## 2.1    Impact on Mobility

## Source details

**Canonical URL:** [2.1    Impact on Mobility](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020234-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020234-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020234-print-pdf.pdf.json)

---

### Data and Identification
- High-frequency proxies used:
  - Google Community Mobility Reports: daily attendance rates at various locations relative to pre-crisis levels (national and, for various countries, sub-national).
  - Indeed job postings: daily job postings in 22 countries, disaggregated by employment sector.
- Time horizon of analysis:
  - "first seven months of the pandemic"
  - Mobility and job postings computed as the daily average over "the first semester of 2020".
  - Real GDP growth for "the first semester of 2020" computed with respect to "the first semester of 2019".
- Econometric approach:
  - Local projections regressing the mobility index on the stringency of lockdowns and the number of COVID-19 infections.
  - Regressions control for the stage of the epidemic to separate lockdown effects from voluntary social distancing.
  - Sub-national estimations focus on regions less affected by COVID-19 in countries that adopted national lockdowns to strengthen identification; identification relies on national lockdowns often imposed in response to localized outbreaks and thus being largely exogenous to low-infection regions.
- Identification and methodology details:
  - Empirical approach: local projection regressions (Jordà, 2005) estimated with country fixed effects and time fixed effects.
  - Key variables: mobi,t+h; ln∆casesi,t−p; locki,t−p (Oxford stringency index reconstructed to exclude public information campaigns).
  - Controls: a week worth of lags of dependent variable, lags of cases and lockdown stringency, country fixed effects, time fixed effects; standard errors clustered at the country level.
  - Sample coverage:
    - National analysis: 128 countries between early February and mid-July, 2020.
    - Subnational robustness: 422 subnational regions in 15 G20 countries (regions with the largest case counts and any region with more than 20 percent of the country’s total cases were excluded).

### Main Findings on Mobility and Economic Activity
- Both lockdowns and voluntary social distancing in response to rising COVID-19 infections had strong detrimental effects on economic activity as proxied by mobility and job postings.
- Lockdowns and voluntary social distancing played comparable roles in driving the drop in mobility across the full set of countries.
- Heterogeneity across country groups:
  - Advanced economies: the contribution of voluntary distancing was stronger, reflecting greater ability to work from home and to sustain temporary unemployment via personal savings and government benefits.
  - Low-income countries: lockdowns played a much stronger role, reflecting limited financial means to refrain from economic activities.
- Correlations:
  - The collapse in mobility over the first six months of 2020 correlates well with the decline in real GDP growth.
  - Job postings display a tight negative correlation with unemployment rates over the same period.
- Quantitative effects (national and subnational):
  - A full lockdown leads to a very significant decline in mobility:
    - National-level estimate: impact reaches about 25 percent after a week and mobility then resumes gradually.
    - Subnational estimates: negative effect corroborated; impact is modestly larger and more persistent than national estimates.
  - For a given lockdown stringency, rising COVID-19 cases induce voluntary reductions in mobility:
    - National estimate: a doubling of daily COVID-19 cases leads to a contraction in mobility by about 2 percent.
    - Subnational estimate: a doubling of daily COVID-19 cases leads to a contraction in mobility of 1.7 percent after 30 days.

### Behavior around Lockdown Measures and Asymmetries
- Lifting lockdowns is not sufficient for a strong sustained recovery if health risks remain; voluntary distancing can persist.
- Easing lockdowns tends to have a positive effect on mobility but:
  - "the impact is weaker than that of tightening lockdowns."
  - There is an asymmetric effect: tightening produces larger mobility reductions than loosening produces mobility increases.
- Interaction and asymmetric evidence:
  - Introduction of a full lockdown leads to a decline in mobility of about 26 percent one week after tightening.
  - Lifting restrictions boosts mobility by only about 18 percent over the same period; the difference is statistically significant.
- Policy implication:
  - Economies are likely to operate below potential as long as health risks persist; policymakers should be cautious about lifting lockdowns prematurely.

### Lockdowns, Infections, and Timing
- Using the same empirical framework, lockdowns can substantially reduce infections.
- Effects on COVID-19 cases tend to materialize a few weeks after lockdown introduction, consistent with the incubation period of the virus and testing times.
- Rapid intervention matters:
  - Lockdowns are particularly effective in curbing infections if introduced at an early stage of a country’s epidemic.
- Reassessment of the lives-versus-livelihoods trade-off:
  - Because infections themselves depress economic activity through voluntary distancing, lockdowns that reduce infections may facilitate faster economic recovery, potentially compensating short-term economic costs with higher future economic activity.

### Nonlinearities and Optimal Design
- Marginal effects reported:
  - More stringent lockdowns have decreasing marginal costs in restricting mobility (progressively smaller additional economic damages).
  - Lockdowns display increasing marginal benefits in reducing infections.
- Policy implication:
  - To reduce infections by a given amount at the lowest short-run economic cost, "more stringent shorter-lived lockdowns could be preferable to mild prolonged measures."

### Organization and Scope
- Section 2 uses high-frequency proxies to assess economic impacts of lockdowns and voluntary social distancing.
- The analysis covers a large set of countries including advanced, emerging, and low-income economies.
- The study emphasizes the complementarity of mobility and job postings as high-frequency proxies and the importance of controlling for epidemiological dynamics in causal identification.

*Source: wpiea2020234-print-pdf — "2.1    Impact on Mobility"*

### 2.1    Impact on Mobility .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .

### 2.1    Impact on Mobility

### Data and Identification
- High-frequency proxies used:
  - Google Community Mobility Reports: daily attendance rates at various locations relative to pre-crisis levels (national and, for various countries, sub-national).
  - Indeed job postings: daily job postings in 22 countries, disaggregated by employment sector.
- Time horizon of analysis:
  - "first seven months of the pandemic"
  - Mobility and job postings computed as the daily average over "the first semester of 2020".
  - Real GDP growth for "the first semester of 2020" computed with respect to "the first semester of 2019".
- Econometric approach:
  - Local projections regressing the mobility index on the stringency of lockdowns and the number of COVID-19 infections.
  - Regressions control for the stage of the epidemic to separate lockdown effects from voluntary social distancing.
  - Sub-national estimations focus on regions less affected by COVID-19 in countries that adopted national lockdowns to strengthen identification; identification relies on national lockdowns often imposed in response to localized outbreaks and thus being largely exogenous to low-infection regions.

### Main Findings on Mobility and Economic Activity
- Both lockdowns and voluntary social distancing in response to rising COVID-19 infections had strong detrimental effects on economic activity as proxied by mobility and job postings.
- Lockdowns and voluntary social distancing played comparable roles in driving the drop in mobility across the full set of countries.
- Heterogeneity across country groups:
  - Advanced economies: the contribution of voluntary distancing was stronger, reflecting greater ability to work from home and to sustain temporary unemployment via personal savings and government benefits.
  - Low-income countries: lockdowns played a much stronger role, reflecting limited financial means to refrain from economic activities.
- Correlations:
  - The collapse in mobility over the first six months of 2020 correlates well with the decline in real GDP growth.
  - Job postings display a tight negative correlation with unemployment rates over the same period.

### Behavior around Lockdown Measures
- Lifting lockdowns is not sufficient for a strong sustained recovery if health risks remain; voluntary distancing can persist.
- Easing lockdowns tends to have a positive effect on mobility but:
  - "the impact is weaker than that of tightening lockdowns."
  - There is an asymmetric effect: tightening produces larger mobility reductions than loosening produces mobility increases.
- Policy implication: economies are likely to operate below potential as long as health risks persist; policymakers should be cautious about lifting lockdowns prematurely.

### Lockdowns, Infections, and Timing
- Using the same empirical framework, lockdowns can substantially reduce infections.
- Effects on COVID-19 cases tend to materialize a few weeks after lockdown introduction, consistent with the incubation period of the virus and testing times.
- Rapid intervention matters:
  - Lockdowns are particularly effective in curbing infections if introduced at an early stage of a country’s epidemic.
- Reassessment of the lives-versus-livelihoods trade-off:
  - Because infections themselves depress economic activity through voluntary distancing, lockdowns that reduce infections may facilitate faster economic recovery, potentially compensating short-term economic costs with higher future economic activity.

### Nonlinearities and Optimal Design
- Marginal effects:
  - More stringent lockdowns have decreasing marginal costs in restricting mobility (progressively smaller additional economic damages).
  - Lockdowns display increasing marginal benefits in reducing infections.
- Policy implication:
  - To reduce infections by a given amount at the lowest short-run economic cost, "more stringent shorter-lived lockdowns could be preferable to mild prolonged measures."

### Organization and Scope
- Section 2 uses high-frequency proxies to assess economic impacts of lockdowns and voluntary social distancing.
- The analysis covers a large set of countries including advanced, emerging, and low-income economies.
- The study emphasizes the complementarity of mobility and job postings as high-frequency proxies and the importance of controlling for epidemiological dynamics in causal identification.

*Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020234-print-pdf.pdf*

### 2.1    Impact on Mobility

### 2.1    Impact on Mobility

### Identification and methodology
- Empirical approach: local projection regressions (Jordà, 2005) estimated with country fixed effects and time fixed effects (see equations (1)–(4)).
- Key variables preserved from the source:
  - mobi,t+h: level of mobility for country i at time t+h (h is the horizon).
  - ln∆casesi,t−p: log of daily COVID-19 cases used to track the stage of the pandemic.
  - locki,t−p: lockdown stringency index (University of Oxford’s Coronavirus Government Response Tracker) reconstructed to exclude public information campaigns.
- Controls and estimation details:
  - Lags of dependent variable included (a week worth of lags).
  - Lags of cases and lockdown stringency included to control for persistence.
  - Country fixed effects and time fixed effects included.
  - Standard errors clustered at the country level.
- Sample coverage:
  - National analysis: 128 countries between early February and mid-July, 2020.
  - Subnational robustness: data for 422 subnational regions in 15 G20 countries that adopted national lockdowns; regions with the largest case counts and any region with more than 20 percent of the country’s total cases were excluded for identification.

### Effects of lockdowns on mobility (national and subnational results)
- A full lockdown (includes all measures used by governments during the pandemic) leads to a very significant decline in mobility:
  - National-level estimate: impact reaches about 25 percent after a week and mobility then resumes gradually.
  - Subnational estimates: negative effect corroborated; impact is modestly larger and more persistent than national estimates.
- Figure-based findings (as reported):
  - Panel 2a (national data) and panel 2b (subnational data) show similar shapes in mobility response to a full lockdown.
- Robustness checks reported in the source:
  - Results robust to controlling for COVID-19 deaths instead of cases.
  - Results robust to using Google mobility sub-indicators, controlling for testing, contact tracing, and public information campaigns.
  - Results robust to testing for cross-country heterogeneity by population density and indicators of governance and social capital.
  - A richer lag structure does not affect the results.

### Voluntary social distancing (responsiveness to rising infections)
- For a given lockdown stringency, rising COVID-19 cases induce voluntary reductions in mobility:
  - National estimate: a doubling of daily COVID-19 cases leads to a contraction in mobility by about 2 percent.
  - Subnational estimate: a doubling of daily COVID-19 cases leads to a contraction in mobility of 1.7 percent after 30 days.
- Interpretation and caveats noted in the source:
  - Voluntary social distancing captures behavioral responses to rising infections for a given level of lockdown stringency.
  - The analysis may underestimate true voluntary distancing because people also react to public health announcements, news, and government actions.
  - Reverse causality (higher mobility leading to faster spread) is acknowledged; dynamic estimation structure is intended to alleviate this endogeneity concern.

### Relative contributions by country group (advanced, emerging, low-income)
- Extended specification (equation (2)) allows impacts to differ across:
  - Advanced economies (AEi), emerging markets (EMi), and low-income countries (omitted category).
- Contributions to mobility decline during the first three months of each country’s epidemic are computed by multiplying average coefficients over a 30-day local projection horizon by average values of the corresponding variables for each country group.
- Cross-country averages (as described in Figure 4):
  - Both lockdowns and voluntary social distancing had a large impact on mobility, playing a roughly similar role across the full set of countries.
  - Voluntary social distancing was significantly stronger in advanced economies (facilitated by teleworking, contactless delivery, personal savings, social security).
  - Voluntary social distancing was quite limited in low-income countries, where the drop in mobility was mostly due to lockdowns.

### Implications for recovery and asymmetric effects of easing vs tightening
- Interaction analysis (equation (3)) shows that the effect of lockdowns depends on the stage of the pandemic:
  - γh0 (interaction term) reveals a weaker lockdown impact when national infections are relatively high.
  - Interacting lockdown stringency with global cases shows similar results: lockdowns have a weaker impact on mobility when global cases are high, indicating behavior responds to global health developments as well as national conditions.
  - The interaction difference is reported as statistically significant when interacting with both national and global cases.
- Asymmetric response to tightening versus easing (equation (4)):
  - Introduction of a full lockdown leads to a decline in mobility of about 26 percent one week after tightening.
  - Lifting restrictions boosts mobility by only about 18 percent over the same period; the difference is statistically significant.
- Policy implication emphasized:
  - Lifting lockdowns can only lead to a partial rebound in economic activity if health risks persist because voluntary social distancing will likely keep mobility compressed.
  - These findings caution against expecting a sharp economic rebound solely from easing lockdown measures if the virus continues to spread at a constant pace.

*Source: IMF Working Paper section "2.1    Impact on Mobility" from the provided PDF content.*

### 2.2    Impact on Job Postings

### 2.2    Impact on Job Postings

### Data and empirical specification
- Daily job postings data from Indeed for 22 countries covering January 1 to June 28, 2020.
- Panel regression re-estimates equation (1) substituting the level of mobility with the log of the number of job postings.
- Specification includes seven lags of the dependent and independent variables, and country and time fixed effects.
- Indeed sample includes primarily advanced economies.

### Aggregate impacts on job postings
- A full lockdown is associated with a decline in job postings of about 12 percent two weeks after the introduction of the lockdown.
- A doubling in daily COVID-19 cases leads to a 2 percent decline in job postings after 30 days.
- Both lockdowns and voluntary social distancing have negative and significant effects on job postings.
- Using estimated coefficients and average values during the first three months of each country’s epidemic, both lockdowns and voluntary social distancing contributed to the drop in job postings; the contribution of voluntary social distancing was relatively stronger.

### Sectoral dynamics around stay-at-home orders
- Job postings are analyzed by sector: contact-intensive sectors (food, hospitality, personal care) versus less-contact intensive sector (manufacturing).
- Stock of job postings normalized to 100 forty days before introduction of stay-at-home orders; time zero denotes introduction of stay-at-home orders.
- For each sector, the date of the first decline in job postings larger than one standard deviation is identified.
- Contact-intensive sectors started to decline between 1 and 2 weeks before the introduction of stay-at-home orders, indicating the importance of voluntary social distancing.
- Decline in manufacturing job postings broadly coincided with introduction of stay-at-home orders, suggesting lockdowns were the driving force in less-contact intensive sectors.

### Dynamics when lockdowns were eased
- Lifting restrictions led only to a marginal recovery in job postings.
- This corroborates findings based on mobility data, warning against expecting a sudden economic rebound from merely easing lockdown measures.

### Key takeaways
- Both lockdown policies and voluntary social distancing materially reduced job postings during the initial phase of the pandemic.
- Voluntary social distancing played a relatively stronger role in the Indeed sample of primarily advanced economies.
- Sectoral evidence shows earlier declines in contact-intensive sectors driven by voluntary avoidance, while less-contact sectors reacted mainly to official stay-at-home orders.
- Easing lockdowns produced only marginal improvements in job postings, indicating limited immediate rebound from reopening.

*Source: wpiea2020234-print-pdf - 2.2    Impact on Job Postings*

### Appendix A.  Data Sources and Country Coverage

### Appendix A.  Data Sources and Country Coverage

### Data sources used in the analysis
- Table A.1 lists the data sources used in the analysis.
- Indicators and sources:
  - Contact tracing — Oxford COVID-19 Government Response Tracker
  - COVID-19 cases — Oxford COVID-19 Government Response Tracker
  - Humidity — Air Quality Open Data Platform
  - Lockdown stringency index — Oxford COVID-19 Government Response Tracker
  - Mobility — Google Community Mobility Reports, Baidu for China
  - Stock of job postings — Indeed
  - Temperature — Air Quality Open Data Platform
  - Testing — Oxford COVID-19 Government Response Tracker

### Sample sizes and coverage summary
- For the analysis relying on high-frequency indicators:
  - 22 countries when job postings are used
  - 128 countries when mobility is used
- Subnational mobility sample:
  - 422 units for 15 G20 countries
- Analysis of infections:
  - 89 countries for which information on temperature, humidity, testing, and contact tracing is available
- Subnational infections sample:
  - 373 units for G20 15 countries

### Country coverage (Table A.2)
- Afghanistan — Mn, In
- Algeria — In
- Angola — Mn
- Argentina — Mn, Ms, In, Is
- Aruba — Mn
- Australia — Mn, Ms, In, Is, Jp
- Austria — Mn, In, Jp
- Bahrain — Mn, In
- Bangladesh — Mn, In
- Barbados — Mn
- Belarus — Mn
- Belgium — Mn, In, Jp
- Belize — Mn
- Benin — Mn
- Bolivia — Mn, In
- Bosnia and Herzegovina — Mn, In
- Botswana — Mn
- Brazil — Mn, Ms, In, Is, Jp
- Bulgaria — Mn, In
- Burkina Faso — Mn
- Cambodia — Mn
- Cameroon — Mn
- Canada — Mn, Ms, In, Is, Jp
- Chile — Mn, In
- China — Mn, Ms, In, Is
- Colombia — Mn, In
- Costa Rica — Mn, In
- Croatia — Mn, In
- Czech Republic — Mn, In
- Cˆote d’Ivoire — Mn, In
- Cyprus — In
- Denmark — Mn, In
- Dominican Republic — Mn
- Ecuador — Mn, In
- Egypt — Mn
- El Salvador — Mn, In
- Estonia — Mn, In
- Ethiopia — In
- Fiji — Mn
- Finland — Mn, In
- France — Mn, Ms, In, Is, Jp
- Gabon — Mn
- Georgia — Mn, In
- Germany — Mn, Ms, In, Is, Jp
- Ghana — Mn, In
- Greece — Mn, In
- Guatemala — Mn, In
- Guinea — In
- Haiti — Mn
- Honduras — Mn
- Hong Kong SAR — Mn, In, Jp
- Hungary — Mn, In
- Iceland — In
- India — Mn, Ms, In, Is
- Indonesia — Mn, Ms, In, Is
- Iran — In
- Iraq — Mn, In
- Ireland — Mn, In, Jp
- Israel — Mn, In
- Italy — Mn, Ms, In, Is, Jp
- Jamaica — Mn
- Japan — Mn, Ms, In, Is, Jp
- Jordan — Mn, In
- Kazakhstan — Mn, In
- Kenya — Mn
- Korea — Mn, In
- Kosovo — In
- Kuwait — Mn, In
- Kyrgyz Republic — Mn, In
- Lao P.D.R. — Mn, In
- Latvia — Mn
- Lebanon — Mn
- Libya — Mn
- Lithuania — Mn, In
- Luxembourg — Mn
- Macao SAR — In
- Madagascar — (not listed)
- Malawi — (not listed)
- Malaysia — Mn, In
- Mali — Mn, In
- Malta — (not listed)
- Mauritania — (not listed)
- Mauritius — Mn
- Mexico — Mn, Ms, In, Is, Jp
- Moldova — Mn
- Mongolia — Mn, In
- Montenegro — (not listed)
- Morocco — Mn
- Mozambique — Mn
- Myanmar — Mn, In
- Namibia — Mn
- Nepal — Mn, In
- Netherlands — Mn, In, Jp
- New Zealand — Mn, In, Jp
- Nicaragua — Mn
- Niger — Mn
- Nigeria — Mn
- Norway — Mn, In
- Oman — Mn
- Pakistan — Mn, In
- Panama — Mn
- Papua New Guinea — Mn
- Paraguay — Mn
- Peru — Mn, In
- Philippines — Mn, In
- Poland — Mn, In, Jp
- Portugal — Mn, In
- Puerto Rico — Mn
- Qatar — Mn
- Romania — Mn, In
- Russia — Mn, In
- Rwanda — Mn
- Saudi Arabia — Mn, Ms, In, Is
- Senegal — Mn
- Serbia — Mn, In
- Singapore — Mn, In, Jp
- Slovak Republic — Mn, In
- Slovenia — Mn
- South Africa — Mn, Ms, In, Is
- Spain — Mn, In, Jp
- Sri Lanka — Mn, In
- Sweden — Mn, In, Jp
- Switzerland — Mn, In, Jp
- Taiwan Province of China — Mn
- Tajikistan — Mn, In
- Tanzania — Mn
- Thailand — Mn, In
- Togo — Mn
- Trinidad and Tobago — Mn
- Turkey — Mn, In
- Uganda — Mn, In
- Ukraine — Mn, In
- United Arab Emirates — Mn, In, Jp
- United Kingdom — Mn, Ms, In, Is, Jp
- United States — Mn, In, Jp
- Uruguay — Mn
- Uzbekistan — In
- Venezuela — Mn
- Vietnam — Mn, In
- Yemen — Mn
- Zambia — Mn
- Zimbabwe — Mn

Notes: Mn = national-level regressions of mobility; Ms = subnational-level regressions of mobility; In = national-level regressions of infections; Is = subnational-level regressions of infections; Jp = job postings.

*Appendix A. Data Sources and Country Coverage — wpiea2020234-print-pdf*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020234-print-pdf.pdf_