RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
3
CHAPTER
RECESSIONS AND RECOVERIES IN LABOR MARKETS:
PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
Despite remarkable adaptation and extraordinary Introduction
policy support in many economies, economic turmoil Over a year since its onset, the COVID‑19 pan-
and labor market dislocations from the COVID-19 demic continues to generate widespread economic
pandemic shock continue, with highly unequal effects disruptions and worker dislocations. Even with the
across workers. Youth and the lower-skilled are among extraordinary policy support already deployed (out-
the most heavily impacted, with sharp rises in unem- lined in Chapter 1 of the April 2021 World Economic
ployment rates, which already tend to be at higher Outlook (WEO) and of the April 2021 Fiscal Monitor),
levels. Some of these effects reflect the asymmetric, average unemployment rates are up and labor force
sectoral, and occupational nature of the COVID-19 participation down compared with their pre-pandemic
shock, with less-skill-intensive sectors tending to be averages in both advanced and emerging market and
hit harder. The shock is also accelerating preexisting developing economies, according to the latest data
employment trends, hastening a shift away from sectors (Figures 3.1 and 3.2, panels 1 and 2).
that are more vulnerable to automation. Worker The employment impacts from the pandemic
reallocation across sectors and occupations is more likely have been highly unequal across groups of workers
after an unemployment spell, but it comes at a high (Figures 3.1 and 3.2, panels 3–8). In particular, youth
cost, as average earnings fall for those who switch. Job and the lower-skilled have been hit harder in the
retention policies—those aimed at maintaining existing average advanced and emerging market and developing
employment matches—can help reduce job separations, economies, with larger rises in unemployment rates
particularly for the lower-skilled, while measures to and declines in labor force participation. Women in
support worker reallocation can boost job finding emerging market and developing economies have seen a
prospects. A new, model-based analysis shows how job slightly higher rise in unemployment and larger drop in
retention policies are extremely powerful at reducing participation than men, on average, while in advanced
scarring and mitigating the unequal impacts of a pan- economies there is little difference in average unem-
demic shock across workers, while reallocation policies ployment across genders.1 These movements in unem-
supporting job creation can help ease the adjustment ployment and labor force participation rates imply that
to the more permanent effects of the COVID-19 shock average employment rates have declined across groups.
on the labor market. Retention measures are best while In the near term, the consequences for these more
the shock is acute and social distancing high to preserve vulnerable demographic groups are potentially dire, as
ultimately viable job matches, with support relying they face earnings losses and difficult searches for job
more on reallocation measures as the pandemic sub- opportunities after unemployment spells. Even after the
sides. Careful monitoring of the intensity of the pan- pandemic abates, some of the effects on the structure
demic (including cases and deaths, the extent of social of employment may be persistent, with some sectors
distancing, and rollout of vaccines) is needed to gauge and occupations (job types) permanently shrinking and
when the economy can cope with the reduction of job others growing.2 For these persistent effects, the speed
retention support and switch toward greater reliance on
reallocation. 1Early in the crisis, studies indicated that women’s employment
was impacted more than men’s in some advanced economies,
unlike most previous downturns (Alon and others 2020). However,
The authors of this chapter are John Bluedorn (lead), Francesca with some recovery as the year proceeded, the average differences
Caselli, Wenjie Chen, Niels-Jakob Hansen, Jorge Mondragon, Ippei have diminished. See Bluedorn and others (2021) for a more
Shibata, and Marina M. Tavares, with support from Youyou Huang, in-depth exploration of the phenomenon.
Christopher Johns, and Cynthia Nyakeri. Yi Ji also provided data 2Barrero, Bloom, and Davis (2020) focuses on the experience of
support. The chapter benefited from discussions with Tito Boeri and the United States and argues that 32 percent to 42 percent of layoffs
from comments by internal seminar participants and reviewers. from the COVID-19 pandemic shock are likely to be permanent.
International Monetary Fund | April 2021 63WORLD ECONOMIC OUTLOOK: Managing Divergent Recoveries
Figure 3.1. Labor Market Conditions in Advanced Economies Figure 3.2. Labor Market Conditions in Emerging Market and
(Percentage points) Developing Economies
(Percentage points)
The COVID-19 pandemic has caused large worker dislocations in advanced
economies, with highly unequal impacts across workers, on average, hitting youth The COVID-19 shock has led to sharp deteriorations in labor markets in emerging
and the lower-skilled harder. market and developing economies, hurting youth, women, and the lower-skilled
worse, on average.
Unemployment Rate Labor Force Participation Rate
10 1. Total 1.0 100 2. Total 0.0 Unemployment Rate Labor Force Participation Rate
8 0.8 80 –0.1 12 1. Total 2 80 2. Total 0
10
6 0.6 60 –0.2 60 –1
8
4 0.4 40 –0.3 6 1 40 –2
2 0.2 20 –0.4 4
20 –3
0 0.0 0 –0.5 2
Average Change as of Average Change as of 0 0 0 –4
2018–19 2020 2018–19 2020 Average Change as of Average Change as of
(right scale) (right scale) 2018–19 2020 2018–19 2020
(right scale) (right scale)
20 3. By Age 4.0 120 4. By Age 1.0
Prime Youth 100 Prime Youth 0.5 30 3. By Age 4 120 4. By Age 2
15 3.0 Prime Youth 100
80 0.0 Prime Youth
3
10 2.0 60 –0.5 20 80 0
40 –1.0 2 60
5 1.0 10 40 –2
20 –1.5 1
0 0.0 0 –2.0 20
Average Change as of Average Change as of 0 0 0 –4
2018–19 2020 2018–19 2020 Average Change as of Average Change as of
(right scale) (right scale) 2018–19 2020 2018–19 2020
(right scale) (right scale)
8 5. By Gender 1.5 120 6. By Gender 0.2
Men Women 16 5. By Gender 2 120 6. By Gender 1
Men Women 100
6 0.0 Men Women 100 Men Women
1.0 80 12 0
80
4 60 –0.2
8 1 60 –1
0.5 40
2 –0.4 40
20 4 –2
20
0 0.0 0 –0.6
Average Change as of Average Change as of 0 0 0 –3
2018–19 2020 2018–19 2020 Average Change as of Average Change as of
(right scale) (right scale) 2018–19 2020 2018–19 2020
(right scale) (right scale)
10 7. By Skill 2.0 120 8. By Skill 0.8
14 7. By Skill 2 120 8. By Skill 2
8 Higher-skilled 100 Higher-skilled 0.4 12
1.5 Higher-skilled 100 Higher-skilled 1
Lower-skilled 80 Lower-skilled
6 0.0 10 Lower-skilled 80 Lower-skilled
1.0 60 8 0
4 –0.4 1 60
40 6 –1
0.5 4 40
2 20 –0.8
20 –2
2
0 0.0 0 –1.2
Average Change as of Average Change as of 0 0 0 –3
2018–19 2020 2018–19 2020 Average Change as of Average Change as of
(right scale) (right scale) 2018–19 2020 2018–19 2020
(right scale) (right scale)
Sources: International Labour Organization; Organisation for Economic
Sources: International Labour Organization; Organisation for Economic
Co-operation and Development; and IMF staff calculations.
Co-operation and Development; and IMF staff calculations.
Note: “Change” is the average change in the indicated variable across countries in
Note: “Change” is the average change in the indicated variable across countries in
the group, calculated relative to its average value over 2018–19.
the group, calculated relative to its average value over 2018–19.
Higher-skilled = tertiary education and above; Lower-skilled = above secondary
Higher-skilled = tertiary education and above; Lower-skilled = above secondary
and nontertiary education and below. Prime age = 25 to 54 years old; Youth = 15
and nontertiary education and below. Prime age = 25 to 54 years old; Youth = 15
to 24 years old. To account for sample coverage changes, the average within the
to 24 years old. To account for sample coverage changes, the average within the
group over time is calculated from the normalized time fixed effects from a
group over time is calculated from the normalized time fixed effects from a
regression of the indicated variable on country and time fixed effects
regression of the indicated variable on country and time fixed effects
(Karabarbounis and Neiman 2014). See Online Annex 3.1 for further details.
(Karabarbounis and Neiman 2014). See Online Annex 3.1 for further details.
64 International Monetary Fund | April 2021CHAPTER 3 RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
with which economies can reemploy and reallocate The main findings of the chapter are:
workers across sectors and occupations will determine •• The COVID-19 pandemic shock is accelerating preex-
how long lived the effects on employment are. isting employment trends with uneven impacts across
With an eye to understanding the potential after- demographic groups. The shock has hit sectors that are
math of the COVID-19 shock, this chapter studies more vulnerable to automation harder. Around the
unemployment, labor market transitions (job findings, world, youth and the lower-skilled are more heavily
separations, and employment changes across sectors impacted, on average, partly reflecting differences in
and occupations), and earnings over the business cycle workforce composition across sectors. In emerging
and across demographic groups. It investigates how market and developing economies, women’s unem-
policies—specifically those supporting job retention ployment has risen more than men’s, on average, while
(preserving and maintaining existing employment in advanced economies there is not much difference.
matches) and worker reallocation (fostering new •• The pandemic recession is likely to inflict sizable costs on
matches, assisting job search, and helping workers unemployed workers, particularly the lower‑skilled. While
obtain useful new skills)—can mitigate the damage it is not uncommon for workers to reallocate across
done by the shock. Given that the ultimate effects sectors and occupations after spells of unemployment,
of the pandemic on the economy’s structure remain such reallocation is costly. On average, workers finding
highly uncertain and may vary across countries, the reemployment in an occupation different from their
chapter uses a newly developed labor market model to previous job experience an average earnings penalty
examine how policies and the shock’s persistence inter- of about 15 percent, pointing to large costs—both
act. Drawing on empirical and model-based analyses, personal and social—from reallocation via unem-
the chapter investigates the following key questions: ployment.3 Lower-skilled workers experience a triple
•• What is the sectoral character of the COVID‑19 whammy: they are more likely to be employed in
pandemic recession so far and how does it compare sectors more negatively impacted by the pandemic; are
with past recessions? more likely to become unemployed in downturns; and,
•• How have labor market inflows and outflows across those who are able to find a new job, are more likely to
sectors behaved in recessions and recoveries? Do need to switch occupations and suffer an earnings fall.
recessions tend to amplify sectoral employment •• Both retention and reallocation policies can help
trends (in vulnerability to automation)? mitigate the impact on workers. The persistence
•• How do individual-level labor market outcomes and asymmetry of the pandemic shock are crucial
(including sectoral and occupational employment for the choice between retention and reallocation.
transitions and associated earnings gains/losses) Job retention policies—such as wage subsidies and
behave and differ across demographic groups (such short-term work schemes—are effective in lowering
as age, gender, and skill) and the business cycle? separations, while worker reallocation policies—such
•• How effective are labor market policies encourag- as hiring incentives, job search-and-matching assis-
ing job retention versus worker reallocation against tance, and retraining programs—boost job finding
the adverse effects from asymmetric shocks across and on-the-job occupational switches by those still
sectors and occupations? Does the persistence of the in employment. Historically, the lower-skilled have
shock matter? tended to benefit more from job retention policies,
while worker reallocation policies have bolstered
Importantly, the chapter reflects on what the women’s and youth’s prospects more.
findings imply for the labor market during and after oo For a transitory and asymmetric shock (such as a
the COVID-19 pandemic recession and the role of lockdown or sharp rise in social distancing affect-
policies. Due to data availability constraints, much of ing sectors differently), job retention policies are
the historical empirical analysis is based on a sample extremely powerful in reducing unemployment
of largely advanced economies over the past 30 years. and providing near-term income insurance.
As such, the patterns in labor markets identified and
assessments of policy effectiveness and options may
3See Helliwell and Huang (2014) and Reichert and Tauchmann
be less applicable to economies where large shares of
(2017) for evidence on the large social costs of unemployment aris-
employment are informal (as in some emerging market ing from spillovers across individuals to the larger labor market and
and developing economies). increasing perceptions of job insecurity.
International Monetary Fund | April 2021 65WORLD ECONOMIC OUTLOOK: Managing Divergent Recoveries
oo For a permanent shock (such as a permanent shift There are some important caveats to the findings.
in demand across sectors or drop in productivity First, country and time coverage vary across empirical
in some sectors), worker reallocation policies that exercises because of differences in data availability and
foster job creation perform better in the long are typically more representative of advanced econo-
term and hasten adjustment toward the new mies’ experiences. Recent studies of emerging market
equilibrium. and developing economies suggest that economies with
oo Where the shock is a mix of transitory and per- larger shares of informal employment are suffering
manent components, a policy package that favors initially sharper declines in employment from the pan-
job retention while social distancing is pervasive, demic, but that they may also be poised to experience
and then reallocation once it lifts, better mitigates faster labor market recoveries after the shock passes
unemployment dynamics. as informal jobs can be (re)created more quickly.5
The lack of channels to provide job retention support
Taken together, the findings suggest that countries to informally employed workers may also mean that
with fiscal space should maintain support for job greater reliance on policies such as cash transfers may
retention until the pandemic abates markedly, helping be needed to provide income insurance.6 Second, given
to avoid socially costly unemployment spells and to that national policies and individual labor market
dampen the effects on more disadvantaged worker outcomes may be affected by many different variables
groups. In particular, the findings suggest that the use for which the analysis is unable to fully account, the
of retention policies could be linked to the duration estimated effects of national-level job retention and
and intensity of the pandemic. Uncertainties about worker reallocation policies on individual-level labor
the pandemic and its path mean that the phaseout market transition probabilities should be interpreted as
of such measures is more complicated in practice; it associational rather than causal. Third, the model-based
requires careful monitoring of the pandemic (including analysis should be considered illustrative, highlighting
rollout of vaccines) and judgment of the economy’s key considerations relevant to the choice between job
ability to weather a reduction in support. Although retention and worker reallocation support. Uncer-
the model-based analysis is unable to take account of tainties about the size and structure of permanent
tight fiscal space constraints, the powerful effects of job effects from the COVID-19 shock are large, and past
retention policies in avoiding deeper and more pro- recoveries may not be fully representative. Policy-
tracted employment deterioration from the pandemic makers may need to be nimble in their responses (see
suggest that such measures should be prioritized. also Chapter 2).
Policies could also be designed to target more- This chapter begins with a look at differences in
affected worker groups—for example, increasing the labor market impact of the COVID‑19 pandemic
wage subsidies for youth or lower-skilled workers—to recession across sectors; how past downturns compare;
discourage firms from letting these workers go and and the relationship between sectoral reallocation and
reduce the unequal impact of the shock. As a recovery the business cycle through the lens of worker flows,
gets under way, a more vigorous deployment of worker focusing on vulnerability to automation. It then turns
reallocation support can hasten labor market adjust- to individual-level labor market transitions, earnings
ment. However, it is important to be realistic about changes, and differences across demographic groups. It
how quickly progress in reallocation—particularly the also estimates how these have varied across past busi-
long-term shifting of workers from occupations more- ness cycles and what these patterns may imply for the
to less-vulnerable to automation—can be achieved COVID-19 shock. The penultimate section presents
given skill mismatches. Human capital investments empirical estimates of the associations of job retention
to help workers reskill for new occupations will
5For
take time.4 in-depth looks at specific emerging market and developing
economies and how informality in employment may affect the
impact of the COVID-19 shock, see Alfaro, Becerra, and Eslava
4See World Bank (2018, 2019) for how policymakers can adjust (2020); Balde, Boly, and Avenyo (2020); Kesar and others (2020);
policies and improve education and lifetime learning systems to and Levya and Urrutia (2020), among others. Historically, greater
help workers adapt to the changing nature of work as technology informality has been associated with a lower cyclical sensitivity of
advances. See also Edelberg and Shevlin (2021) for a discussion of employment (Ahn and others 2019).
how policies to boost workforce training may help ease the employ- 6See Díez and others (2020) for a discussion of delivery modalities
ment recovery from the pandemic in the United States. for support to informal workers during the pandemic.
66 International Monetary Fund | April 2021CHAPTER 3 RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
and worker reallocation policies with labor market Figure 3.3. Sectoral Employment Growth and the Business
transitions and the findings from a model-based Cycle
analysis illustrating the effectiveness of these policies in
responding to a lockdown or social-distancing shock. COVID-19 has hit sectors unevenly, with the most-impacted different than in past
recessions, but still hastening an uptick in automation trends.
The chapter concludes with a summary of the main
takeaways and policy implications. More vulnerable to automation Less vulnerable to automation
4 1. Average in 2020
(Year-over-year percent change)
Sectoral Shocks, Trends in Reallocation, and the 2
Business Cycle
0
Reflecting the larger direct impact of the pandemic
on more contact-intensive work and sectors, the –2
COVID‑19 shock has been highly asymmetric in its
–4
employment effects across sectors (Figure 3.3, panel 1;
see also Chapter 2). –6
Total
economy
Agri.
Min./Ener.
Manuf.
Constr.
Trade/Acc.
Arts/Serv.
Info./Com.
Fin./Ins.
Real Est.
Pro. Serv.
Edu./Heal.
The COVID-19 Shock’s Impacts Differ across Sectors
In advanced economies, the sharpest drops in 4 2. Average in Past Recessions
(Year-over-year percent change)
employment were in the wholesale and retail trade, 2
transportation, accommodation and food service, 0
and arts and entertainment sectors, unlike during
–2
previous recessions over the past 50 years, when the
manufacturing and construction sectors were typically –4
the most negatively impacted (Figure 3.3, panel 2). –6
Some sectors, such as information and communica- –8
tion and finance and insurance, have even experienced
Total
economy
Agri.
Mining
Manuf.
Utilities
Constr.
Trade
Transport
Acc./Food
Arts/Serv.
Info./Com.
Fin./Ins.
Real Est.
Pro. Serv.
Public Ad.
Educ.
Health
employment growth during the pandemic, further
highlighting divergent fortunes. Interestingly, the
broad sectoral pattern is similar to that observed in 4 3. Average Trend Growth
(Year-over-year percent change)
previous recessions, which seem to accelerate preexist-
2
ing structural trends hastening a shift in employment
away from sectors more vulnerable to automation
0
(Figure 3.3, panel 3).7
–2
The Shock Hits Workers Unequally, with Youth and the
–4
Lower-Skilled More Affected
Total
economy
Agri.
Mining
Manuf.
Utilities
Constr.
Trade
Transport
Acc./Food
Arts/Serv.
Info./Com.
Fin./Ins.
Real Est.
Pro. Serv.
Public Ad.
Educ.
Health
Inequalities in the labor market impacts of the
pandemic across demographic groups highlighted
in the introduction may in part reflect these asym-
Sources: Choi and others (2018); EU KLEMS; International Labour Organization;
metric sectoral impacts of the COVID‑19 shock. Organisation for Economic Co-operation and Development; Statistics Canada; US
Bureau of Economic Analysis; World KLEMS; and IMF staff calculations.
7Some recent studies have also classified jobs according to their Note: Sector groupings in panel 1 are slightly different from those in panels 2 and
“teleworkability” (for example, Dingel and Neiman 2020). Most 3 due to reporting differences in the quarterly sectoral national data. Total
economy indicates employment for the economy as a whole. Sectors are classified
teleworkable jobs are found in sectors that are classified as less vul-
according to ISIC Revision 4. Sectors are classified as more (less) vulnerable to
nerable to automation, meaning there is also a trend toward greater automation if more (less) than half their share of employment is in occupations
teleworkability in employment. However, there are some differences. classified as highly exposed to routinization (Carrillo-Tudela and others 2016).
Sectors that are less vulnerable to automation but not teleworkable Underlying data for panel 1 cover 2019:Q1–2020:Q4 and for panels 2 and 3 span
include utilities and arts and entertainment, while sectors that are 1970–2019, as available. Patterns in average trend growth are similar over the
teleworkable but more vulnerable to automation include administra- shorter period, 2010–19. See Online Annex 3.1 for further details, including the list
tive services. See Online Annex 3.1 for a tabulation. of abbreviations.
International Monetary Fund | April 2021 67WORLD ECONOMIC OUTLOOK: Managing Divergent Recoveries
Figure 3.4. Changes in Sectoral Online Job Posting Trends When split according to the proportion of these demo-
(Percent; gap in trend from a year ago, indexed to February 1, 2020)
graphic groups represented in a given sector, the latest
high-frequency data on trends in online job postings
Sectoral workforce composition accounts for some of COVID-19’s unequal impact
across groups of workers. suggest that sectors that tend to have more youth,
women, or lower-skilled workers are likely to have
10 1. By Age underperformed more than other sectors (Figure 3.4).
0 In other words, demographic differences in employ-
ment across sectors and occupations—such as a con-
–10 Higher share of youth centration of workers from disadvantaged groups—are
Lower share of youth
–20 likely contributing to differences in outcomes across
groups in the current crisis.8
–30
–40
Past Recessions Suggest COVID-19 Shock Requires
–50
Feb. Apr. Jun. Aug. Oct. Dec. Feb. Worker Reallocation
2020 20 20 20 20 20 21
Based on past shocks, it seems likely that some of
10 2. By Gender this uneven sectoral impact from the COVID‑19 pan-
demic shock reflects a longer-lived labor reallocation
0
shock that is contributing to the unemployment rise.
–10 Higher share of women As seen in the behavior of gross worker flows, built
Lower share of women
–20 up to the country level from microdata on workers,
recessions are typically characterized by declines in
–30
gross hiring rates (hires into new or existing jobs as a
–40 share of employment) and rises in gross separations
(job terminations, whether voluntary or involuntary,
–50
Feb. Apr. Jun. Aug. Oct. Dec. Feb. as a share of employment), consistent with a rise
2020 20 20 20 20 20 21
in unemployment during downturns (Figure 3.5).9
10 3. By Skill Level
8See Cajner and others (2020) on how the sectoral nature of the
0 COVID-19 shock may drive much of the disparity in effects across
Higher share of lower-skilled workers worker groups. Dam and others (2021) and Klein and Smith’s
–10
Lower share of lower-skilled workers (2021) early analysis of the COVID-19 pandemic’s impact in the
–20 United States indicate that workers from ethnic minorities (African
American and Hispanic) have been disproportionately hurt. Previous
–30 research has also pointed out the unequal effects of downturns, with
historically more disadvantaged groups (youth and ethnic minorities,
–40 among others) more likely to experience protracted unemployment
and income losses (Altonji and Blank 2004; Raaum and Røed 2006;
–50 Oreopoulos, von Wachter, and Heisz 2012; among others). Earlier
Feb. Apr. Jun. Aug. Oct. Dec. Feb.
2020 20 20 20 20 20 21 work has also suggested that composition of employment across
sectors and occupations, and hence unequal exposure to shocks, may
account for some differences (Davis and von Wachter 2011; Peiró,
Sources: EU Labour Force Survey; Indeed; Integrated Public Use Microdata Series, Belaire-Franch, and Gonzalo 2012; Albanesi and Șahin 2018). Beyond
Current Population Survey; and IMF staff calculations.
differences in the sectoral or occupational exposure to the shock,
Note: Data are as of February 16, 2021. Higher (lower) demographic representation
in employment by sector is defined as whether the share of young or lower-skilled other features that could be associated with sector of employment and
workers is above (below) the economy-wide average or whether the share of occupation may contribute to inequalities across worker groups (for
women employed is above (below) 50 percent in a sector. The sample includes a example, the prevalence of temporary versus permanent employment
mix of advanced and emerging market economies. Vertical line = March 10, 2020 contracts, strength of worker bargaining power). See Kikuchi, Kitao,
(Italy enters country-wide lockdown). See Online Annex 3.1 for further details, and Mikoshiba (2020), which finds that more employment on tem-
available at www.imf.org/en/Publications/WEO. porary contracts may account for the large impact of the COVID-19
shock on women in Japan in the early phase of the pandemic.
9Recessions are years of negative real GDP growth. Recoveries are
years after a recession when output remains below its previous histor-
ical maximum. See Online Annex 3.1, available at www.imf.org/en/
Publications/WEO, for a description of the business cycle dating
algorithm used to identify phases.
68 International Monetary Fund | April 2021CHAPTER 3 RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
Figure 3.5. Labor Market Turnover across Business Cycles Figure 3.6. Sectoral Employment, by Vulnerability to
(Percent) Automation, Skill Level, and Business Cycle
Hiring falls and separations rise in recessions compared with expansions, Employment trends favoring higher-skilled sectors that are less vulnerable to
reversing somewhat in recoveries. automation occur more as a result of joblessness spells than on-the-job sectoral
changes, accelerating during recessions.
18
Expansion Recession Recovery
Other sector Unemployment Nonparticipation Total
15 90 1. Average Employment Share1 2. Average Net Hiring Rates 5
(Percent) (Percentage points)
4
More lower-skilled
12 80
More vulnerable to 3
automation
70 2
9
1
60
6 0
50 –1
3 1990 2000 10 20 More vulnerable Less vulnerable
sectors sectors
0 6 3. Average Net Hiring Rates 4. Average Net Hiring Rates 5
Gross hiring rate Gross separation rate Gross job-to-job hiring rate in Recessions in Recoveries
4 (Percentage points) (Percentage points) 4
Sources: EU Labour Force Survey; Integrated Public Use Microdata Series, Current 3
Population Survey; and IMF staff calculations. 2
Note: Hiring and separation rates and their components are calculated as annual 2
hires/separations divided by average employment over the current and previous
0
years. All rates are statistically significantly different, except those for job-to-job 1
hiring rates for recession and recovery and those for separation rates for recovery
and expansion. See Online Annex 3.1 for further details about the data and –2 0
business cycle dating.
–4 –1
More vulnerable Less vulnerable More vulnerable Less vulnerable
sectors sectors sectors sectors
The job-to-job hiring rate (hires from the employed as
a share of employment) also tends to drop, suggesting Sources: EU Labour Force Survey; Integrated Public Use Microdata Series, Current
that reallocation through job-to-job changes is inhib- Population Survey; and IMF staff calculations.
Note: Sectors are classified as more vulnerable to automation if more than half
ited during downturns. Within the job-to-job flows, their share of employment is in occupations classified as highly exposed to
about two-thirds of all flows are within the same sec- routinization (Carrillo-Tudela and others 2016). Sectors are classified as more
lower-skilled if the sectoral share of lower-skilled employment is greater than the
tor. All of these mechanisms are likely to be operating economy-wide average. Net hiring rates are calculated as the difference between
during the COVID‑19 pandemic recession. annual hires and separations, divided by the average employment over the current
and previous year. See Online Annex 3.1 for further details.
1
To account for sample coverage changes, the average share of employment in
working-age population across selected economies over time is calculated
Sectors More Vulnerable to Automation Are Harder Hit, according to the normalized time fixed effects from a regression of the indicated
variable on country and time fixed effects (Karabarbounis and Neiman 2014).
Similar to Past Recessions
Over time, employment has been shifting away
from sectors that are more vulnerable to automation, tends to work more through joblessness, its social
and the share of employed workers with lower skills costs can be high, particularly during recessions
has fallen (Figure 3.6, panel 1). The shift reflects in when sectors that are more vulnerable to automation
part direct movement of workers from more vul- exhibit large outflows into unemployment, as is likely
nerable to less vulnerable sectors, but more often it with the COVID-19 shock (Figure 3.6, panel 3).
results from net hiring of workers from unemploy- Indeed, as remarked above, employment in sectors
ment and nonparticipation (Figure 3.6, panel 2). that are more vulnerable to automation has declined
This suggests that sectoral reallocations often happen more steeply during the COVID-19 pandemic, simi-
after a spell of nonemployment. Because reallocation larly to earlier recessions.
International Monetary Fund | April 2021 69WORLD ECONOMIC OUTLOOK: Managing Divergent Recoveries
In sum, the COVID-19 pandemic shock has been Figure 3.7. Labor Market Transition Probabilities across
Business Cycles and Demographic Groups
highly asymmetric in its employment impacts across
sectors and demographic groups. Moreover, if the past
Individual labor market transitions exhibit business cycle patterns similar to those
is any guide, these effects may have a long half-life of worker flows, but there is significant variation in prospects across demographic
and entail the need for some reallocation. In partic- groups, with youth and the lower-skilled at particular disadvantage in the labor
market.
ular, the shock is accelerating preexisting automation
trends, leading more vulnerable sectors to shrink, and 40 1. Average Probabilities across Business Cycles
encouraging employment growth in expanding sectors. (Percent)
Differences in workforce composition across sectors 30
imply that some worker groups—particularly the Expansion Recession Recovery
lower-skilled—face more tenuous job prospects. 20
Labor Market Transitions, Inequality, 10
and Recessions
0
An alternative perspective to aggregate worker flows Job finding Job separation On-the-job
emerges from an examination of individual-level labor sectoral switch
market transitions—such as an unemployed person
15 2. Average Probabilities across Demographic Groups1
finding a job, an employed person losing or separating (Percentage point deviation from the base group)
from a job, and sectoral and occupational changes in 10
employment (either on the job or after an unemploy- 5
ment spell)—which allows for demographic differ-
0
ences in prospects to be identified. As shown here,
lower-skilled workers are likely to be particularly hurt –5
Women Lower-skilled Youth
by the COVID-19 pandemic recession. –10
–15
Job finding Job separation On-the-job
Job Finding Is Lower and Job Separation Higher in sectoral switch
Recessions than in Expansions
6 3. Average Probabilities across Demographic Groups in Recessions1
The probability of finding a job is lower in recessions (Percentage point deviation from the base group in recessions)
and recoveries than in expansions, while the reverse is true 4
for job separations (Figure 3.7, panel 1). The likelihood of
2
switching the sector of employment while on the job also
tends to follow the cycle—rising in expansions and falling 0
in recessions—although the estimated difference across –2
business cycle phases is not statistically significant.10
–4 Women Lower-skilled Youth
These average labor market transition likelihoods
mask systematic differences across demographic groups. –6
Job finding Job separation On-the-job
Using a linear probability model augmented with sectoral switch
individual-level characteristics, the average effects of these
characteristics on labor market transitions are estimated. Sources: EU Labour Force Survey; Integrated Public Use Microdata Series, Current
The results suggest that finding a job is easier for young Population Survey; and IMF staff calculations.
Note: Job finding calculations comprise individuals who were unemployed in the
than prime-age workers while, on average, it is more previous year and are employed in the current year. Job separation calculations
difficult for women than men and the lower-skilled than comprise individuals who were employed in the previous year and are unemployed
in the current year. On-the-job sectoral switches comprise individuals who are
the higher-skilled (Figure 3.7, panel 2). Losing a job employed in the previous and current years and changed their sector of
tends to be more likely for the young or lower-skilled, occupation. The whiskers indicate the 95 percent confidence band. See Online
Annex 3.1 for further details.
1
Base group is prime-age and higher-skilled men.
10The procyclicality of sectoral switches in employment is also found
in the literature (Carrillo-Tudela, Hobijn, and Visschers 2014; Carrillo-
Tudela and Visschers 2014; and Carrillo-Tudela and others 2016).
70 International Monetary Fund | April 2021CHAPTER 3 RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
while the separation likelihood for women appears about changes do not occur in a vacuum; they likely depend
the same as that of men. At the same time, youth are on a worker’s employment history.
also more likely than prime‑age individuals to change Based on a panel data set of individuals from a
jobs across sectors while employed. sample of European economies, the probability of
an occupational switch and earnings change reflect
this dependence. Among those who are “on the job”
Youth and the Lower-Skilled Were Also Most Affected in (continuously employed over the past two years), occu-
Past Recessions pational switch incidence is only about 10 percent; for a
Zooming in on transitions during past recessions, worker reemployed after a one-year unemployment spell
systematic differences across groups are also evident (“via unemployment”), it is nearly five times higher,
(Figure 3.7, panel 3). Youth tend to be particularly at almost 50 percent (Figure 3.8, panel 1).13 In other
disadvantaged in finding a job and more likely to lose words, workers appear to generally prefer sticking with
one than prime-age workers in a downturn. Histori- their current occupation, unless circumstances—such as
cally, women have seen smaller drops in job finding prolonged unemployment—force them to switch.
and rises in separations than men during a recession. These worker preferences are also evident in the
The story for the lower-skilled is more complex, with earnings changes associated with occupational switches
both a higher likelihood of finding a job than the when comparing those who switched with those
higher-skilled, but also of losing it in a recession. How- who stayed in their original occupations (Figure 3.8,
ever, the separation effect likely dominates, leading the panel 2). Among the employed, those who switched
lower-skilled to be more prone to end up unemployed occupations saw an average earnings gain of about
in a recession than the higher-skilled. On-the-job sec- 2 percent, suggesting that they changed occupations
toral switches in employment show no clear pattern. because it was advantageous. In contrast, among unem-
These findings suggest that past recessions showed ployed workers who successfully found new employ-
many similar features to the current crisis, with youth ment, those who switched occupations saw an average
and the lower-skilled particularly disadvantaged in earnings penalty of about 15 percent, indicating that
the labor market. The earlier signs that women in they may have had to take a less desirable job.14
advanced economies were also hurt more on average The state of the business cycle does not appear to
by the COVID-19 shock—different from the typical significantly impact the occupational switch probabilities
patterns of previous recessions—appear to be fading. and the associated earnings changes.15 Even so, the fact
that unemployment rises in a recession and that the inci-
dence of occupational switches is larger after unemploy-
Switches in Occupations Are More Frequent after ment spells, indicates that mechanically there are likely
Unemployment Spells and Inflict Earnings Penalties to be more occupational switches and more workers
Beyond shifts in sectoral employment, labor market suffering earnings penalties on reemployment after reces-
adjustment may also reflect workers changing not only sions, including the COVID‑19 pandemic recession.
jobs, but occupations.11 This dimension has become
particularly relevant with the COVID‑19 shock, given
13The probability of an occupational switch via nonparticipation is
the premium placed on occupations that allow indi-
similar to the probability via unemployment.
viduals to work from home.12 However, occupational 14Although it is not possible to precisely compare the magnitudes
switches by workers and their associated earnings of this measure in the literature because of differences in the sample
of countries and level of disaggregation of occupation categories,
these results are broadly in line with previous studies—see Huckfeldt
11For the analysis here, these are classified into broad categories, (2018) and Gertler, Huckfeldt, and Trigari (2020). The stylized facts
such as managers, clerical support workers, craftspeople, and plant are also consistent with theories of sequential bargaining in which a
and machine operators, as per the International Standard Classifica- worker’s bargaining position is affected by their recent employment
tion of Occupations 2008 major groups occupational classification. history (see, for example, Postel-Vinay and Robin 2002; Cahuc,
See Online Annex 3.1 for more details. Postel-Vinay, and Robin 2006; and Jarosch 2015). An earnings pen-
12For instance, Hensvik, Le Barbanchon, and Rathelot (2021) finds alty with an occupational switch after an unemployment spell also
that job seekers tend to redirect their search toward less severely hit arises in a model of selective hiring (Huckfeldt 2018). Furthermore,
occupations, beyond what is predicted by the drop in vacancies during the earnings change is due mainly to changes in the hourly wage
the COVID-19 pandemic. See also Shibata (forthcoming), which change and not changes in hours worked.
finds that more teleworkable jobs are more insulated from the business 15The one exception is the earnings change associated with an
cycle, including the pandemic recession, in the United States. on-the-job occupational switch, which is smaller during a recession.
International Monetary Fund | April 2021 71WORLD ECONOMIC OUTLOOK: Managing Divergent Recoveries
Figure 3.8. Occupational Switches statistically significant). Youth also see larger earnings
gains from on-the-job occupational switches. Com-
Occupational switches after periods of unemployment are common but costly in
earnings. paring the lower-skilled to the higher-skilled, there are
no statistically significant differences in occupational
80 1. Occupational Switch 2. Earnings Change due to 20 switch incidence nor their associated earnings changes,
Probability Occupational Switch
(Percent, conditioned on the (Percent, deviation of
although there are some signs that the lower-skilled
10
60 same labor market switchers from stayers) may experience a larger earnings penalty after an occu-
transition) pational switch via unemployment.
0
40 These findings on occupational switches and their
–10 associated earnings changes across demographic groups
20 do not differ much between expansion and recession
–20
periods. However, among lower-skilled workers able
0 –30 to find reemployment, the likelihood of switching
On-the-job Via unemployment On-the-job Via unemployment occupations via unemployment increases during a
recession.17 This is particularly worrisome in light of
Women Lower-skilled Youth the COVID‑19 pandemic recession, given that it sug-
gests that the lower-skilled are likely being hit with a
16 3. Occupational Switch 4. Earnings Change due to 80
Probability1 Occupational Switch1 triple whammy: they are more likely to be employed in
12 (Percentage points, (Percentage points, 60 sectors more negatively impacted by the pandemic; are
deviation from base group) deviation from base group)
8 40 more likely to become unemployed in downturns; and
those who find a new job are also more likely to have
4 20
had to switch occupations and suffer an associated
0 0 earnings penalty.
–4 –20
–8 –40 Policy Responses to the COVID-19 Shock:
On-the-job Via unemployment On-the-job Via unemployment
Job Retention versus Worker Reallocation
Sources: EU Statistics on Income and Living Conditions; and IMF staff calculations. As the previous sections have shown, labor market
Note: Occupational switches on-the-job are calculated from individuals who are transitions tend to track the business cycle, with the
employed in the current and previous year and switched occupations.
Occupational switches via unemployment are calculated from individuals who are probabilities of job separation rising and job find-
employed in the current year and were unemployed last year and switched ing falling with adverse shocks, and youth and the
occupations (based on their occupation of record two years before when last
employed). The whiskers indicate the 95 percent confidence band. See Online lower-skilled tending to be hurt even more, on average.
Annex 3.1 for further details. Can policies help mitigate these effects while also eas-
1
Base group is prime-age and higher-skilled men.
ing any needed labor market adjustment?
The COVID‑19 pandemic has prompted extraor-
When comparing the incidence and earnings dinary policy support in many countries, devoted
consequences of occupational switches across demo- largely to preserving employment relationships and
graphic groups, some notable differences are apparent. providing workers with income insurance (often
Women are less likely than men to switch occupations, through expanded eligibility for and generosity
either while on the job or after a period of unemploy- of unemployment benefits; Figure 3.9).18 As the
ment. However, once women switch occupations, the pandemic continues, discussion focuses more and
associated earnings change (whether gain or penalty)
tends to be larger than it is for men.16 Youth are much 17It is important to emphasize that the results shown here on
more likely than prime-age individuals to switch occupational switch probabilities and associated earnings changes for
occupations, either on the job or via unemployment the lower-skilled already select for lower-skilled workers who found
a job after a period of unemployment and exclude lower-skilled
(although the difference via unemployment is not workers who could not find a job.
18See the IMF’s COVID-19 Policy Tracker for details on specific
measures. Importantly, any disincentives for reemployment from
16See Montenovo and others (2020) and Shibata (forthcoming) extensions to unemployment benefit schemes—key insurance for
for related evidence on the distributional impacts of COVID-19 in those who have lost jobs—appear to be markedly reduced during
the US labor market. recessions (Schmieder, von Wachter, and Bender 2012).
72 International Monetary Fund | April 2021CHAPTER 3 RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
Figure 3.9. Public Spending on Retention and Reallocation between retention and reallocation policies in respond-
Policies: Before COVID-19 and Response to COVID-19
(Percent of GDP)
ing to an adverse lockdown or social-distancing shock.
The laboratory of the model enables key features of
Average public spending to preserve employment after the COVID-19 shock is the pandemic shock—such as its asymmetric impacts
dramatically larger than job retention spending in the past. The rise in health across occupations—and policies to be considered.
sector spending alone is on par with average spending on reallocation in the past.
2.0
Empirical Estimates of Labor Market Policy Effectiveness
Building on the analysis of individual-level labor
market transitions, variables capturing spending as
1.5 a share of average income per unemployed person
on labor market policies aimed at job retention and
worker reallocation are included in the linear proba-
1.0 bility model. Although this model incorporates fixed
effects (country and time) and macroeconomic controls
(such as the output gap), omitted variables correlated
with the labor market policy variables remain a con-
0.5 cern, such that the results should be interpreted as
associational rather than causal.20
Focusing only on relationships that were estimated
0.0 to be statistically significant, job retention policies are
Retention, Reallocation, Preserving Health care sector found to lower job separation probabilities, on average,
pre–COVID-19 pre–COVID-19 employment, 2020 increase, 2020
while worker reallocation policies raise the likelihood
Sources: IMF, COVID-19 Policy Tracker; Organisation for Economic Co-operation of job finding and on-the-job occupational switches,
and Development; and IMF staff calculations. consistent with what many models of such policies
Note: Bars show the average public spending on the indicated area as a share of
GDP. See Online Annex 3.1 for further details.
suggest (Figure 3.10, panel 1).21 At the same time,
retention policies also appear to be associated with a
higher overall likelihood of on-the-job occupational
more on the roles of two broad sets of policies and switches and reallocation policies with a lower separa-
when to use them: those aimed at job retention tion probability, which are more puzzling. These results
(maintaining existing matches between workers and may reflect imperfect measurement of job retention
employers) and those aimed at worker reallocation and worker reallocation policies as aggregates of
(creating new jobs and facilitating workers’ shift spending to improve labor market functioning. These
away from shrinking and toward growing sectors and include spending on training programs—delivered
occupations).19 either on the job in the case of retention, or outside
To make some headway on this question, this of work where reallocation is the aim. To the extent
section first provides an empirical assessment of the that such programs increase a worker’s productivity,
effects of country-level public spending in the broad they may also raise their value to their employers
areas of job retention and worker reallocation pol-
icies on individual-level labor market transitions.
20Although the fixed effects do effectively capture the average
However, recognizing that these estimates should be
impacts of country-specific characteristics (such as the stringency
interpreted as associational rather than causal, and of labor market regulations and the structure of labor market
that the pandemic shock possesses features not seen institutions) on the outcomes, the impacts of these characteristics
in recent history, it then presents a newly developed on the effectiveness of the policy interventions explored here cannot
be independently assessed. The policy effects shown represent the
search-and-matching model to study the choice average policy effect.
21In a canonical Diamond-Mortensen-Pissarides search-and-matching
19Specifically, policy tools to encourage job retention include model of the labor market, layoff taxes (a kind of job retention
wage subsidies, short-term work schemes, and partial unemployment policy) reduce job destruction while having an ambiguous effect on
benefits, while those that foster worker reallocation include hiring job creation (Pissarides 2000). In contrast, hiring subsidies (a kind of
and start-up incentives, job search-and-matching assistance, and worker reallocation policy) in the model increase both job creation and
retraining programs. job destruction.
International Monetary Fund | April 2021 73WORLD ECONOMIC OUTLOOK: Managing Divergent Recoveries
Figure 3.10. Effects of Job Retention and Worker (reducing separations) as well as their interest in and
Reallocation Policies
capability of switching occupations while on the job.
Retention and reallocation policies may also have
Job retention and worker reallocation policies can help mitigate adverse shocks
and improve labor market functioning. different impacts across demographic groups, poten-
tially reflecting demographic differences in employ-
20 1. Overall Effects of Policies on Labor Market Transition ment in sectors and occupations benefiting from these
Probabilities
16
(Percent) policies as well as direct targeting of specific groups.22
12
Job retention Worker reallocation
The empirical results suggest that job retention policies
8 have tended to lower job separation probabilities more
4 for the lower-skilled than the higher-skilled, while
0 worker reallocation policies have tended to boost job
–4 finding chances for youth and women more than for
–8 prime-age individuals and men (Figure 3.10, panels 2
–12 and 3). The results are consistent with a greater risk
Job finding Job separation On-the-job occ. switch of layoff for the lower-skilled after an adverse shock
and, thus, their greater benefit from retention policies.
Women Lower-skilled Youth
In the case of youth, the results may reflect a greater
4 2. Differential Effects of Job Retention Policies across
capability to benefit from reallocation spending related
Demographic Groups1 to training. Women’s typically weaker labor force
2 (Percentage points) attachment may also translate into a greater sensitivity
to reallocation policies that enhance job finding.
0
–2
Economic Policy Responses to a Pandemic: Model-Based
–4 Analysis of Job Retention and Worker Reallocation
The preceding empirical analysis suggests that
–6
Job finding Job separation On-the-job On-the-job retention and reallocation policies can be effective tools
sec. switch occ. switch to respond to the labor market deterioration caused
by the COVID-19 pandemic recession. As remarked
4 3. Differential Effects of Job Reallocation Policies across
Demographic Groups1 earlier, to address concerns that the empirical estimates
3 (Percentage points) are associational and better disentangle the effects of
2
policies, this chapter also presents a newly-developed
labor market search-and-matching model to study the
1 roles of job retention versus worker reallocation poli-
0 cies in responding to the COVID-19 shock.23
The model incorporates several features that are
–1
essential to a better understanding of labor market
–2 support measures at this juncture. There are two occu-
Job finding Job separation On-the-job On-the-job
sec. switch occ. switch pations in the economy, which differ in their contact
intensity (and exposure to the pandemic shock). Work-
Sources: EU Labour Force Survey; EU Statistics on Income and Living Conditions; ers in the two occupations differ in their productivities.
Integrated Public Use Microdata Series, Current Population Survey; and IMF staff
calculations. Firms enter and exit freely in the model, paying a
Note: Panel 1 shows the percent change in the indicated transition probability cost to post a vacancy (create a job). Firms also make
(relative to its average value) associated with a 1-percentage-point increase in the
indicated policy spending as a share of average income per unemployed person.
different employment offers, depending on workers’
Panels 2 and 3 show the percentage points of the indicated transition probability
as deviations from the base group. Only estimated effects that are statistically 22These estimated differential effects are likely better identified
significant at the 95 percent level are shown. See Online Annex 3.1 for further
details, including for the specific means of the labor market transition than those for the overall policy effects, given that they are adjusted
probabilities. occ. = occupational; sec. = sectoral. for the impact of any omitted variables by country-year that could
1 be confounded with labor market policies.
Base group is prime-age and higher-skilled men.
23The model calibration is partially informed by the empirical
results. See Online Annex 3.1 for further details.
74 International Monetary Fund | April 2021CHAPTER 3 RECESSIONS AND RECOVERIES IN LABOR MARKETS: PATTERNS, POLICIES, AND RESPONSES TO THE COVID-19 SHOCK
productivity.24 As in the empirical results, workers in Figure 3.11. Model Simulations with Lockdown Shocks and
the model who switch occupations while on the job Labor Market Policies
experience a modest earnings gain, whereas workers
The unemployment rise is larger for the same-size lockdown shock when part of
who switch after an unemployment spell see a marked the shock is permanent. Retention policies are powerful in reducing
drop in earnings. An unemployment benefit system unemployment over the short term, while reallocation policies work better over the
long term and after a permanent shock.
operates in the background, offering some insurance to
unemployed workers.
Unemployment Rate and Distribution of Workers, by Occupation
The COVID-19 shock is modeled as an adverse (No-policy scenario)
“lockdown” shock associated with an increase in social
distancing that hurts one of the two occupations Less-impacted occ. More-impacted occ.
Unemployment rate
more than the other. The shock is set to replicate (right scale) (right scale)
the initial increase in unemployment observed in the
30 1. Transitory Shock 70 30 2. Hybrid Shock 70
United States and is presumed to last for four periods (Percent) (Percent)
(quarters). Given uncertainties about the persistence 60 60
of the shock, two cases are considered: (1) a transitory 20 20
50 50
shock, where productivities return to their initial levels
40 40
after the shock abates; and (2) a more likely hybrid 10 10
shock, which is largely transitory but with some per- 30 30
manent component (specifically, half of the shock to 0 20 0 20
the more-impacted occupation is permanent). 0 3 6 9 12 0 3 6 9 12
Three policy scenarios are considered and compared
against a no-policy intervention benchmark: (1) job Effect and Cost of Labor Market Policies during Different Shocks
(Deviation from no-policy scenario)
retention support, in which the government provides
transfer payments to firms to support a portion of their 2 3. Unemployment Rate 4. Deficit 60
Retention
wage bill when the match between a firm and worker (Percentage points) (Percent)
Reallocation
becomes unprofitable; (2) worker reallocation support, 40
0 Package
in which the government offers a subsidy to firms to 20
reduce their vacancy cost and stimulate job creation; –2
Retention 0
and (3) a package, which first provides job retention Reallocation
–4 Package
support and then worker reallocation support.25 In –20
the first two scenarios, support is coincident with the –6 –40
transitory component of the shock (for four quarters), 0 4 8 12 16 0 4 8 12 16
while in the package, worker reallocation support is
3 5. Unemployment Rate 6. Unemployment Rate 0.3
offered after the transitory component has passed but under Retention under Reallocation
the permanent effects are still unfolding. (Percentage points) (Percentage points)
To get a sense of what the persistence of the lock- 0 0.0
Transitory shock
down shock means for the economy, consider the Hybrid shock
no-policy intervention benchmarks under the transi- –3 –0.3
Transitory shock
tory and hybrid shocks (Figure 3.11, panels 1 and 2). Hybrid shock
When the shock hits the economy, it reduces the out- –6 –0.6
put produced by firms and workers, making some job 0 4 8 12 0 4 8 12
matches unprofitable and leading to job losses and a
sharp rise in unemployment. Given that a firm’s profit- Source: IMF staff calculations.
Note: The x-axis indicates the number of quarters after the shock starts. Package
ability increases with worker productivity, lower-skilled comprises a sequence of retention and reallocation policies. Panels 3 and 4 show
responses to the hybrid shock. See Online Annex 3.1 for the definition of different
shocks and policy measures. occ. = occupation.
24Wages are fixed for the duration of the job match once the firm
and the worker agree.
25Government transfers for job retention have an upper limit
calibrated to replicate public expenditure on job retention policies
observed in the data. Policies are financed using public debt in the
short term, which the government pays back over time.
International Monetary Fund | April 2021 75You can also read
