see the box on pp. 46 f.). for G20 countries*
In the United States, real GDP is now back to
just above its pre-​crisis level. By contrast, in the
G20 aggregate Individual G20
largest euro area Member States, it is still some countries
Monthly, lin scale
way off. 125
Weekly deaths related to COVID-19
per million inhabitants
100

75
Impact of the pandemic 50
on mobility and global 25
­economic activity 0
Stringency of government
Influence of One explanation for the mixed picture among containment measures1
government-​ 100
imposed and
countries could be differences in their re- Index points

self-​imposed sponses to the pandemic in terms of the meas- 80

changes in
behaviour …
ures taken and their duration. In March and 60
April 2020, many governments took swift ac- 40
tion against the spread of infection, imposing
20
far-​reaching restrictions on social and eco-
0
nomic life. Even until recently, some countries
Real GDP
implemented drastic measures in an attempt to Q4 2019 =100, quarterly, log scale
110
stop the virus from spreading. Others took a
100
less tough stance, for example because the
local infection figures allowed it, because they 90

relied more strongly on voluntary changes in

80
behaviour, or because additional restrictions 75
appeared too costly in light of the economic 2019 2020 2021
situation. The complex interactions between
Sources: Johns Hopkins University, Oxford COVID-19 Govern-
the pandemic situation, the measures taken, ment Response Tracker, national statistics and Bundesbank cal-
culations. * Excluding EU aggregate. GDP weighting at market
changes in behaviour, and economic activity exchange rates for restrictions and economic activity. 1 Values
of zero indicate no containment measures; values of 100 rep-
present a major challenge for empirical studies. resent strictest possible containment measures.
Although it seems plausible to assume that the Deutsche Bundesbank

sharp drop in GDP in Germany in the second

quarter of 2020 was the result of government mobility data are analysed as a chain in the
measures, other countries that took a less strict causal link between government-​imposed or
stance also experienced considerable declines. self-​imposed constraints, on the one hand, and
Even in the absence of administrative meas- economic output, on the other. For instance,
ures, infection rates brought about noticeable the data on the movement patterns of mobile
changes in behaviour.5 phone users show significant shifts around

… can be ana- One way to overcome these difficulties and

lysed using high-​ 5 In Sweden, which initially imposed very few administra-
frequency data
identify cause-​effect relationships is to analyse tive measures, relying instead on behavioural guidelines,
GDP shrank by just over 8% in the second quarter of 2020
high-​frequency data. In actual fact, daily infec-
and thus by only slightly less than Germany’s GDP.
tion figures are available for almost all coun- 6 Inputs to the index calculations include government re-
strictions on schools, businesses, public transport, events
tries. The scale of the constraints owing to gov-
and gatherings as well as various restrictions on the mobil-
ernment containment measures can be ap- ity of citizens. Index values of zero indicate no containment
measures, while values of 100 represent the strictest pos-
proximated using a stringency index developed sible containment measures. For a description of the index,
by the University of Oxford.6 High-​frequency see Hale et al. (2021).
Deutsche Bundesbank
Monthly Report
October 2021
46

The reasons for the Chinese economy’s comparatively good

performance in the pandemic
China, the country in which the pandemic The decisive reason why Chinese exports
originated, had already endured a massive performed remarkably well lay in the ability
decline in economic activity at the begin- of Chinese industry to deliver quickly and in
ning of 2020. It recovered surprisingly large quantities those goods for which de-
quickly from this slump, which affected sev- mand picked up on account of the pan-
eral services sectors as well as the industrial demic. Such goods initially included medical
sector. Whilst other countries’ economies personal protective equipment. In addition,
were getting caught up in the downward the transition of many employees to work-
spiral of the pandemic, China’s real gross ing from home caused a global spike in
domestic product had already returned to additional demand for IT equipment. More-
pre-crisis levels by the second quarter of over, global consumer demand shifted as
2020. Shortly thereafter, it even returned to well: owing to containment measures or
its original growth trajectory. voluntary changes in behaviour, households
strongly curbed their consumption of high-
The Chinese authorities’ rigorous contain- contact services but conversely increasingly
ment policy accounted for a material share acquired goods such as electronic devices
of the dynamic recovery; it brought infec- or furniture. Chinese exporters, which are
tion counts down quickly and sustainably. specialised in consumer goods, therefore
High-contact services, in particular, were benefited enormously from this.
the beneficiaries, whereas industrial pro-
duction recovered rapidly on the back of, In order to quantify the significance of the
above all, foreign business.1 In 2020, Chi- product range effect for China’s successful
na’s goods exports (on a US dollar basis) export performance, we decompose Chi-
picked up by 3½% even though global im- nese exports for the past two years into ap-
port expenditure dropped by around 6% in proximately 5,000 product groups and
the same year.2 compare each to global exports.3 We then
ask what China’s export revenues for 2020
would have been if, in each product group,
the country had participated in global trade
China: pandemic-related windfalls in
growth at 2019 product-specific global
selected export categories* market shares. It turns out that the hypo-
US$ billion thetical export value calculated in this fash-
ion for 2020 would have been merely 1.4%
16
14
12 Protective garments
for medical use
10 (HS 621010)
1 The considerable expansion in public investment ac-
8 Medical test Vaccines tivity was an additional key reason.
kits (HS 300220)
6
(HS 300215) 2 China’s goods imports in 2020, on the other hand,
4 did not quite sustain their 2019 levels in value. The
Face masks
2 (HS 630790) sharp fall in commodity prices was the decisive factor,
0 however. If these products are factored out, the result
is an increase of just under 3%. German exporters,
M A M J J A S O N D J F M A M J J A
too, benefited from China’s essentially quite strong im-
2020 2021
port demand (see Deutsche Bundesbank (2020)).
Source: Bundesbank calculations based on Trade Data Monit- 3 The disaggregation is based on the classification of
or. * Export value less the average export value for the same goods in the Harmonized Commodity Description and
month in 2018 and 2019.
Coding System (HS) at the six-digit level. Data were
Deutsche Bundesbank
taken from the Trade Data Monitor database.
 Deutsche Bundesbank
Monthly Report
October 2021
47

below 2019 export revenues, whereas The aggregate economic upturn in China
global trade contracted by 6%. decelerated markedly in the first three quar-
ters of this year. Exports, in turn, were an
The product range effect therefore had a important factor in this development, too.
stabilising effect on Chinese exports in a dif- They continued to expand briskly in the first
ficult global economic environment. How- quarter. However, as the pandemic receded
ever, the fact that Chinese exports actually and goods consumption in the industrial
rose requires further explanation. In individ- countries began to return to normal, the
ual product groups, China gained consider- export boom seems to have been dissipat-
able global export shares. This increase was ing since then.5 The Chinese economy is
particularly impressive for personal protect- thus once again increasingly dependent on
ive equipment products, which include, for drivers of domestic growth.
instance, face masks, global demand for
which veritably skyrocketed within just a
few weeks after the outbreak of the pan-
demic. China succeeded in extremely ramp-
ing up its production capacity within a short
period of time and was thus able to almost 4 These included not only face masks (HS code:
single-handedly accommodate the increase 630790) but also protective garments for medical use
(621010), medical test kits (300215), disinfectants
in global demand. On the whole, pandemic- (380894) and diagnostic or laboratory reagents
related medical products contributed just (382200).
5 Although revenue from goods exports was still up
over 2½ percentage points to Chinese ex- nearly 25% year-on-year in the third quarter of 2021,
port growth in the past year.4 this was probably due in large part to price increases.

times when waves of the pandemic occurred.

Whilst infection rates were high, the time users
spent at home generally increased consider- 7 For each country, this can be calculated from the mean
ably; conversely, far fewer individuals were at values of the following sub-​indices of the national Google
mobility reports: retail and recreation, grocery and phar-
the workplace, and consumers stayed away macy, transit stations, and workplaces. Each sub-​indicator
from restaurants and recreational facilities. taken by itself measures the percentage change in visitor
numbers as compared to the reference date in January-​
February 2020. To reduce fluctuations over the course of
Empirical esti- A Bundesbank empirical study based on work the week, the mobility index is included in the estimates as
mates for a a seven-​day moving average. See Google LLC (2021).
large group
by the IMF on developments during the first 8 See International Monetary Fund (2020b) and, for a
of countries few months of the pandemic analyses these re- more detailed discussion of the approach and results,
Caselli­et al. (2021).
lationships using a broad-​based measure of 9 Measured using the stringency index of the Oxford
mobility7 for a large group of countries span- COVID-​19 Government Response Tracker, adjusted for the
influence of public information events.
ning 128 economies.8 One topic of particular 10 Infection rates are captured using the number of new
interest is the mobility response to a tightening infections per 100,000 inhabitants within seven days, as
taken from the data provided by Johns Hopkins University.
of government-​imposed restrictions9 as well as The choice of indicator is based on the assumption that the
population adapts its behaviour to current infection rates
to rising infection numbers, based on which
rather than on the basis of indicators which only provide a
voluntary changes in behaviour are assumed to lagged picture of developments in the pandemic. Persistent
differences in national testing strategies are taken into ac-
have taken place.10 Owing to the high fre-
count in the estimates through country fixed effects. Speci-
quency of data, assumptions about the incuba- fications that use death counts as pandemic indicators in-
stead produce results of similar quality.
tion period and the length of political decision-​
11 Specifically, it is assumed that unexpected changes to
making processes allow pandemic shocks to be government containment measures or unforeseen devel-
opments in infection rates directly affect mobility. Mobility
identified.11 shocks, on the other hand, are only assumed to have a
lagged impact on the other variables.
Deutsche Bundesbank
Monthly Report
October 2021
48

The impulse-​ response functions estimated Higher infection

High-frequency mobility and rates and con-
economic indicators using local projections for the period from tainment meas-
Median values across all available countries
January 2020 through June 2021 suggest that ures strongly
curbed mobility
government restrictions did, in fact, strongly
Mobility 1
+ 10 Deviation from pre-crisis figure (%) curb mobility.12 They show that the immediate
7-day average
0 introduction of the strictest containment meas-
– 10 ures, which involved nationwide stay-​at-​home
– 20 requirements and extensive business closures,
– 30 inter alia, reduced mobility by almost 35%.13,14
– 40 After just over one week, the estimated effects
– 50 gradually start to wear off, probably mainly due
– 60 to the success of the measures in slowing
Macroeconomic activity 2
Deviations from the projection baseline down the rate of infection and a subsequent
of November 2019, weekly,
enlarged scale
+ 5 easing of the restrictions. The results also show
0 that, in and of themselves, increasing infection
– 5
rates caused a clear decline in mobility. In any
– 10
case, after a doubling of the seven-​day total of
– 15
new infections per 100,000 inhabitants, mobil-
– 20
ity decreased by just under 5% on average, and
2020 2021 had barely recovered even almost two months
Sources: Google COVID-19 Community Mobility Report, OECD later. Given that government restrictions ini-
and Bundesbank calculations. 1 Mean of the following sub-
indices of the national Google mobility reports: retail and re- tially remained unchanged, this response was
creation, grocery and pharmacy, transit stations, and work-
places. 2 According to the OECD Economic Activity Tracker. probably largely due to the public taking their
Deutsche Bundesbank own safety precautions.15 Because the number
Global mobility responses to of infections not only doubled, but multiplied
pandemic shocks* several times over in each of the last pandemic
% waves, voluntary changes in behaviour prob-
Tightening of government containment measures ably played a key role in the observed declines
0 100-point increase in stringency index)

– 10 12 The impulse-​response functions reflect the estimated

90% confidence interval1 coefficients of regressions that explain future develop-
– 20 ments in mobility using the pandemic and containment
variables. The projection equations also take into account
– 30 realisations of all variables (including the dependent vari-
ables) during the previous two weeks, as well as time and
– 40 country fixed effects. This approach is therefore broadly
Increase of infection rates
equivalent to an estimate using a panel vector autoregres-
(doubling in seven-day total of new sion (VAR) model. See also Jordà (2005) as well as Plagborg-​
infections per 100,000 inhabitants) +2 Møller and Wolf (2021).
13 Because the mobility indices capture the percentage de-
0 viation of mobility from a reference level in January-​
February 2020, the impulse-​response functions reflect,
–2 strictly speaking, the responses attributable to the meas-
ures in percentage points. This is approximately equal to
–4 the percentage change in the mobility level. All responses
described here are significantly different from zero at the
–6 10% level at least.
14 In countries such as Italy, where similarly strict rules
0 10 20 30 40 50 were briefly in force in the second quarter of 2020, this
Days after shock occured corresponds to around half of the decline in mobility actu-
Source: Bundesbank calculations. * Impulse-response function
ally observed.
derived from local projections. Estimation equations regress 15 The effects, which initially increase over time, probably
mobility indices on contemporaneous and lagged indicators of also reflect the lagged tightening of containment measures
the stringency of government containment measures and pan- to at least some extent, however. A correlation of this kind
demic developments as well as time and country fixed effects.
1 Based on clustered standard errors. is suggested, inter alia, by separately estimated impulse-​
Deutsche Bundesbank
response functions for the adjustment of containment
measures after a doubling of infection numbers.
 Deutsche Bundesbank
Monthly Report
October 2021
49

Global mobility responses to the tightening of specific containment measures*

Percentage changes in the measure of mobility following a 100-point increase in the relevant stringency index

Ban on public events Restrictions on gatherings Restrictions on School and university

+4 internal movement closures

–4 90% confidence interval1

–8

+ 4 Stay-at-home Restrictions on International travel Workplace closures

requirements public transport restrictions

– 4

– 8

– 12

0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50
Days Days Days Days

Source: Bundesbank calculations. * Impulse-response function derived from local projections. Estimation equations regress mobility in-
dices on contemporary and lagged indicators for the stringency of specific and more general government containment measures and
pandemic developments as well as time and country fixed effects. 1 Based on clustered standard errors.
Deutsche Bundesbank

in mobility and activity alongside government-​ erings, international travel and public events
mandated containment measures. This obser- was much smaller.19 All in all, the results there-
vation is consistent with the IMF’s findings for fore suggest that those measures that probably
the first few months of the pandemic as well as
a large number of academic studies on this
topic.16 16 Chernozhukov et al. (2021), for instance, confirm that
stay-​at-​home orders and business closures were highly ef-
fective. However, other studies, including Gupta et al.
Strict bans, stay-​ A closer look at the individual containment (2020) and Goolsbee and Syverson (2021), note that con-
at-​home require- siderable changes in mobility behaviour were already visible
ments and
measures reveals that they had very different ef- before the tightening or easing of containment measures
workplace clos- fects on mobility and thus also on economic ac- and highlight the role of self-​imposed behavioural adjust-
ures particularly ments. Even the strong deterioration in the US labour mar-
limiting
tivity and the course of the pandemic.17 For ex- ket in the first few months of the pandemic can probably
ample, although behavioural recommendations only partly be explained by government containment
measures; see, for instance, Baek et al. (2021) as well as
alone – such as those regarding on-​site working Kong and Prinz (2020).
or social distancing – also noticeably reduced 17 In the following analyses, in addition to the respective
measure in question for a given type of containment meas-
mobility,18 mandatory measures such as ure, the local projections incorporate a second indicator
government-​ imposed workplace closures or that summarises the stringency of the restrictions in all
other categories.
stay-​at-​home requirements had a much greater 18 This distinction takes advantage of the fact that, at the
level of the components of the Oxford COVID-​19 Govern-
impact. This is also true in comparison to regu-
ment Response Tracker, behavioural recommendations are
lations that encroached on other areas of pub- differentiated from mandatory measures of varying de-
grees of magnitude.
lic and private life. While strict workplace clos-
19 However, it also appears that measures – such as re-
ures and stay-​at-​home requirements by them- strictions on international travel – that have become part of
everyday life in many places since the outbreak of the pan-
selves reduced mobility by almost 10% and 8%, demic were accompanied by mobility restrictions that per-
respectively, the impact of restrictions on gath- sisted for comparatively long periods.
 Deutsche Bundesbank
Monthly Report
October 2021
50

a doubling of the infection rate are also statis-

Global economic activity after
pandemic shocks* tically significant. According to the estimation,
%
they reduce GDP by up to 1%.

Tightening of government containment measures

+3 (100-point increase in stringency index)

0 Differences in how the euro

–3 area countries’ economies
–6
90% confidence interval1
were affected
–9
The analysis up to now has largely excluded the Economic effects
Increase in infection rates + 0.3 of the corona-
(doubling of seven-day total of new heterogeneity between countries. However, virus crisis rela-
infections per 100,000 inhabitants) 0 the pandemic has weighed on the individual tively heteroge-
neous across
– 0.3 countries to differing degrees and at different countries
– 0.6 times. A comparison of China with the rest of
– 0.9 the world illustrates this in particular. In that
country, many production plants came to a
– 1.2
standstill as early as in January 2020. A few
– 1.5
months later, economic activity then slumped
0 1 2 3 4 5 6 7
elsewhere. A part was played by the disruption
Weeks after shock occured
to international value chains as a result of sup-
Source: Bundesbank calculations. * Impulse-response function
derived from local projections. Estimation equations regress ply shortfalls for Chinese products (see the box
activity indices on contemporaneous and lagged indicators of
the stringency of government containment measures and pan- on pp. 52 ff.).
demic developments as well as time and country fixed effects.
1 Based on clustered standard errors.
Deutsche Bundesbank The euro area economies also pulled through Drops in activity
varied widely in
the crisis very differently. This already became the euro area
entailed particularly severe economic costs had apparent in the first half of 2020. At the peak
the strongest impact on mobility behaviour. of the crisis in the second quarter, GDP losses
(in each case compared with pre-​crisis levels)
Economic activ- Because measures of economic activity are ranged from 4½% in Lithuania to 22% in
ity also strongly
impaired
generally only available on a quarterly basis, Spain.22 Even one year later, following a rapid
analysing GDP losses using a high-​frequency but partly bumpy recovery, the situation has re-
dataset is not entirely straightforward. How- mained mixed. While GDP in some Member
ever, the OECD evaluates internet search quer- States had exceeded its respective pre-​crisis
ies on key economic topics every week and level once again as early as in the second quar-
tracks overall economic activity on this basis.20
The resulting indicators, which are available for
46 countries, provide at least a rough high-​ 20 See OECD (2020).
21 This is a fairly conservative estimate. Identifying the ef-
frequency picture of GDP trends since the start fects of government containment measures in this study is
of 2020. Analysing local projections based made more difficult by the fact that many of the mostly ad-
vanced economies in the sample tightened or eased restric-
on these data confirms, for instance, that tions in the same week. Much of the variation is therefore
both self-​imposed and government-​imposed explained by time fixed effects. If these are stripped out of
the estimations, the calculated decline in economic activity
changes in behaviour during the pandemic after tightening government containment measures to the
greatest extent possible is almost three times as large.
were associated with considerable declines in
22 Ireland recorded only a slight decline in real GDP in the
activity. Taken in isolation, tightening govern- second quarter of 2020 and economic activity already ex-
ceeded its pre-​crisis level in the following quarter. However,
ment containment measures to the greatest
the statistical reporting of economic output in Ireland has
extent possible would directly reduce economic been largely determined by the strategic planning of multi-
national enterprises for several years (see Deutsche Bun-
activity by around 6% below the expected desbank (2018)). For this reason, Ireland is excluded from
baseline before the pandemic.21 The effects of the following analyses.
 Deutsche Bundesbank
Monthly Report
October 2021
51

ter of 2021, in Spain it remained just over 8%

Correlation of GDP losses in the
lower. The backlogs were somewhat lower in euro area with selected indicators*
Italy, at 4%, as well as in Germany and France,
each at 3%.

Pandemic devel- The mixed picture across countries is partly at-

opments, con-
tainment meas-
tributable to differences in infection rates. Sim- Mean GDP losses1
up to …
ures and mobil- ple correlation analyses indicate this, at least.
ity behaviour … Q2 … Q2
correlate
Particularly in the first half of 2020, economic Item 2020 2021
strongly with losses clearly coincided with the intensity of the Direct and indirect effects of infection
direct economic rates
losses
pandemic.23 The stringency of government
COVID-19 death rates2 0.53 0.27
containment measures24 and the mobility indi- Oxford index3 0.65 0.77
Mobility behaviour4 – 0.72 – 0.38
cator25 turn out to be even more closely associ-
ated with GDP losses in the first half of 2020. Economic structure
Share, tourism5 0.70 0.74
Overall, the findings support the hypothesis Share, hotel and restaurant sector6 0.51 0.66
that, above all, the countries which had to
Fiscal support measures
shoulder sharp declines in GDP were those that Change in government fiscal balance7 – 0.44 – 0.50
were hit particularly hard by the pandemic, in
* Euro area excluding Ireland. Table looks at difference between
which stringent and/​or more protracted con- the mean values in Q1 and Q2 2020, and from Q1 2020 to Q2
tainment measures were in force, and whose 2021. 1 Calculated as the mean difference in real GDP to the
level of Q4 2019. 2 Number of deaths of or with COVID-19 over
residents restricted their mobility more sharply. seven days per 100,000 inhabitants. 3 Oxford COVID-19 Gov-
ernment Response Tracker (excluding the influence of public in-
formation campaigns); a higher index level indicates stricter re-
Relationships However, the relationship between pandemic strictions (no data are available for Malta). 4 Mean of the sub-
become weaker indices of the national Google mobility reports: retail and recre-
over time
developments and mobility, on the one hand,
ation, grocery and pharmacy, transit stations, and workplaces.
and economic developments, on the other, Each sub-index, taken in isolation, measures the percentage
change in the number of visitors compared with the reference
later weakened. The correlation of the indica-
day in January/February 2020 (no data are available for Cyprus).
tors with GDP losses cumulated since the onset 5 OECD indicator: share in gross value added of the sectors dir-
ectly related to tourism (2018 or the earlier, most recently avail-
of the pandemic fell. This is likely to be due, able year; no data are available for Belgium and Cyprus). 6 Share
inter alia, to enterprises adapting better to the of gross value added (2019). 7 Difference between the current
general government fiscal balance for 2020 and the value fore-
pandemic conditions through the deployment cast in the macroeconomic projections published by the Euro-
of hygiene measures and increased remote system in December 2019; as a percentage of GDP from 2019.

working, and households’ increased use of Deutsche Bundesbank

contactless distribution channels on account of

the restrictions.

Countries with As contacts were reduced during the pan-

significant hotel
and restaurant
demic, economic sectors such as the food and
industry or tour- beverage and accommodation sectors, but also 23 The intensity of the pandemic is approximated using of-
ism sector so far ficial death figures. For a comparison of how different
worst affected
transport services and cultural activities suf-
countries were affected by the pandemic, it is preferable to
by the crisis fered heavy losses. Economies for which these use death rates as an indicator rather than infection rates,
sectors play a key role were thus hit particularly as the latter largely depend on the respective testing strat-
egy.
hard. The correlation analysis suggests that the 24 As before, the stringency of government containment
measures is approximated using the stringency index of the
significance of the economic structure for GDP
Oxford COVID-​19 Government Response Tracker (excluding
growth even increased over time. For instance, the influence of public information campaigns).
25 The broad-​based measure of mobility is again used as a
it reveals a close relationship between the share
mean of the sub-​indices of the national Google mobility re-
of the accommodation sector in aggregate ports: retail and recreation, grocery and pharmacy, transit
stations, and workplaces.
gross value added26 and cumulative GDP losses 26 Shares in aggregate gross value added from 2019 were
up to the second quarter of 2021. A broadly taken into account.
Deutsche Bundesbank
Monthly Report
October 2021
52

The role of the disruption of Chinese supply chains

in production slumps in the United States and the EU
in spring 2020
The government of the People’s Republic of One way of approximating these effects is
China responded to the outbreak of the to compare developments in sectors which
coronavirus pandemic as from the end are dependent to varying degrees on inputs
January 2020 by ordering businesses to supplied by China. To this end, we use
shut down and imposing extensive restric- input-output tables to calculate a measure
tions on labour mobility. Owing to the of dependence at a detailed level of break-
major importance of Chinese firms for inter- down of industries in the United States and
national goods trade, this disrupted numer- the European Union.1 All sectors are broken
ous supply chains. Many industrial firms the down into more highly exposed and less
world over cited this as a key factor limiting highly exposed industrial sectors based on
production in March and April 2020. How-
ever, in many places the burdens caused
directly by the pandemic increased more or
less simultaneously, too. This makes it all
the more difficult to identify the role played 1 For the United States, we use the input-output tables
of the Bureau of Economic Analysis for 2012 and cus-
by those value chain disruptions that are at-
toms values provided by the US Census Bureau for
tributable to China in the global production 2019. Only intermediate inputs and capital goods are
included in the calculation. For the EU, we use Euro-
slump in spring 2020. stat’s supply and use tables for 2017 and customs val-
ues for 2019.

Impact of large dependence on Chinese inputs in spring 2020*

%, baseline period: February 2020

USA EU
+ 10 Production + 10

0 0

– 10 – 10

– 20 90% confidence interval – 20

– 30 – 30

– 40 – 40

+3 Producer prices +3

+2 +2

+1 +1

0 0

–1 –1

–2 –2
D J F M A M J J A D J F M A M J J A
2019 2020 2019 2020

Sources: Federal Reserve Board, Bureau of Economic Analysis, Census Bureau, Haver Analytics, Eurostat and Bundesbank calculations.
* The sample contains US and EU manufacturing sectors. The data refer to differences between sectors which are particularly depend-
ent on Chinese inputs and less-dependent sectors.
Deutsche Bundesbank
 Deutsche Bundesbank
Monthly Report
October 2021
53

the cost shares of inputs from China.2 In a The role of China-specific import frictions
panel data analysis with monthly data on in the slump in industrial production in
sectors’ production, employment and pro- spring 2020*
ducer prices, we can then gauge the impact Compared to December 2019

of large dependence on Chinese inputs.3 Components in percentage points

China-specific import friction
Domestic economic shocks 1
The estimations show that, in industries Other 2

that are highly dependent on Chinese input + 5 USA

supplies, production dropped off signifi- 0

cantly more sharply in March and April – 5

2020 than in other industries. The differ- – 10

ence in April was nearly 10% for the United – 15

States and even approached 27% for the – 20

Production (%)
European Union. This discrepancy did not – 25
Euro area
persist, however. As from July 2020, there + 5

have been no significant differences be- 0

tween industrial sectors that are more de- – 5
pendent on or less dependent on Chinese – 10
inputs. This is likely to be due to the rapid – 15
lifting of restrictions in China, as a conse- – 20
quence of which its foreign trade had al- – 25
ready recovered fully in April. – 30

– 35
J F M A M J J A
2020

Sources: Federal Reserve Board, Bureau of Economic Analysis,

2 The median share of Chinese inputs in production Census Bureau, Haver Analytics, Eurostat and Bundesbank cal-
costs is roughly 1% in both the United States and the culations. * Contributions of contemporaneous and past real-
EU. In those industries with above-median exposure to isations of shocks derived from a recursively identified structur-
al VAR model. 1 Direct shocks to domestic consumption or to
Chinese inputs, the average share is well above 2%. domestic industrial production. 2 Disruptions to trade with the
3 For the United States, the analysis incorporates the rest of the world and deterministic component.
four-digit NAICS manufacturing industries over the Deutsche Bundesbank
January 2019 to March 2021 period from the G.17 Re-
lease of the Board of Governors of the Federal Reserve
System. For the EU, data from Eurostat’s (primarily
three-digit) NACE manufacturing sectors are available In addition, the estimates show that produ-
for the same period. The regression controls for time cer prices in industries that are particularly
fixed effects and industry-specific fixed effects. It also
controls for general dependence on imported inputs dependent on Chinese inputs picked up
and the degree of trade openness over time. Some slightly at the outbreak of the pandemic,
sectors which could be particularly affected by domes-
tic restrictions were omitted from the analysis, such as whereas prices in less exposed industries
transport goods and clothing production. See Khalil fell. In April and May 2020, prices in par-
and Weber (2021). Meier and Pinto (2020) present a
similar analysis for the role of international value chains ticularly dependent sectors were 1% to 2%
at the beginning of the pandemic in the United States. higher for the United States. In the EU,
In addition, Santacreu et al. (2021) also shed light on
the role of large exposure to imported inputs. where developments were very similar, the
4 The fuel and coal processing industries were omitted corresponding price differential was around
from the study in order to factor out energy price de-
velopments. 1%.4 Thus, shortfalls in intermediate goods
5 It can be shown for the United States, for which de- imports from China probably resulted in
tailed employment data are available, that increased
dependence on China also considerably dampened supply-side disruptions in large parts of the
employment in spring 2020. This indicates that there US and EU manufacturing sector.5
are complementarities between inputs and labour in
the manufacturing sector, at least in the short term.
Deutsche Bundesbank
Monthly Report
October 2021
54

Alongside the sector-level impact of Chi- vere and lasting damage to the global econ-
nese supply shortfalls, the macroeconomic omy.
effects are also of interest. An analysis using
a structural vector autoregressive (SVAR)
model is a promising way of capturing
these effects. Private goods consumption,
manufacturing production, goods imports
from the rest of the world (excluding China)
and goods imports from China all feed into
the model, which is estimated separately
for the United States and the euro area.6

A historical shock decomposition based on

the estimation results initially shows that
the decline in US and euro area goods im-
ports from China in February and March
2020 can be attributed to China-specific
trade frictions.7 This turmoil also contrib-
uted to the considerable production short-
falls in the US and euro area manufacturing
industry in the spring months of 2020.
These shortfalls subsided distinctly in both
regions within just a few months. Accord-
ing to the shock decomposition, however,
domestic economic developments in each
of those regions constituted the more im-
portant factor accounting for the drop-off
in industrial production.

To sum up, therefore, the analyses indicate

that the disruptions to cross-border value
chains resulting from the Chinese contain-
ment measures markedly weakened indus-
6 All data for the United States are seasonally and
trial activity in the United States and the price adjusted. For the euro area, private goods con-
euro area at the outbreak of the pandemic. sumption is approximated by retail sales in value terms
(excluding private motor vehicles). All data for the euro
The disruptions in trade with China, how- area are seasonally adjusted. By ranking the variables
ever, were not the main reason for the and through recursive identification, it is assumed in
the model that goods imports from China contempor-
slump in production at that time. Moreover, aneously respond to unexpected disruptions in private
the strains were short-lived as China was consumption, industrial production and trade with
third countries. China-specific trade disruptions, in
able to ramp up the manufacture of inputs turn, impact directly only on goods imports from
relatively quickly. Based on this experience, China; other variables respond with a time lag. The ap-
proach is based on Kilian et al. (2021) and is described
it also stands to reason that the latest pro- in more detail by Khalil and Weber (2021). The Euro-
duction disruptions caused by local flare- pean Central Bank’s BEAR toolbox was used for the
estimations (see Dieppe et al. (2016)).
ups of the coronavirus in some emerging 7 For both regions, the slump in imports from China
market economies will not result in any se- can be explained almost entirely by China-specific
trade disruptions, whereas other shocks do not play
any significant role.
 Deutsche Bundesbank
Monthly Report
October 2021
55

defined tourism indicator27 also correlates The euro area’s relatively poor economic per- Pandemic and
economic policy
somewhat more strongly with the cumulative formance was probably partly down to differ- as possible
losses in activity than with the depth of the im- ences in the course of the pandemic and in the explanatory
factors­
mediate slump in the first half of 2020. This is responses taken to it. Already in the first wave
probably a reflection of the fact that, in later of infection, the self-​imposed and government-​
waves of the pandemic, administrative contain- mandated behavioural adjustments were more
ment measures were focused almost exclu- stringent in the euro area than in the United
sively on high-​contact services sectors, whilst, States. This also applied to the responses to the
in the first wave of the pandemic, even all eco- resurgence of the pandemic at the turn of
nomic sectors seen as not directly essential 2020-21. Economic policy might also have
were temporary shut down. been a key factor. Although monetary policy
was eased swiftly and decisively on both sides
Fiscal measures To cushion the economic fallout from the pan- of the Atlantic, at the beginning of the crisis,
particularly size-
able in the
demic, the euro area countries took extensive there was greater scope for doing so in the
worst affected fiscal measures. The deviation of the general United States. The sequence of extensive stimu-
countries
government fiscal balance for 2020 from the lus packages in the United States also suggests
value forecast in the last pre-​crisis Eurosystem that US fiscal policy might have supported the
projection can be used as an indicator for the economy to a greater extent.30
overall fiscal stimulus.28 It reveals a close rela-
tionship with economic development; that is, These questions are addressed in our own em- According to
SVAR analysis,
in countries with more pronounced GDP losses, pirical analysis below. Structural vector autore- more stringent
the deficit widened even more. This is likely to gression (SVAR) models provide a framework restrictions on
behaviour in
reflect the fact that, in particularly hard hit for analysing the relative significance of the euro area
countries, the automatic stabilisers responded various explanatory factors.31 In the model significant­
robustly and governments took extensive fiscal used, the relationship between economic activ-
measures.

Causes of the transatlantic

growth differential
Gap in growth There were also considerable differences in
between United 27 The indicator captures the share in gross value added of
States and euro
economic developments between the euro
the sectors directly related to tourism and reflects both do-
area since out- area as a whole and the United States during mestic and overseas tourism. See OECD (2021a).
break of pan- 28 Compared with the value from the macroeconomic
demic
the coronavirus crisis.29 Real GDP in the euro
projections published by the Eurosystem in December
area contracted by 6.5% in 2020. In the United 2019; the fiscal balance was referenced in each case to
States, the decline was roughly half this nominal GDP in 2019. See European Central Bank (2019).
29 A number of institutions, including the European Cen-
amount. In addition, from the final quarter of tral Bank (2021), the International Monetary Fund (2021)
2020 and the first quarter of 2021 onwards, re- and Banco de España (2021), as well as economists at the
Banque de France (Chatelais (2021)), looked into the
covery in the euro area was rather bumpy. In growth differential between the United States and the euro
the second quarter of 2021, economic output area or Europe. These partly descriptive, partly model-​
based analyses saw more stringent administrative and self-​
in the euro area was still 3% lower than its pre-​ imposed restrictions in the euro area as the key factor be-
hind the transatlantic growth differential. In addition, dif-
crisis level, whilst in the United States it already
ferences in fiscal support, the economic structure, the de-
slightly exceeded its pre-​crisis level. These dif- gree of openness, and the underlying pace of growth also
played a certain role.
ferences also remain when taking into account
30 For an evaluation of the latest major stimulus pro-
the stronger economic growth in the United gramme, see Deutsche Bundesbank (2021a).
31 The models described below were estimated using
States in the period prior to the crisis. Bayesian methods employing the European Central Bank’s
BEAR toolbox. See Dieppe et al. (2016).
 Deutsche Bundesbank
Monthly Report
October 2021
56

icy is estimated based on counterfactual simu-

Historical decomposition of the growth
rate of real GDP * lations which assume a scenario where all
Quarter-on-quarter change
forms of support provided by economic policy
during the past one-​and-​a-​half years are ex-
+ 8 United States
cluded and monetary policy interest rates, gov-
+ 4
ernment expenditure and tax rates are set as
0 expected before the crisis.36 According to the
– 4 Decomposition in percentage points simulations, considerably stronger fiscal policy
– 8
Lockdown1 responses by the United States go a long way
Monetary policy shock
– 12 Fiscal shock
towards explaining its more favourable eco-
Other shocks nomic development in comparison with the
Euro area
Real GDP (%)
+ 12
euro area.37 The relative explanatory contribu-

+ 8

+ 4

0 32 The fiscal policy stance is approximated based on the

cyclically adjusted primary balance as a percentage of po-
– 4 tential output, incorporating measures on the expenditure
and revenue sides (such as tax cuts). Unexpected fiscal pol-
– 8 icy measures are identified with the help of sign restric-
tions. It is assumed that these reduce the cyclically adjusted
– 12
Q1 Q2 Q3 Q4 Q1 Q2
primary balance and, at the same time, stimulate GDP
2020 2021 growth. The monetary policy stance is approximated based
on the shadow rate (according to estimates by Krippner
* Contributions of contemporaneous and past realisations of (2013)). Monetary policy shocks lower the shadow rate
shocks derived from a structural VAR model (with an exogen- and, at the same time, boost real GDP growth and con-
ous variable) with zero and sign restrictions. 1 Contains contri-
bution of the constant. sumer price inflation. It is also assumed that they do not
Deutsche Bundesbank
have an immediate impact on the cyclically adjusted fiscal
primary balance.
33 On the basis of the Goldman Sachs Effective Lockdown
Index, both government-​mandated measures (according to
ity, monetary and fiscal policy indicators32 and the Oxford COVID-​19 Government Response Tracker) and
restrictions on behaviour33 during the pan- self-​imposed behavioural adjustments (according to Google
mobility reports) are taken into account. See Hatzius et al.
demic are estimated separately for each eco- (2020).
nomic area. According to a historical shock de- 34 The model takes into account not only the major role
played by automatic stabilisers in the euro area owing to
composition based on these estimations, the the social security systems (according to Dolls et al. (2012),
weaker economic development in the euro automatic stabilisers in the euro area cushion around 49%
of the idiosyncratic unemployment shock; in the United
area overall since the outbreak of the corona- States the figure is only 34%), but also the fact that the
virus crisis was primarily attributable to more United States frequently pursues an active stabilisation pol-
icy in times of crisis.
stringent containment measures and behav- 35 NiGEM is the global economic model developed by the
ioural adjustments. It indicates that the extent UK-​based National Institute of Economic and Social Re-
search (NIESR). It models economic interconnectedness be-
of extraordinary fiscal and monetary policy tween 60 economies and regions via foreign trade and the
measures did not play a major role. However, interest rate-​exchange rate nexus. The model has New
Keynesian features, especially forward-​looking elements on
the model only shows those responses that go the financial and labour markets. For further information,
beyond the usual responses relative to the in- see [Link]
36 Here, six fiscal policy variables (government consump-
tensity of the crisis as being fiscal or monetary tion, public investment, transfers, income tax, corporation
policy shocks.34 tax and VAT rates) and the short-​term interest rates from
the first quarter of 2020 onwards were replaced by values
set by the NIESR in the NiGEM forecast baseline from Janu-
ary 2020.
According to In order to reflect the effects of monetary and
37 For example, fiscal policy in the United States contrib-
NiGEM simula-
tions, greater US
fiscal policy responses in a more comprehen- uted around 3½ percentage points to the average quar-
terly growth rate since the beginning of the coronavirus cri-
fiscal expansion sive way, i.e. including automatic stabilisers and
sis, whilst in the euro area this effect was estimated to be
also significant
conventional policy responses, the NiGEM35 1¾ percentage points. In the simulations for the euro area,
the funds from the Next Generation EU programme were
global macroeconometric model is used. In this not yet taken into account as they have not yet been de-
context, the impact of monetary and fiscal pol- ployed.
 Deutsche Bundesbank
Monthly Report
October 2021
57

tion of monetary policy for the growth differ-

Supporting effects of monetary and
ential was significantly smaller.38 fiscal policy since the beginning of the
pandemic according to NiGEM
simulations*
Summary and economic Mean deviations of real GDP from the baseline (%)

­policy conclusions 4.0

Monetary policy
3.5
Various reasons All in all, the findings point to a series of im-
behind differ-
ences in how
portant reasons why countries’ economies 3.0
Fiscal policy
individual econ- were affected to differing degrees by the 2.5
omies were
affected
coronavirus pandemic. The pandemic did not
2.0
rage everywhere to the same extent. There
were also differences in the containment meas- 1.5
ures taken by governments. Sectoral particular-
1.0
ities in individual economies likewise played a
role. For example, the Chinese economy bene- 0.5

fited from its range of exports, which suited

0
people’s needs particularly well during the pan- United States Euro area
Memo item:
difference 1
demic. Conversely, the major importance of
Source: Bundesbank calculations based on the NiGEM global
tourism not least in some euro area countries is macroeconometric model. * Estimates based on the simulated
development of the respective economy under the assumption
likely to have contributed to the comparatively that the monetary and fiscal policy instruments would have fol-
lowed the paths expected prior to the outbreak of the pan-
sharp contraction in economic output. Finally, demic. Effects are averaged over the period from Q1 2020 to
Q2 2021. 1 Difference between the effects of economic policy
both in the euro area and in the United States, in the United States and the euro area in percentage points.
the fiscal and monetary policy responses cush- Deutsche Bundesbank

ioned the immediate impact of the crisis con-

siderably; in the United States the supporting ticipated that in 2024 global economic output Longer-​term
impact of the
effects were probably even greater. would remain 2¼% below the level expected crisis probably
prior to the outbreak of the pandemic.39 As is relatively low
in industrial
Recovery pro- The economic recovery has now made good already the case for the short-​term effects, the countries …
cess not entirely
smooth
progress in many places. The increasing per- picture is mixed across countries. For the ad-
centage of the population that is fully vaccin- vanced economies, the longer-​term damage
ated has contributed significantly to this. In will tend to be fairly minor. By contrast, in many
addition, households and enterprises have developing and emerging market economies
learned to deal with the challenges of the pan- where vaccination campaigns have often been
demic and government containment measures progressing at only a sluggish pace, a compre-
are being used in a more targeted manner and hensive recovery will be lagged. This increases
more sparingly. The recovery has not been en- the risk that economic scars will remain.
tirely smooth, however. Drags on growth in-
clude shortages of key intermediate inputs and
38 However, such estimates of the relative significance of
delivery delays. Pandemic-​induced shifts in de- different factors for economic development during the
mand contributed significantly to this. The un- pandemic are subject to a particularly high degree of un-
certainty. First, the findings depend crucially on the challen-
expectedly rapid speed of the recovery itself ging measurement of economic policy drivers and
pandemic-​related restrictions. Second, it is not known to
was probably another reason.
what extent experience from previous economic cycles
which underlies all models can be applied to the excep-
tional situation of the past two years.
While the short-​term economic consequences
39 This would mean losses would be significantly smaller
of the pandemic are now better understood, its than following the global financial and economic crisis of
2008-09, where they amounted to 8¾% after four years,
longer-​term impact has only been able to be measured in terms of predictions by the World Economic
roughly estimated so far. The IMF recently an- Outlook in October 2007.
Deutsche Bundesbank
Monthly Report
October 2021
58

Remote working and its impact on labour productivity

The coronavirus pandemic has seen an in- developments can be seen in other indus-
crease in remote working, meaning that trial countries, too, in which remote work-
there has been a surge in the use of associ- ing also increased broadly in response to
ated digital technologies (such as video the pandemic.4
conferences and cloud services). This devel-
opment is frequently accompanied by the The main reason for the sudden rise in
hope that it will prove sustainable and is working from home was the necessity to
potentially a sign of a broader push to- practise social distancing owing to the pan-
wards digitalisation1 that could strengthen demic. Working from home arrangements
productivity growth over the coming years. are also likely to be used more frequently
after the crisis than before on account of
In the case of Germany, the results of a rep- the investments made, learning and net-
resentative survey of firms conducted by work effects, as well as the wealth of posi-
the Bundesbank in May 2021 indicate that
almost three-fifths of enterprises have made
greater use of working from home arrange-
ments since the onset of the crisis.2 This
1 Alongside remote working, the use of online distri-
ratio was significantly higher in some ser- bution channels (e-commerce) and digital payment
vices sectors, such as the financial and in- systems, for example, has also received a boost. For in-
stance, online retailers’ sales have risen strongly in
surance activities sector or in the informa- both the USA and the EU.
2 This is a regular survey of firms conducted by the
tion and communication sector, while nat-
Deutsche Bundesbank. The participating enterprises
urally far fewer enterprises made use of make up a representative selection of Germany’s cor-
porate landscape; see Deutsche Bundesbank (2021b).
such arrangements in more contact-intensive
3 These findings are consistent with those of other
sectors, including the accommodation and studies; see, for example, Alipour et al. (2020).
4 For instance, according to Eurostat, the share of em-
food service activities sector, the retail trade
ployees in the EU who worked at least partly from
and also the construction sector. Larger en- home increased by around one-half on the year to just
under 23% in 2020. The American Time Use Survey re-
terprises, in particular, also made more ex- ported that this share increased from 22% in 2019 to
tensive use of remote working.3 Very similar 42% in 2020 in the USA.

Increase in working from home*

Percentage of enterprises

Slight increase Significant increase

Financial and insurance activities
Information and communication
Other services
Mining and electricity
Business support service activities
Manufacturing
Transportation and storage
Wholesale trade
Construction
Retail trade
Manufacture of food products
Accommodation and food service activities

0 10 20 30 40 50 60 70 80 90

Source: Bundesbank Online Panel Firms (BOP-F). * Percentage of enterprises that provided the response “increased slightly” or “in-
creased significantly” to the question “How has the use of the following digital technologies in your enterprise changed since the onset
of the coronavirus pandemic?” in the section entitled “Working from home/teleworking.”
Deutsche Bundesbank
 Deutsche Bundesbank
Monthly Report
October 2021
59

tive experiences with using this working Expected impact of the increase in
model.5 working from home on firm productivity*
Percentage of enterprises

It is still difficult to predict what impact this Expectation of an increase

60 in productivity
development will have on labour productiv- Expectation of a decrease
in productivity
ity. However, a number of studies based on
experiments or on corporate or employee 40

surveys indicate that employees are at least

as productive when working from home as 20
they are when working in traditional offices.6
Amongst other factors, the possibility of
0
structuring the working day more flexibly Enterprises with ...
... a slight ... a significant ... a significant
and also the time saved from not having to increase in increase in increase in both
working from working working from
commute play a significant role in this con- home1 from home1 home and
using other
text. By contrast, other studies find evidence technologies 2

of productivity-reducing effects.7 This view is Source: Bundesbank Online Panel Firms (BOP-F). * Distribution
of responses to the question “How do you expect the in-
mainly supported by increased communica- creased use of digital technologies in your enterprise to affect
productivity in your enterprise in the long term?” in conjunc-
tion costs in some cases and potentially tion with changes in the use of the digital technology “working
from home” previously reported by the surveyed firms. 1 En-
shortened periods of focused work. In add- terprises that have increased “working from home” slightly or
significantly and, at the same time, reported no increase in the
ition to these direct effects, however, in- use of other digital technologies included in the survey (exclud-
ing video conferences). 2 Enterprises that have significantly in-
creased remote working could also lead to creased “working from home” as well as the use of at least
one other digital technology included in the survey.
improved job matching in the longer run, Deutsche Bundesbank

with positions being filled by qualified em-

ployees from other regions or even other Overall, there is therefore some evidence to
countries, thereby enabling efficiency gains.8 suggest that the surge in remote working
The use of remote working could likewise re- seen since the outbreak of the crisis could
duce expenditure on workplace infrastruc- have a moderate productivity-enhancing
ture, for instance for office rents, and thus effect. However, a considerable amount of
have a productivity-enhancing effect. In view further research is still required with regard
of the reduction in spatial frictions, the trend to this relationship. It also still remains to be
towards remote working could also result in seen whether the pandemic has triggered
productivity-boosting reallocation effects be- a more extensive and broader- based
tween enterprises and sectors. It is conceiv- productivity-boosting push towards digitali-
able that the more productive members of sation.
the workforce will increasingly switch to more
successful and attractive enterprises, which 5 Studies supporting this assessment can be found, for
example, in Ozimek (2020), Alipour et al. (2021), Bar-
might bolster the success of these enter- rero et al. (2021), Erdsiek (2021) and OECD (2021b).
prises and augment aggregate productivity.9 Network effects in this context arise from the use of
video conferencing applications, for example. If these
applications are used by many people, then their use-
The results of the Bundesbank’s survey of fulness increases for individual users.
6 See Angelici and Profeta (2020), Barrero et al. (2021),
firms support an optimistic assessment Bloom et al. (2015), Deole et al. (2021), Erdsiek (2021),
overall. The vast majority of the surveyed Etheridge et al. (2020) and Statistics Canada (2021).
7 See Gibbs et al. (2021) and Morikawa (2021).
enterprises that relocated activities to a 8 See Kakkad et al. (2021) and Wolter et al. (2021).
home working structure during the crisis 9 In the longer term, however, the greater concentra-
tion of enterprises potentially associated with this de-
expect this development to be conducive to velopment could also reduce incentives to innovate
their enterprise’s productivity. and impair productivity growth.
 Deutsche Bundesbank
Monthly Report
October 2021
60

… also owing In the industrial countries, longer-​term damage for the international community to push ahead
to economic
policy support
is likely to also be limited by the rapid fiscal and with vaccination campaigns around the world.
measures monetary policy response. It boosted macro-
economic demand and employment,40 averted Economic policy in the industrial countries Fiscal policy
should not sup-
numerous corporate insolvencies41 and pre- should support macroeconomic recovery until port economic
vented major turmoil in the banking and finan- the end of the pandemic and thus try to avoid recovery for
longer than
cial systems. This created an environment in knock-​on damage. Thereafter, however, fiscal necessary
which investment activity was able to hold up consolidation needs to be tackled. Here, it is
relatively well. Overall, government measures not just a question of avoiding overstimulating
made a substantial contribution to containing and thus “overheating” the economy. The past
the negative impact on labour and capital input one-​and-​a-​half years have also shown how im-
and aggregate productivity.42 portant it is to have fiscal policy buffers in times
of crisis.
Productivity Certain developments could even provide the
boost through
accelerated
economy with additional momentum in future, Beyond this, thought should already be given Structural
change should
digitalisation?­ including, in particular, the push towards digi- today to the fact that the coronavirus crisis will not be hindered
talisation triggered by the pandemic. The pan- probably result in longer-​term changes to the once pandemic
has been over-
demic conditions forced many enterprises to economic structure. Certain business models come
digitalise their processes or business models. might no longer be sustainable in the long term.
This could fuel productivity growth over the However, many of the measures taken during
next few years. This is also suggested by the the crisis were rightly targeted at keeping firms
expectations of enterprises in Germany with re- from going under in the light of the high degree
gard to the increased use of remote working of uncertainty. In this way, the number of mar-
(see the box on pp. 58 f.). ket exits has fallen distinctly since the onset of
the crisis, not least on account of the insolvency
Global vaccin- A turning point in the pandemic was the devel- moratoria adopted in many countries. As the
ation campaign
needs to be
opment of effective vaccines. A large part of pandemic is gradually overcome, these forms of
driven forward the population is now vaccinated in the indus- assistance should be scaled back so that the ne-
trial countries; however, in most cases vaccin- cessary structural change is not hindered.
ation rates are not high enough to enable all
protective measures to be lifted. In many de-
40 For example, during the crisis relatively few jobs were
veloping and emerging market economies, vac- lost in the advanced economies. The unemployment rate,
cines are still in scarce supply. In the world’s having been as low as 4.8% in 2019 for the group of ad-
vanced economies, rose to 6.6% in 2020 in the aftermath
poorest countries, just 1½% of the population of the dramatic economic slump at the beginning of the
has been fully vaccinated so far. This not only pandemic. The increase was considerably smaller outside
the United States, and there, too, the unemployment rate
means that millions of people have largely no then began to go back down rapidly. As a result, the
protection against the virus but it is also en- longer-​term effects on the potential labour force are likely
to be modest in the current crisis.
couraging more dangerous strains of the virus 41 For example, in Germany in 2020, the number of cor-
porate insolvencies fell by around 15% compared with the
to develop. In addition, new waves of infection
previous year and was thus at its lowest level since the
could trigger renewed economic setbacks in introduction of the current insolvency framework in 1999
(see Federal Statistical Office (2021)). According to Eurostat
the developing and emerging market econ-
data, the EU as a whole recorded a decline of around 23%.
omies. This would also affect the industrial For the United States, too, Crane et al. (2021) find evidence
of fairly low exit rates since the outbreak of the crisis.
countries via international trade and the global 42 For a model-​based analysis of the measures taken in
financial system. It thus remains a priority issue Germany, see Hinterlang et al. (2021).
 Deutsche Bundesbank
Monthly Report
October 2021
61

List of references
Alipour, J.-V., O. Falck and S. Schüller (2020), Germany’s Capacities to Work from Home, CESifo
Working Papers, No 8227.

Alipour, J.-V., C. Langer and L. O’Kane (2021), Is Working from Home Here to Stay? A Look at 35
Million Job Ads, ifo Schnelldienst, No 9/​2021, pp. 46-52.

Angelici, M. and P. Profeta (2020), Smart-​Working: Work Flexibility without Constraints, CESifo
Working Papers, No 8165.

Baek, C., P. McCrory, T. Messer and P. Mui (2021), Unemployment Effects of Stay-​at-​Home Orders:
Evidence from High Frequency Claims Data, The Review of Economics and Statistics, forthcoming.

Banco de España (2021), Some determinants of the growth differential between the euro area and
the United States since the onset of the pandemic, Economic Bulletin, Issue 2/​2021, pp. 25-28.

Barrero, J. M., N. Bloom and S. J. Davis (2021), Why Working From Home Will Stick, NBER Work-
ing Paper, No 28731.

Bloom, N., J. Liang, J. Roberts and Z. J. Ying (2015), Does Working from Home Work? Evidence
from a Chinese Experiment, The Quarterly Journal of Economics, Vol. 130(1), pp. 165-218.

Caselli, F., F. Grigoli and D. Sandri (2021), Protecting Lives and Livelihoods with Early and Tight
Lockdowns, The B. E. Journal of Macroeconomics, forthcoming.

Chatelais, N. (2021), Covid-​19 and divergence in GDP declines between Europe and the United
States, Eco Notepad, blog post No 229, [Link]
blog-entry/covid-19-and-divergence-gdp-declines-between-europe-and-united-states

Chernozhukov, V., H. Kasahara and P. Schrimpf (2021), Causal impact of masks, policies, behavior
on early covid-​19 pandemic in the U. S., Journal of Econometrics, Vol. 220(1), pp. 23-62.

Crane, L. D., R. A. Decker, A. Flaaen, A. Hamins-​Puertolas and C. Kurz (2021), Business Exit During
the COVID-​19 Pandemic: Non-​Traditional Measures in Historical Context, Federal Reserve Board,
Finance­and Economics Discussion Series, No 2020-089.

Deole, S. S., M. Deter and Y. Huang (2021), Home Sweet Home: Working from home and employee­
performance during the COVID-​19 pandemic in the UK, GLO Discussion Paper, No 791.

Deutsche Bundesbank (2021a), The macroeconomic implications of the American Rescue Plan,
Monthly Report, May 2021, pp. 16 ff.

Deutsche Bundesbank (2021b), Assessments and expectations of firms in the pandemic: findings
from the Bundesbank Online Panel Firms, Monthly Report, April 2021, pp. 33-56.

Deutsche Bundesbank (2020), China’s foreign trade withstanding the global economic crisis,
Monthly Report, August 2020, pp. 15 f.
Deutsche Bundesbank
Monthly Report
October 2021
62

Deutsche Bundesbank (2018), Activities of multinational enterprise groups and national economic
statistics, Monthly Report, October 2018, pp. 65-78.

Dieppe, A., R. Legrand and B. van Roye (2016), The BEAR toolbox, ECB Working Paper No 1934.

Dolls, M., C. Fuest and A. Peichl (2012), Automatic stabilizers and economic crisis: US vs. Europe,
Journal of Public Economics, Vol. 96, pp. 279-294.

Dong, E., H. Du and L. Gardner (2020), An interactive web-​based dashboard to track COVID-​19 in
real time, The Lancet Infectious Diseases, Vol. 20(5), pp. 533 f.

Erdsiek, D. (2021), Working From Home During COVID-​19 and Beyond: Survey Evidence From
Employers­, ZEW Discussion Paper, No 21-051.

Etheridge, B., Y. Wang and L. Tang (2020), Worker Productivity during Lockdown and Working
from Home: Evidence from Self-​Reports, ISER Working Paper, No 2020-12.

European Central Bank (2021), Economic developments in the euro area and the United States in
2020, Economic Bulletin, Issue 2/​2021, pp. 44-49.

European Central Bank (2019), Eurosystem staff macroeconomic projections for the euro area,
December­ 2019.

Federal Statistical Office (2021), 15.5% fewer business insolvencies in 2020, Press release No 161
of 31 March 2021, [Link]
C784D1753BA099E7AC65E7A5660674BB.live712

Gibbs, M., F. Mengel and C. Siemroth (2021), Work from Home & Productivity: Evidence from Per-
sonnel & Analytics Data on IT Professionals, IZA Discussion Paper, No 14336.

Google LLC (2021), Google COVID-​19 Community Mobility Reports, [Link]

covid19/mobility/?hl=en-GB, accessed on 8 October 2021.

Goolsbee, A. and C. Syverson (2021), Fear, lockdown, and diversion: Comparing drivers of pan-
demic economic decline 2020, Journal of Public Economics, Vol. 193.

Gupta, S., K. Simon and C. Wing (2020), Mandated and Voluntary Social Distancing During the
COVID-​19 Epidemic, Brookings Papers on Economic Activity, COVID-​19 and the Economy: Part One
(Summer 2020), pp. 269-315.

Hale, T., J. Anania, N. Angrist, T. Boby, E. Cameron-​Blake, M. Di Folco, L. Ellen, R. Goldszmidt,

L. Hallas, B. Kira, M. Luciano, S. Majumdar, R. Nagesh, A. Petherick, T. Phillips, H. Tatlow, S. Webster,­
A. Wood and Y. Zhang (2021), Variation in government responses to COVID-​19, BSG Working
Paper, No 2020/​032, Version 12.0, July 2021.

Hatzius, J., A. Tilton and D. Struyven (2020), Measuring the Impact of Lockdowns and Social
Distancing­on Global GDP, Goldman Sachs, Global Economics Analyst, Newsletter of 26 April
2020.
 Deutsche Bundesbank
Monthly Report
October 2021
63

Hinterlang, N., S. Moyen, O. Röhe and N. Stähler (2021), Gauging the Effects of the German
COVID-​19 Fiscal Stimulus Package, Deutsche Bundesbank Discussion Paper, forthcoming.

International Monetary Fund (2021), Differences In Output Performance Between Europe And the
US During COVID-​19, Regional Economic Outlook Update: Europe, April 2021, p. 7.

International Monetary Fund (2020a), World Economic Outlook Update: A Crisis Like No Other, An
Uncertain Recovery, June 2020.

International Monetary Fund (2020b), The Great Lockdown: Dissecting the Economic Effects,
World Economic Outlook, October 2020, pp. 65-84.

Jordà, Ò. (2005), Estimation and Inference of Impulse Responses by Local Projections, American
Economic Review, Vol. 95(1), pp. 161-182.

Kakkad, J., C. Palmou, D. Britto and J. Browne (2021), Anywhere Jobs: Reshaping the Geography
of Work, Tony Blair Institute For Global Change.

Khalil, M. and M.-D. Weber (2021), Chinese supply chain shocks, Deutsche Bundesbank Discussion
Paper, mimeo.

Kilian, L., N. Nomikos and X. Zhou (2021), Container Trade and the U. S. Recovery, CEPR Discus-
sion Paper, No 16277.

Kong, E. and D. Prinz (2020), Disentangling policy effects using proxy data: Which shutdown
policies­affected unemployment during the COVID-​19 pandemic?, Journal of Public Economics,
Vol. 189(1).

Krippner, L. (2013), Measuring the stance of monetary policy in zero lower bound environments,
Economics Letters, Vol. 118(1), pp. 135-138.

Meier, M. and E. Pinto (2020), Covid-​19 Supply Chain Disruptions, Collaborative Research Center
Transregio 224, Discussion Paper, No 239.

Morikawa, M. (2021), Work-​from-​Home Productivity during the COVID-​19 Pandemic: Evidence

from Surveys of Employees and Employers, SSPJ Discussion Paper, No 20-007.

National Bureau of Economic Research (2021), Determination of the April 2020 Trough in US Eco-
nomic Activity, Business Cycle Dating Committee Announcement, 19 July 2021.

OECD (2021a), Tourism GDP (indicator), [Link] accessed on 13 Sep-

tember 2021.

OECD (2021b), Measuring telework in the COVID-​19 pandemic, OECD Digital Economy Papers,
No 314.

OECD (2020), The OECD Weekly Tracker of activity based on Google Trends, OECD Economic Out-
look, Vol. 2020(2), December 2020, pp. 60-72.
Deutsche Bundesbank
Monthly Report
October 2021
64

Ozimek, A. (2020), The Future of Remote Work, SSRN Paper, [Link]

ssrn.3638597

Plagborg-​Møller, M. and C. K. Wolf (2021), Local Projections and VARs Estimate the Same Impulse
Responses, Econometrica, Vol. 89(2), pp. 955-980.

Santacreu, A. M., F. Leibovici and J. LaBelle (2021), Global Value Chains and U.S. Economic Activ-
ity During COVID-​19, Federal Reserve Bank of St. Louis Review, Vol. 103(3).

Statistics Canada (2021), Working from home: Productivity and Preferences, [Link]
[Link]/n1/en/daily-quotidien/210401/[Link]?st=hhC17sAo

World Health Organization (2021), The true death toll of COVID-​19: Estimating global excess mor-
tality, Data story, May 2021, [Link]
estimating-global-excess-mortality, accessed on 30 August 2021.

Wolter, M. I., A. Mönnig, T. Maier, C. Schneemann, S. Steeg, E. Weber and G. Zika (2021), Lang-
fristige Folgen der Covid-​19-​Pandemie für Wirtschaft, Branchen und Berufe, IAB Research Report,
2/​2021.