## 2021. Using surprises in vaccines administered, we find that an unexpected increase in

## Source details

**Canonical URL:** [2021. Using surprises in vaccines administered, we find that an unexpected increase in](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021248-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021248-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021248-print-pdf.pdf.json)

---

### Key findings
- Using surprises in vaccines administered, an unexpected increase in vaccination per capita is associated with a significant increase in economic activity.
- A surprise increase in vaccinations by 10 percent of population is associated with an increase in daily per capita NO2 emissions of around 0.112, which is about a 0.28 standard deviation increase in NO2, or an increase of almost 30 percent relative to median.
- A 10-percentage points vaccine surprise is associated with an increase in mobility of 5 points.
- The marginal economic benefits of vaccines are larger when vaccination rates are higher (evidence of non-linear effects).
- Economic gains from vaccines are lower when strict containment measures are in place.
- Economic gains from vaccines are lower if the country is experiencing a severe outbreak during the vaccine rollout.
- Spillovers: an increase in COVID-19 cases in trading partners slows domestic economic activity, while an increase in COVID-19 vaccinations in main trading partners has a positive and statistically significant effect on domestic economic activity.
- Quantitative comparison: a 10-percentage points vaccine surprise has about the same impact on economic activity as going from a full lockdown (stringency index of 1) to containment measures equivalent to a stringency level of 0.5 (comparison noted as heuristic and caveated in the paper).

### Data and identification
- Sample and period:
  - Daily high-frequency proxies of economic activity for a sample over the period December 16, 2020 to June 20, 2021 (paper references "46 countries" in the introduction).
- High-frequency economic proxies:
  - NO2 coverage: 64 countries starting January 1, 2021.
  - CO coverage: 59 countries starting January 1, 2021.
  - Google mobility daily data available for 135 countries beginning February 15, 2020.
- COVID-19 data sources:
  - Vaccines data: Our World in Data, disaggregated by first and second shots, covering up to 202 countries starting in December 2020.
  - COVID-19 cases: CSSE at Johns Hopkins University, coverage begins January 22, 2020 for 208 countries.
  - Expected vaccinations time series: Airfinity’s supply-driven model.
- Vaccine surprise variable:
  - Constructed as the difference between actual percent of population fully vaccinated and the expected percent predicted by Airfinity’s supply-driven rollout model; used to improve causal identification and reduce endogeneity.
- Trade-weighted spillover variable:
  - Constructed using bilateral trade weights from 2019 Directional of Trade Statistics, scaled so weights sum to 1 for each country.

### Methodology
- Main estimation framework:
  - Country-time panel at daily frequency with country and time fixed effects.
- Baseline specification controls:
  - Lagged levels of COVID-19 cases, lagged high-frequency economic indicators (NO2, CO, mobility), stringency of containment measures, and country-specific time trends.
- Identification strategy:
  - Leverages high-frequency variation and the vaccine surprise measure to reduce endogeneity.
- Non-linearities and heterogeneity:
  - Tested by including squares of vaccination variables and interactions between vaccine surprises and quartiles of country-specific conditions (stringency, new cases).
  - Semi-parametric approach using quartile ("bin") interactions and alternative smooth transition specifications.
- Spillover analysis:
  - Trading-partner cases and vaccinations included via trade-weighted aggregates in the regression (equations (4) and (5)).
- Estimation and inference:
  - OLS with standard errors clustered at the country level; robustness checks include Driscoll-Kraay standard errors.

### Quantitative results and key statistics
- NO2 (main proxy):
  - Surprises in vaccines administered (per capita) coefficients imply: a 10 percent surprise → ~0.112 increase in daily per capita NO2 (≈0.28 standard deviation; ≈30 percent relative to median).
  - Table 3 reports Surprises in vaccines administered (per capita) 0.011232*** and 0.011086*** in key columns.
- Mobility:
  - 10-percentage points vaccine surprise → increase in mobility of 5 points (example comparison to U.S. mobility changes between March 2021 and May 2021).
  - Table 3 reports mobility-related coefficients (e.g., 0.058095 in one column).
- CO:
  - Positive effects observed but statistically weaker and typically significant only with lags.
- Lag structure:
  - Impact of vaccination increases with greater lags; CO effects become significant over time as protection accumulates and behavior adjusts.
- Spillover magnitudes:
  - One standard deviation increase in foreign COVID-19 cases → 0.10 standard deviation decrease in domestic NO2 emissions (lag ~21 days).
  - One standard deviation increase in foreign COVID-19 vaccines → 0.13 standard deviation increase in NO2 emissions (lag ~20 days).
  - US vaccinations per trade-weighted exposure: positive spillover; Figure 7 indicates economic gains materialize with a 20-day lag.
- Table-level sample & fit indicators (selected):
  - Table 3 observations: 6,226; 6,226; 5,879; 5,879; 6,208; 6,208; 5,861; 5,861; 6,215; 6,215; 5,868; 5,868.
  - Table 3 R-squared range: 0.326 to 0.443 across columns.

### Heterogeneity and non-linear effects
- Non-linearities:
  - Square terms and interaction specifications indicate economic benefits of vaccination rise as the share of vaccinated population increases; vaccine surprises have larger effects when baseline vaccination levels are higher.
  - Table 4 reports First dose squared 0.000131*; Second dose squared 0.000214***; interaction terms by quartile show larger coefficients in 2nd–4th quartiles (e.g., 2nd Quartile of second doses administered * Surprises 0.024510**, 3rd Quartile 0.025691**, 4th Quartile 0.019766*).
- Role of containment measures:
  - Interaction terms with containment-stringency quartiles are negative for higher quartiles and significantly different from zero for the 4th quartile (Table 5: 4th Quartile * Surprises -0.012473**).
  - Interpretation: stringent containment measures dampen the positive impact of vaccine surprises on NO2 emissions.
- Role of outbreak severity:
  - Interaction with new cases quartiles: negative and significant for higher quartiles (Table 5: 3rd Quartile * Surprises -0.029437**, 4th Quartile * Surprises -0.026988*).
  - Interpretation: high contemporaneous case counts reduce the local economic effect of vaccines due to voluntary distancing.
- Semi-parametric quartile interactions:
  - Confirm variation of vaccine effectiveness across levels of stringency and outbreak severity; similar qualitative patterns when using mobility as dependent variable though statistically weaker.

### Spillovers and policy implications
- Cross-border effects:
  - Foreign COVID-19 cases: negative spillover—trade-weighted foreign new cases per capita coefficient -37.319212** (Table 6, col 1).
  - Foreign vaccinations: positive spillover—foreign vaccines administered per capita (trade weighted) 0.068932** (Table 6, col 2).
  - US vaccinations (trade-weighted): 0.247082* (Table 6, col 3).
  - Timing: spillover effects materialize with lags of around 20–21 days and persist.
- Policy implications:
  - Speedy and equitable access to vaccines across nations is important for both partner health outcomes and domestic economic recovery.
  - Containment measures and outbreak severity must be considered when assessing economic dividends from vaccine rollouts.
  - Expectations matter: surprises relative to expected vaccine rollout drive economic responses, underscoring the role of credible and predictable supply and distribution.
  - Results support sharing vaccine oversupply in advanced economies to speed global recovery.

### Robustness and validation
- Robustness checks:
  - Results robust to winsorizing (1% winsorize coefficient 0.011192***), dropping late-starting vaccinators, dropping very early vaccinators, dropping one region or one country at a time (Annex Table 3 shows multiple columns with Surprises coefficients: range of reported estimates includes 0.008216** to 0.012232*** across specifications).
- Validation of proxies:
  - NO2 validated as a proxy for economic activity via correlations with Industrial Production (IP), Purchasing Managers Index (PMI), and OECD Composite Leading Indicator (CLI).
  - Table 1 key coefficients: IP 0.365665***, 0.173088***, 8.497571*; PMI 0.253454***; CLI 0.004974**.
  - Historical association between NO2 and GDP growth shown using OECD emissions data (1990–2018) in Table 2 (e.g., GDP growth coefficients 0.341**, 0.326*, 0.307*).

### Annex summaries (selected)
- Annex Tables 4 and 5:
  - Repeat non-linear regressions using CO and mobility as dependent variables; results generally less precisely estimated compared to NO2.
  - Annex Table 4 (CO): selected coefficients include First dose per capita -0.001680** (col 1) and First dose squared 0.000032** (col 1).
  - Annex Table 5 (Mobility): selected coefficients include First dose squared 0.005015* (col 1) and High second doses * Surprises 0.687222***.
- Annex Table A.1. Summary statistics (selected):
  - First dose per 100 inhabitants: Obs.: 23,257; Mean: 13.26; Std. Dev.: 17.73; Min: 0.00; Max: 116.15; N. of countries: 202.
  - Second dose per 100 inhabitants: Obs.: 15,257; Mean: 9.59; Std. Dev.: 14.27; Min: 0.00; Max: 114.86; N. of countries: 180.
  - Surprises in vaccines administered: Obs.: 18,845; Mean: -0.62; Std. Dev.: 5.70; Min: -24.26; Max: 54.50; N. of countries: 182.
  - NO2 per 1M inhabitants (daily): Obs.: 31,529; Mean: 1.37; Std. Dev.: 3.42; Min: 0.00; Max: 66.84; N. of countries: 62.
  - Mobility (daily): Obs.: 63,740; Mean: -21.66; Std. Dev.: 25.07; Min: -100.00; Max: 181.00; N. of countries: 135.

*Source: wpiea2021248-print-pdf - 2021. https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021248-print-pdf.pdf*

### 2021. Using surprises in vaccines administered, we find that an unexpected increase in

### wpiea2021248-print-pdf - 2021. Using surprises in vaccines administered, we find that an unexpected increase in

### Key findings
- Using surprises in vaccines administered, an unexpected increase in vaccination per capita is associated with a significant increase in economic activity.
- A surprise increase in vaccinations by 10 percent of population is associated with an increase in daily per capita NO2 emissions of around 0.112, which is about a 0.28 standard deviation increase in NO2, or an increase of almost 30 percent relative to median.
- A 10-percentage points vaccine surprise is associated with an increase in mobility of 5 points.
- The marginal economic benefits of vaccines are larger when vaccination rates are higher (evidence of non-linear effects).
- Economic gains from vaccines are lower when strict containment measures are in place.
- Economic gains from vaccines are lower if the country is experiencing a severe outbreak during the vaccine rollout.
- Spillovers: an increase in COVID-19 cases in trading partners slows domestic economic activity, while an increase in COVID-19 vaccinations in main trading partners has a positive and statistically significant effect on domestic economic activity.
- Quantitative comparison: a 10-percentage points vaccine surprise has about the same impact on economic activity as going from a full lockdown (stringency index of 1) to containment measures equivalent to a stringency level of 0.5 (comparison noted as heuristic and caveated in the paper).

### Data and identification
- Sample and period: daily high-frequency proxies of economic activity for a sample (paper references "46 countries" in the introduction) over the period December 16, 2020 to June 20, 2021.
- High-frequency economic proxies assembled: nitrogen dioxide (NO2) emissions, carbon monoxide (CO) emissions, and Google retail & recreation mobility indices.
  - NO2 coverage: 64 countries starting January 1, 2021.
  - CO coverage: 59 countries starting January 1, 2021.
  - Google mobility daily data available for 135 countries beginning February 15, 2020.
- COVID-19 data:
  - Vaccines data sourced from Our World in Data, disaggregated by first and second shots, covering up to 202 countries starting in December 2020.
  - COVID-19 cases from the CSSE at Johns Hopkins University, coverage begins January 22, 2020 for 208 countries.
  - Expected vaccinations time series obtained from Airfinity’s supply-driven model.
- Vaccine surprise variable: constructed as the difference between actual percent of population fully vaccinated and the expected percent predicted by Airfinity’s supply-driven rollout model; used to improve causal identification and reduce endogeneity.
- Trade-weighted spillover variable: constructed using bilateral trade weights from 2019 Directional of Trade Statistics, scaled so weights sum to 1 for each country.

### Methodology
- Main estimation framework: country-time panel at daily frequency with country and time fixed effects.
- Baseline specification includes lagged levels of COVID-19 cases, lagged high-frequency economic indicators (NO2, CO, mobility), stringency of containment measures, and country-specific time trends.
- Primary identification leverages high-frequency variation and the vaccine surprise measure to reduce endogeneity.
- Non-linearities and heterogeneity:
  - Tested by including squares of vaccination variables and interactions between vaccine surprises and quartiles of country-specific conditions (stringency, new cases).
  - Semi-parametric approach using quartile ("bin") interactions and alternative smooth transition specifications.
- Spillover analysis: trading-partner cases and vaccinations included via trade-weighted aggregates in the regression (equations (4) and (5)).
- Estimation: OLS with standard errors clustered at the country level; robustness checks include Driscoll-Kraay standard errors.

### Quantitative results and interpretation
- NO2 (main proxy): vaccine surprises strongly associated with higher NO2; coefficient magnitude implies the stated 10 percent surprise → ~0.112 increase in daily per capita NO2 (≈0.28 standard deviation; ≈30 percent relative to median).
- Mobility: 10-percentage points vaccine surprise → increase in mobility of 5 points (example comparison to U.S. mobility changes between March 2021 and May 2021).
- CO: positive effects observed but statistically weaker and typically significant only with lags.
- Lag structure: impact of vaccination increases with greater lags; CO effects become significant over time as protection accumulates and behavior adjusts.

### Heterogeneity and non-linear effects
- Non-linearities: square terms and interaction specifications indicate economic benefits of vaccination rise as the share of vaccinated population increases; vaccine surprises have larger effects when baseline vaccination levels are higher.
- Country-specific conditions:
  - Higher stringency of containment measures dampens the economic gains from vaccine surprises.
  - Higher contemporaneous severity of outbreaks reduces the local economic effect of vaccines due to continued voluntary social distancing.
- Semi-parametric quartile interactions confirm the variation of vaccine effectiveness across levels of stringency and outbreak severity.

### Spillovers and policy implications
- Cross-border effects:
  - Increase in COVID-19 cases in trading partners → slowdown in domestic economic activity through trade linkages.
  - Increase in vaccinations in trading partners → positive, statistically significant boost to domestic economic activity.
- Policy implications:
  - Speedy and equitable access to vaccines across nations is important not only for partner health outcomes but also for domestic economic recovery.
  - Containment measures and outbreak severity must be considered when assessing economic dividends from vaccine rollouts.
  - Expectations matter: surprises relative to expected vaccine rollout drive economic responses, underscoring the role of credible and predictable supply and distribution.

### Robustness and validation
- Results robust to:
  - Winsorizing data to limit outlier influence.
  - Dropping late-starting vaccinators (e.g., Colombia, Vietnam) or very early vaccinators (e.g., United States, United Kingdom).
  - Dropping one region or one country at a time.
- Validation of proxies:
  - NO2 is validated as a robust proxy for economic activity via correlations with industrial production (IP), Purchasing Managers Index (PMI), and OECD Composite Leading Indicator (CLI); historical association between NO2 and GDP growth shown using OECD emissions data (1990–2018).

*Source: wpiea2021248-print-pdf - 2021. https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021248-print-pdf.pdf*

### Annex Tables 4 and 5 repeat the non-linear regressions with changes in CO and mobility

### wpiea2021248-print-pdf - Annex Tables 4 and 5 repeat the non-linear regressions with changes in CO and mobility

### Annex note on alternative dependent variables
- Annex Tables 4 and 5 repeat the non-linear regressions using changes in CO and mobility as the dependent variable respectively.
- Results for these other indicators are generally less precisely estimated compared to NO2.
- Annex Table 6 uses CO (columns 1 and 2) and mobility (columns 3 and 4) as the dependent variable for similar interaction regressions; the results are generally weaker for these measures of economic activity.

### Role of containment measures and severity of outbreak
- Cubic-term test for diminishing returns to vaccination:
  - Cubic terms were added to test for diminishing returns as vaccination rates rise; no evidence for diminishing returns was found.
  - The paper notes this may reflect that not enough countries had reached high vaccination rates to approach herd immunity, and rising transmissibility of variants may have raised herd immunity thresholds.

- Interaction with stringency of containment measures (Table 5, equation 3):
  - An interaction between the vaccine surprise term and containment-stringency quartiles was estimated.
  - Interaction terms are negative for the higher quartiles and significantly different from zero for the 4th quartile.
  - Interpretation: an increase in vaccines leads to a smaller positive impact on NO2 emissions when accompanied by stringent containment measures—movement restrictions can prevent ramping up economic activity in response to higher vaccination rates.
  - Column 3 of Table 5 repeats the analysis with retail mobility as the dependent variable; results are similar to the NO2 regressions.

- Interaction with severity of outbreak (Table 5, equation 2):
  - Interaction between the vaccine surprise variable and number of new cases (7-day moving average) categorized into quartiles was estimated.
  - Interaction terms are negative and significant for the higher quartiles.
  - Interpretation: when cases are high, voluntary social distancing may limit the beneficial impact of vaccines on economic activity.
  - Results are qualitatively similar when using mobility as the dependent variable, though statistically weaker.

### Spillovers from foreign COVID-19 cases and vaccines to economic activity
- Effect of foreign COVID-19 cases on local economic activity (Table 6, column 1):
  - A one standard deviation increase in foreign COVID-19 cases leads to a 0.10 standard deviation decrease in domestic NO2 emissions.
  - The effect is lagged, becoming statistically significant after around 21 days.

- Effect of US vaccinations on foreign economic activity (Table 6, column 2):
  - A US spillover term (US vaccination rates multiplied by each country’s bilateral trade exposure to the US) shows positive spillover effects.
  - Figure 7 indicates economic gains materialize with a 20-day lag.

- Effect of trading-partner vaccinations more broadly (Table 6, column 3):
  - A one standard deviation increase in foreign COVID-19 vaccines leads to a 0.13 standard deviation increase in NO2 emissions.
  - The result is persistent and materializes with a lag; Figure 8 shows impacts persistently higher around the 20-day mark.

- Interpretation and policy relevance:
  - Foreign outbreaks dampen domestic economic activity via trade linkages; foreign vaccinations boost domestic activity.
  - These spillover results support the case for even distribution of vaccines globally and for sharing vaccine oversupply in advanced economies to speed global recovery.

### Key quantitative findings and magnitudes
- A surprise increase in vaccinations by 10 percent of population is associated with:
  - an increase in daily per capita NO2 emissions of about 0.3 standard deviations (an increase of almost 30 percent relative to its median).
  - an increase in mobility of 5 percentage points (for a 10-percentage-point vaccine surprise).
- CO emissions also increase in response to increased vaccine surprises, but with a lag.
- One standard deviation increase in foreign COVID-19 cases → 0.10 standard deviation decrease in domestic NO2 emissions (lag ~21 days).
- One standard deviation increase in foreign COVID-19 vaccines → 0.13 standard deviation increase in domestic NO2 emissions (lag ~20 days).
- US vaccination spillover effects materialize with a 20-day lag.

### Policy implications and conclusions
- Vaccines have a large and statistically significant effect on economic activity across high-frequency indicators (NO2 emissions, CO emissions, Google mobility indices).
- The economic benefits of vaccine rollouts are moderated by:
  - the stringency of containment measures—stringent measures reduce the immediate economic gains from vaccination;
  - the severity of local outbreaks—high case counts mute the economic effect of vaccinations due to voluntary distancing.
- Global coordination in vaccine distribution matters: vaccinating trading partners and sharing vaccine oversupply can boost domestic economic activity and speed global recovery.
- The findings reinforce the importance of vaccinating swiftly and broadly, both domestically and globally, to exit the COVID-19 health and economic crises.

*Source: wpiea2021248-print-pdf - Annex Tables 4 and 5 repeat the non-linear regressions with changes in CO and mobility*

### References

### References

### Cited studies and papers
- Agarwal, Ruchir, and Gita Gopinath. "A proposal to end the COVID-19 pandemic." IMF Staff Discussion Notes 2021, no. 004 (2021).
- Carvalho, V.M., Hansen, S., Ortiz, A., Garcia, J.R., Rodrigo, T., Rodriguez Mora, S. and Ruiz de Aguirre, P., 2020. Tracking the COVID-19 crisis with high-resolution transaction data.
- Coibion, O., Gorodnichenko, Y. and Weber, M., 2020. The cost of the covid-19 crisis: Lockdowns, macroeconomic expectations, and consumer spending (No. w27141). National Bureau of Economic Research.
- Dagan, N., Barda, N., Kepten, E., Miron, O., Perchik, S., Katz, M.A., Hernán, M.A., Lipsitch, M., Reis, B., and Balicer, R.D., 2021. “BNT162b2 mRNA Covid-19 Vaccine in a Nationwide Mass Vaccination Setting”. N Engl J Med, NEJMoa2101765.
- Deb Pragyan, Furceri Davide, Jimenez Daniel, Kothari Siddharth, Ostry Jonathan D., Nour Tawk, 2021. “Determinants of COVID-19 Vaccine Rollouts and Their Ef fects on Health Outcomes”, IMF Working Paper, XX/XXX.
- Deb Pragyan, Furceri Davide, Ostry Jonathan D., Tawk Nour, and Naihan Yang, 2021. “Ef fects of Fiscal Measures during COVID-19”, IMF Working Paper, XX/XXX.
- Deb Pragyan, Furceri Davide, Ostry Jonathan D., Nour Tawk, 2020a. “The Effects of Containment Measures on the COVID-19 Pandemic”, IMF Working Paper, 20/159.
- Deb Pragyan, Furceri Davide, Ostry Jonathan D., Nour Tawk, 2020b. “The Economic Ef fects of COVID-19 Containment Measures”, IMF Working Paper, 20/158.
- Deb Pragyan, Ganslmeier Michael, Furceri Davide, Ostry Jonathan D., Nour Tawk, 2021. “Vaccinate Early and Vaccinate Broadly: On the Health and Economic Ef fects of COVID-19 Vaccines” https://www.researchsquare.com/article/rs-525515/v1
- Fernández-Villaverde, Jesús, and Charles I. Jones. Macroeconomic outcomes and COVID-19: a progress report. No. w28004. National Bureau of Economic Research, 2020.
- International Monetary Fund. “The Great Lockdown: Dissecting the Economic Ef fects”. World Economic Outlook, Chapter 2. October 2020a.
- International Monetary Fund. “COVID-19 Lockdowns and Exits in Asia: Some Lessons”. Asia and Pacific Regional Economic Outlook, Chapter 2. October 2020b.
- International Monetary Fund, 2021. Leveraging Opportunities f rom COVID-19 Vaccines: Early Lessons from Asia. Asia and Pacific Regional Economic Outlook, Chapter 3. Washington, DC.
- Lin, J-T., and Michael Brendon McElroy. "Detection from space of a reduction in anthropogenic emissions of nitrogen oxides during the Chinese economic downturn." Atmospheric Chemistry and Physics 11, no. 15 (2011): 8171-8188.
- Polack, F.P., Thomas, S.J., Kitchin, N., Absalon, J., Gurtman, A., et al., 2020. “C4591001 Clinical Trial Group. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine”. N Engl J Med, vol. 383(27), pp. 2603-2615.
- Ramey, A. Valerie (2011). “Identifying Government Spending Shocks: It's all in the Timing,” The Quarterly Journal of Economics, vol. 126(1), pages 1-50.
- Sampi Bravo, James Robert Ezequiel, and Charl Jooste. "Nowcasting economic activity in times of COVID-19: An approximation from the Google Community Mobility Report." World Bank Policy Research Working Paper 9247 (2020).
- Sandmann, Frank G., Nicholas G. Davies, Anna Vassall, W. John Edmunds, Mark Jit, Fiona Yueqian Sun, C. Julian Villabona-Arenas et al. "The potential health and economic value of SARS-CoV-2 vaccination alongside physical distancing in the UK: a transmission model-based future scenario analysis and economic evaluation." The Lancet Infectious Diseases (2021).
- Voysey, Merryn, Sue Ann Costa Clemens, Shabir A. Madhi, Lily Y. Weckx, Pedro M. Folegatti, Parvinder K. Aley, Brian Angus et al. "Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK." The Lancet 397, no. 10269 (2021): 99-111.

### Figures (captions, sources, notes, and axes labels preserved)
- Figure 1. Vaccine surprises, people fully vaccinated (percent of population)
  - Source: Airfinity, Our World in Data, IMF staff calculations.
  - Axis: percent of population; series: EMDEs AEs; chart marks: -10, -5, 0, 5.
- Figure 2. Vaccine surprises, people fully vaccinated (percent of population)
  - Source: Airfinity, Our World in Data, IMF staff calculations.
  - Axes ticks and labels preserved: -20, -10, 0, 10, 20 (left); -3, -2, -1, 0, 1, 2, 3 (right); time ticks: 01jan2021, 01feb2021, 01mar2021, 01apr2021, 01may2021, 01jun2021; series annotated: USA, India, Israel (right axis).
- Figure 3. Effect of vaccinations on NO2 emissions per capita, at different lags
  - Note (preserved): Coefficient 훽훽 is reported for each lag ℓ (1-40), and based on ∆푌푌푖푖,푡푡 = 훼훼 + 휇휇푖푖 + 훾훾푡푡 + 훽훽훽훽훽훽푖푖,푡푡−푙푙 + 휃휃휃휃푖푖,푡푡−푙푙 + 휀휀푖푖,푡푡 for a sample of 44 countries using daily data from December 20, 2020–June 16, 2021. 푌푌푖푖,푡푡 denotes NO2 emissions per capita and 훽훽훽훽푖푖,푡푡−푙푙 is the surprises in vaccines administered (per capita). 휇휇푖푖 and 훾훾푡푡 are the country and time fixed effects. 휃휃 is a vector of control variables which includes the level of new cases, NO2 and CO emissions per capita, the stringency of containment measures index, and mobility indices at t-1. Lightly shaded bars denote 90 percent confidence bands, and dark-shared bars denote 95 percent confidence bands.
  - Y-axis values shown: 0, .01, .02, .03; Lag in days: 0 1 0 2 0 3 0 4 0 (as formatted in source).
- Figure 4. Effect of vaccinations on CO emissions per capita, at different lags
  - Note: (same specification text as Figure 3 with CO emissions per capita as dependent variable). Lightly shaded bars denote 90 percent confidence bands, and dark-shared bars denote 95 percent confidence bands.
  - Y-axis values shown: -0.001, 0, .001, .002, .003, .004; Lag in days: 0 1 0 2 0 3 0 4 0.
- Figure 5. Effect of vaccinations on mobility, at different lags
  - Note: (same specification text with changes in retail and recreation mobility as dependent variable). Lightly shaded bands denote 90 percent confidence bands, and dark-shared bands denote 95 percent confidence bands.
  - Y-axis values shown: -.5, 0, .5, 1, 1.5; Lag in days: 0 1 0 2 0 3 0 4 0.
- Figure 6. Effect of foreign new COVID-19 cases on economic activity
  - Note: Coefficient 훾훾 is reported for each lag ℓ (1-40), and based on ∆푌푌푖푖,푡푡 = 훼훼 + 휇휇푖푖 + 훾훾푡푡 + 훽훽훽훽훽훽푖푖,푡푡−푙푙 + 훾훾𝑇𝑇푟푟푟푟푟푟푟푟푟푟푟푟 𝑃𝑃푟푟푟푟 𝑃𝑃푟푟푃𝑃푟푟
푗푗,푡푡−푚푚 + 휃휃휃휃푖푖,푡푡−푙푙 + 휀휀푖푖,푡푡 for a sample of 43 countries using daily data from December 20, 2020–June 16, 2021. 푌푌푖푖,푡푡 denotes the number of new COVID-19 cases. 푇𝑇푟푟푟푟푟푟푟푟푟푟푟푟 𝑃𝑃푟푟푟푟 𝑃𝑃푟푟푃𝑃푟푟푖푖,푡푡 is a spillover term for COVID-19 cases in main trading partner countries. 훽훽훽훽푖푖,푡푡−푙푙 denotes the share of the individuals in the population which have received at least one vaccine shot. 휇휇푖푖 and 훾훾푡푡 are the country and time fixed effects. 휃휃 is a vector of control variables which includes the level of new cases, NO2 and CO emissions per capita, the stringency of containment measures index, and mobility indices at t-1. ℓ denotes the lags in the response of new COVID-19 cases. Lightly shaded bars denote 90 percent confidence bands, and dark-shared bars denote 95 percent confidence bands.
  - Y-axis values shown: -80, -60, -40, -20, 0, 20; Lag in days: 0 1 0 2 0 3 0 4 0.
- Figure 7. Effect of US COVID-19 vaccinations on economic activity
  - Note: (same model form as Figure 6 with US vaccinations spillover term). Lightly shaded bars denote 90 percent confidence bands, and dark-shared bars denote 95 percent confidence bands.
  - Y-axis values shown: 0, .1, .2, .3, .4, .5; Lag in days: 0 1 0 2 0 3 0 4 0.
- Figure 8. Effect of foreign COVID-19 vaccines on economic activity
  - Note: (same model form as Figure 6 with foreign vaccinations in main trading partner countries). Lightly shaded bars denote 90 percent confidence bands, and dark-shared bars denote 95 percent confidence bands.
  - Y-axis values shown: -.05, 0, .05, .1, .15; Lag in days: 0 1 0 2 0 3 0 4 0.

### Tables (titles, dependent variables, key reported sample statistics and model notes preserved)
- Table 1. High frequency indicators and monthly measures of economic activity (2019-2021)
  - Dependent variables across columns: NO2, CO, Mobility (three sets).
  - Key coefficients reported in table body: Industrial Production (IP) 0.365665***, 0.173088***, 8.497571*; Purchasing Managers' Index (PMI) 0.253454***, -0.082399, 2.700382; Composite Leading Indicator (CLI) 0.004974**, 0.003803, 1.040853***.
  - Constants and standard errors preserved as in table.
  - Observations: 1,178; 1,070; 1,017; 976; 956; 803; 944; 918; 592 (per respective columns).
  - R-squared: 0.262; 0.201; 0.213; 0.119; 0.061; 0.074; 0.574; 0.519; 0.623.
  - Country FE: Yes for all columns. Time FE: Yes for all columns.
  - No. of countries: 43; 39; 37; 38; 37; 31; 57; 54; 37.
  - Note: Standard error clustered at the country level. Robust standard errors in parentheses. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

- Table 2. NO2 emissions and economic activity—historical data (1990-2018)
  - Reported coefficients: GDP growth 0.341**, 0.326*, 0.307*; Manufacturing VA growth 0.130***, 0.134***, 0.135***; IP growth 0.203*, 0.201**, 0.206**.
  - Time trend and temperature effects reported (e.g., Time trend -0.001***, -0.001, 0.000, -0.002***, -0.001**, -0.001, -0.002**, 0.000, 0.001).
  - Observations: 929; 863; 828; 852; 789; 775; 623; 568; 566.
  - No. of countries: 36; 36; 36; 36; 36; 36; 30; 30; 30.
  - R-squared: 0.061; 0.082; 0.086; 0.051; 0.074; 0.076; 0.058; 0.100; 0.092.
  - Fixed Effects: Yes. Clustered SE: Yes.
  - Note: Standard error clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

- Table 3. Effect of vaccines on economic activity—alternative high-frequency economic indicators
  - Dependent variables across columns: NO2 (cols 1-4), CO (cols 5-8), Mobility (cols 9-12).
  - Key vaccine coefficients: First dose per capita 0.004539**, 0.002492*, 0.004790***, 0.000123, -0.000001, 0.000273, 0.070539, -0.003033, 0.058095 (as reported across relevant columns); Second dose per capita 0.010261***, 0.000626, 0.379216**.
  - Surprises in vaccines administered (per capita) 0.011232***, 0.011086***, 0.000538, 0.000530, 0.496498**, 0.494948** (in respective columns).
  - Observations: 6,226; 6,226; 5,879; 5,879; 6,208; 6,208; 5,861; 5,861; 6,215; 6,215; 5,868; 5,868.
  - R-squared: 0.326; 0.330; 0.332; 0.333; 0.281; 0.281; 0.278; 0.278; 0.432; 0.437; 0.443; 0.443.
  - Country FE: Yes for all columns. Time FE: Yes for all columns.
  - Health controls and country-time trends: Yes for all columns where indicated.
  - No. of countries: 46; 46; 44; 44; 46; 46; 44; 44; 46; 46; 44; 44.
  - Note: Table reports results for equation (1). The dependent variable is NO2 emissions per capita for columns 1-4, CO emissions per capita for columns 5-8 and change in retail and recreation mobility for columns 9-12. Regressions control for stringency of containment measures, other non-pharmaceutical interventions and health policy controls (1 lag), lags of mobility (1 lag), lagged new cases (1 lag), lagged NO2 and CO emissions (1 lag), country-specific time-trends, as well as country and time fixed-effects. Standard errors are clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

- Table 4. Non-linear effects of vaccines on economic activity
  - Dependent variable: NO2 (columns 1-6).
  - Reported terms include First dose per capita and First dose squared (e.g., First dose squared 0.000131*), Second dose per capita (0.008875***, 0.010812*), Second dose squared 0.000214***, Surprises in vaccines administered (per capita) -0.008556, -0.008666, interaction terms and quartile interactions (e.g., 2nd Quartile of second doses administered * Surprises in vaccines 0.024510**; 3rd Quartile 0.025691**; 4th Quartile 0.019766*).
  - Observations: 6,226; 6,226; 6,226; 5,879; 5,879; 5,879.
  - R-squared: 0.328; 0.330; 0.331; 0.335; 0.335; 0.333.
  - Country FE: Yes. Time FE: Yes.
  - No. of countries: 46; 46; 46; 44; 44; 44.
  - P-value F-test: 0.0578.
  - Note: Table reports results for the non-linear impact of vaccines on NO2. Columns 1 through 3 add additional terms for the square of the share of population that is vaccinated to the specification in equation (1). Columns 4 through 6 test use different specifications to test for interactions between the share of population fully vaccinated and the vaccine surprise variable. Regressions control for stringency of containment measures, other non-pharmaceutical interventions and health policy controls (1 lag), lags of mobility (1 lag), lagged new cases (1 lag), lagged NO2 and CO emissions (1 lag), country-specific time-trends, as well as country and time fixed-effects. Standard errors are clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

- Table 5. Effect of vaccines on economic activity—role of containment new cases
  - Dependent variable: NO2 (columns 1 and 2).
  - Surprises in vaccines administered (per capita): 0.014502*** (col 1) and 0.036355** (col 2).
  - Interaction terms with Containment Measures Quartiles: 2nd Quartile * Surprises -0.001964; 3rd Quartile * Surprises -0.005628; 4th Quartile * Surprises -0.012473**.
  - Interaction terms with New Cases Quartiles: 2nd Quartile * Surprises -0.020599; 3rd Quartile * Surprises -0.029437**; 4th Quartile * Surprises -0.026988*.
  - Observations: 5,859; 5,859.
  - R-squared: 0.334; 0.335.
  - Country FE: Yes. Time FE: Yes. Country-Time Trend: Yes.
  - No. of countries: 44; 44.
  - Notes: Table reports results for equation 3. Dependent variable is change in NO2 per capita in columns 1 and 2 and change in retail and recreational mobility in columns 3 and 4. Forecast error in vaccine rollout is interacted with stringency of containment measures (categorized into 4 quartiles) in column 1 and 3. Forecast error in vaccine rollout is interacted with level of new cases (moving average over 7 days and categorized into 4 quartiles) in column 2 and 4. Variables are lagged 1 days. All regressions control for stringency of containment measures and other non-pharmaceutical interventions, lagged mobility and NO2 per capita, country specific time trends, as well as country and time fixed effects. Standard errors are clustered at the country level. *** p<0.01, ** p<0.05, * p<0.1.

- Table 6. Effects of foreign new COVID-19 cases and vaccines on economic activity
  - Dependent variable: NO2 (per capita) across columns 1-3.
  - Surprises in vaccines administered (per capita): 0.009579*** (col 1), 0.010250*** (col 2), 0.011306*** (col 3).
  - Foreign new COVID-19 cases per capita (trade weighted): -37.319212** (col 1).
  - Foreign vaccines administered per capita (trade weighted): 0.068932** (col 2).
  - US Vaccinations per capita (trade weighted): 0.247082* (col 3).
  - Controls and lagged terms reported (e.g., NO2 emissions per capita (lag) -0.568909***, -0.550169***, -0.566578***).
  - Observations: 5,807; 4,606; 5,807.
  - R-squared: 0.334; 0.334; 0.333.
  - Country FE: Yes. Time FE: Yes.
  - Health controls and country-time trends: Yes.
  - No. of countries: 43; 42; 43.
  - Note: Table reports results for equation (4). Dependent variable is NO2 emissions per capita. A spillover term (foreign COVID-19 cases/foreign vaccinations/US vaccinations) (lag 30 days) is introduced to alternately capture the effects of trading partners’ COVID-19 new cases or vaccines on a country’s economic activity using bilateral trade weights (equation 3). Regressions control for stringency of containment measures, other non-pharmaceutical interventions and health policy controls (1 lag), lags of mobility (1 lag), lagged new cases (1 lag), lagged NO2 and CO emissions (1 lag), country-specific time-trends, as well as country and time fixed-effects. Standard errors are clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

*Annex material, figures, tables, and regression notes as presented in the References section of the source PDF.*

### Annex Table A.1. Summary Statistics

### Annex Table A.1. Summary Statistics

### Panel A: Summary Statistics for daily Time-Varying data
- First dose per 100 inhabitants
  - Obs.: 23,257
  - Mean: 13.26
  - Std. Dev.: 17.73
  - Min: 0.00
  - Max: 116.15
  - Source: OWID
  - Starting Date: 16-Dec-20
  - N. of countries: 202
- Second dose per 100 inhabitants
  - Obs.: 15,257
  - Mean: 9.59
  - Std. Dev.: 14.27
  - Min: 0.00
  - Max: 114.86
  - Source: OWID
  - Starting Date: 27-Dec-20
  - N. of countries: 180
- Surprises in vaccines administered
  - Obs.: 18,845
  - Mean: -0.62
  - Std. Dev.: 5.70
  - Min: -24.26
  - Max: 54.50
  - Source: Airfinity
  - Starting Date: 27-Dec-20
  - N. of countries: 182
- COVID-19 cases per 100 inhabitants
  - Obs.: 71,969
  - Mean: 0.01
  - Std. Dev.: 0.02
  - Min: 0.00
  - Max: 1.83
  - Source: JHU
  - Starting Date: 23-Jan-20
  - N. of countries: 205
- NO2 per 1M inhabitants
  - Obs.: 31,529
  - Mean: 1.37
  - Std. Dev.: 3.42
  - Min: 0.00
  - Max: 66.84
  - Source: AqiCN
  - Starting Date: 1-Jan-20
  - N. of countries: 62
- CO per 1M inhabitants
  - Obs.: 27,828
  - Mean: 0.54
  - Std. Dev.: 1.51
  - Min: 0.00
  - Max: 25.73
  - Source: AqiCN
  - Starting Date: 1-Jan-20
  - N. of countries: 57
- Mobility
  - Obs.: 63,740
  - Mean: -21.66
  - Std. Dev.: 25.07
  - Min: -100.00
  - Max: 181.00
  - Source: Google
  - Starting Date: 15-Feb-20
  - N. of countries: 135

### Panel B: Summary Statistics for monthly Time-Varying data
- IP
  - Obs.: 1,115
  - Mean: 3,058.62
  - Std. Dev.: 21,972.43
  - Min: 10.94
  - Max: 238,183.40
  - Source: Haver Analytics
  - Date: Jan-20
  - N. of countries: 63
- PMI
  - Obs.: 990
  - Mean: 50.29
  - Std. Dev.: 7.58
  - Min: 20.68
  - Max: 72.80
  - Source: Haver Analytics
  - Date: Jan-20
  - N. of countries: 55
- CLI
  - Obs.: 663
  - Mean: 97.86
  - Std. Dev.: 3.46
  - Min: 64.75
  - Max: 103.59
  - Source: OECD
  - Date: Jan-20
  - N. of countries: 39
- NO2 (monthly)
  - Obs.: 419
  - Mean: 1.31
  - Std. Dev.: 2.55
  - Min: 0.01
  - Max: 21.00
  - Source: AqiCN
  - Date: Dec-20
  - N. of countries: 61
- CO (monthly)
  - Obs.: 383
  - Mean: 0.48
  - Std. Dev.: 1.23
  - Min: 0.00
  - Max: 10.19
  - Source: AqiCN
  - Date: Dec-20
  - N. of countries: 57
- Mobility (monthly)
  - Obs.: 2,271
  - Mean: -20.15
  - Std. Dev.: 23.16
  - Min: -90.33
  - Max: 176.33
  - Source: Google
  - Date: 1-Feb
  - N. of countries: 135

### Note
- Table headings, variables, sources, and numeric values are reported exactly as in the source.

*Annex Table A.1. Summary Statistics — source PDF content.*

### Annex Table 2. Bivariate regressions of vaccine surprises with various variables

### Regressions: vaccine surprise on single variables (selected coefficients)
- New cases
  - Coefficient: 14.6677
  - Std. error: (16.078)
  - Observations: 5,875 (col. 1)
  - R-squared: 0.442
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Containment measures
  - Coefficient: -4.7802
  - Std. error: (5.014)
  - Observations: 5,848 (col. 2)
  - R-squared: 0.445
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Confirmed vaccine procurement (cumulative)
  - Coefficient: 42.6610
  - Std. error: (83.119)
  - Observations: 5,805 (col. 3)
  - R-squared: 0.459
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Potential vaccine procurement (cumulative)
  - Coefficient: 25.5848
  - Std. error: (80.705)
  - Observations: 5,805 (col. 4)
  - R-squared: 0.451
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Confirmed vaccine procurement (new)
  - Coefficient: -64.1929
  - Std. error: (39.836)
  - Observations: 5,805 (col. 5)
  - R-squared: 0.447
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Potential vaccine procurement (new)
  - Coefficient: -55.4293
  - Std. error: (39.495)
  - Observations: 5,805 (col. 6)
  - R-squared: 0.447
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- NO2 (lag 1)
  - Coefficient: 0.0199
  - Std. error: (0.040)
  - Observations: 5,866 (col. 7)
  - R-squared: 0.436
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- NO2 (lag 30)
  - Coefficient: 0.0693
  - Std. error: (0.068)
  - Observations: 5,790 (col. 8)
  - R-squared: 0.465
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 43
- CO (lag 1)
  - Coefficient: -0.0431
  - Std. error: (0.048)
  - Observations: 5,847 (col. 9)
  - R-squared: 0.436
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- CO (lag 30)
  - Coefficient: -0.0293
  - Std. error: (0.031)
  - Observations: 5,700 (col. 10)
  - R-squared: 0.467
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Mobility (lag 1)
  - Coefficient: 0.0000
  - Std. error: (0.001)
  - Observations: 5,875 (col. 11)
  - R-squared: 0.438
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44
- Mobility (lag 30)
  - Coefficient: 0.0036
  - Std. error: (0.003)
  - Observations: 5,875 (col. 12)
  - R-squared: 0.438
  - Country FE: Yes
  - Time FE: Yes
  - No. of countries: 44

Note: Table reports results from regressing the vaccine surprise variable (difference between per cent of population that is fully vaccinated and the expected share that will be fully vaccinated taken from Airfinity) on various variables. All regressions include country and time fixed effects. Standard errors are clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

### Annex Table 3. Robustness checks—Baseline using NO2 emissions

### Main coefficients and robustness panels
- Dependent variable: NO2 emissions per capita (lagged dependent variable and controls included)
- Surprises in vaccines administered (per capita)
  - Col (1) drop late start: 0.011918***
    - Std. error: (0.003)
  - Col (2) 1% winsorize: 0.011192***
    - Std. error: (0.002)
  - Col (3) drop early after March 1: 0.008219***
    - Std. error: (0.003)
  - Col (4) without 5% before Feb 1: 0.010750***
    - Std. error: (0.003)
  - Col (5) APD: 0.008216**
    - Std. error: (0.003)
  - Col (6) EUR: 0.010365***
    - Std. error: (0.002)
  - Col (7) MCD: 0.012232***
    - Std. error: (0.003)
  - Col (8) WHD AFR: 0.011289***
    - Std. error: (0.002)
- COVID-19 cases per capita (lag)
  - Coefficients vary by column; examples:
    - Col (1): 0.004288 (0.019)
    - Col (3): -0.019220 (0.025)
    - Col (5): 0.020262 (0.019)
- NO2 emissions per capita (lag)
  - Col (1): -0.564944*** (0.033)
  - Col (2): -0.565331*** (0.033)
  - Col (3): -0.582338*** (0.035)
  - Col (4): -0.560856*** (0.037)
  - Col (5): -0.606884*** (0.035)
  - Col (6): -0.548998*** (0.034)
  - Col (7): -0.567048*** (0.033)
  - Col (8): -0.565005*** (0.033)
- CO emissions per capita (lag)
  - Coefficients not statistically significant in reported columns; e.g., Col (1): 0.018059 (0.046)
- Containment measures (lag)
  - Col (1): -0.207061 (0.123)
  - Col (2): -0.229113* (0.129)
  - Col (5): -0.354375** (0.155)
- Mobility (lag)
  - Mostly near zero; Col (6): 0.001335** (0.001)
- Constants (selected)
  - Col (1): 5.976847 (5.387)
  - Col (3): -62.995520*** (6.169)
  - Col (6): 11.078580** (5.231)
- Observations and fit
  - Observations: vary by column; Col (1): 5,879; Col (2): 5,624; Col (3): 5,158; Col (4): 4,878; Col (5): 2,629; Col (6): 5,422; Col (7): 4,823; Col (8): 5,759
  - R-squared: Col (1): 0.332; Col (2): 0.333; Col (3): 0.353; Col (4): 0.341; Col (5): 0.344; Col (6): 0.340; Col (7): 0.343; Col (8): 0.333
  - Country FE: Yes; Time FE: Yes; Health controls and country-time trends: Yes
  - No. of countries: Col (1): 44; Col (2): 41; Col (3): 39; Col (4): 34; Col (5): 23; Col (6): 40; Col (7): 36; Col (8): 43

Note: Table reports results for equation (1). Standard errors clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

### Annex Table 4. Non-linear effects of vaccines on economic activity using CO

### Selected coefficient estimates (dependent variable: CO)
- First dose per capita
  - Col (1): -0.001680** (0.001)
  - Col (2): -0.001771** (0.001)
  - Col (3): 0.000410 (0.001)
- First dose squared
  - Col (1): 0.000032** (0.000)
  - Col (2): 0.000035* (0.000)
- Second dose per capita
  - Col (2): -0.000303 (0.001)
  - Col (3): -0.001211 (0.002)
  - Col (4): 0.001004 (0.002)
- Second dose squared
  - Col (4): 0.000039* (0.000)
- Surprises in vaccines administered (per capita)
  - Col (3): -0.001468 (0.001)
  - Col (4): -0.007415 (0.007)
- Second dose per capita * Surprises in vaccines administered
  - Col (4): 0.000032 (0.000)
- Interactions with second dose quartiles * Surprises in vaccines (Columns 5–6 / alternative specifications)
  - 2nd Quartile * Surprises: 0.008883 (0.006)
  - 3rd Quartile * Surprises: 0.009174 (0.007)
  - 4th Quartile * Surprises: 0.008483 (0.007)
  - Low second doses * Surprises: -0.002158 (0.004)
  - High second doses * Surprises: 0.001272 (0.001)
- Observations: Col (1–3): 6,208; Col (4–6): 5,861
- R-squared: Col (1): 0.282; Col (2): 0.282; Col (3): 0.281; Col (4): 0.278; Col (5): 0.280; Col (6): 0.278
- Country FE: Yes; Time FE: Yes
- No. of countries: Col (1–3): 46; Col (4–6): 44
- P-value F-test: 0.513

Note: Regressions include controls as described in the source. Standard errors clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

### Annex Table 5. Non-linear effects of vaccines on economic activity using mobility

### Selected coefficient estimates (dependent variable: Mobility)
- First dose per capita
  - Col (1): -0.208008 (0.138)
  - Col (2): -0.111641 (0.164)
  - Col (3): 0.068691 (0.090)
- First dose squared
  - Col (1): 0.005015* (0.003)
  - Col (2): 0.002153 (0.003)
- Second dose per capita
  - Col (2): 0.322693** (0.160)
  - Col (3): 0.060115 (0.270)
  - Col (4): 0.128190 (0.208)
- Second dose squared
  - Col (3): 0.006777 (0.004)
- Surprises in vaccines administered (per capita)
  - Col (3): 0.084600 (0.574)
  - Col (4): 0.498283 (0.396)
- Second dose per capita * Surprises in vaccines administered
  - Col (4): 0.008895 (0.008)
- Interactions with second dose quartiles * Surprises in vaccines
  - 2nd Quartile * Surprises: -0.298942 (0.505)
  - 3rd Quartile * Surprises: -0.035946 (0.454)
  - 4th Quartile * Surprises: 0.127307 (0.382)
  - Low second doses * Surprises: -0.204844 (0.590)
  - High second doses * Surprises: 0.687222*** (0.175)
- Observations: Col (1–3): 6,215; Col (4–6): 5,868
- R-squared: Col (1): 0.435; Col (2): 0.438; Col (3): 0.439; Col (4): 0.444; Col (5): 0.444; Col (6): 0.444
- Country FE: Yes; Time FE: Yes
- No. of countries: Col (1–3): 46; Col (4–6): 44
- P-value F-test: 0.136

Note: Regressions control for listed policy and health variables. Standard errors clustered at the country level. ***, **, and * represent statistical significant at 1, 5 and 10 percent respectively.

### Table 6. Effect of vaccines—role of containment and new cases using CO and Mobility

### Interaction results (selected)
- Dependent variables: CO (cols 1–2) and Retail mobility (cols 3–4)
- Surprises in vaccines administered (per capita)
  - CO Col (1): 0.000330 (0.001)
  - CO Col (2): 0.001720 (0.002)
  - Retail Col (3): 0.717312*** (0.147)
  - Retail Col (4): 0.995808*** (0.367)
- Interaction with Containment Measures quartiles * Surprises (1st quartile omitted)
  - 2nd Quartile * Surprises (CO): 0.000709 (0.000)
  - 3rd Quartile * Surprises (CO): 0.000208 (0.001)
  - 4th Quartile * Surprises (CO): -0.001011 (0.001)
  - 2nd Quartile * Surprises (Retail): -0.191644 (0.131)
  - 3rd Quartile * Surprises (Retail): -0.431621** (0.179)
  - 4th Quartile * Surprises (Retail): -0.384407* (0.227)
- Interaction with New Cases quartiles * Surprises (1st quartile omitted)
  - 2nd Quartile * Surprises (CO): -0.001192 (0.002)
  - 3rd Quartile * Surprises (CO): -0.001486 (0.002)
  - 4th Quartile * Surprises (CO): -0.001395 (0.002)
  - 2nd Quartile * Surprises (Retail): -0.402936 (0.354)
  - 3rd Quartile * Surprises (Retail): -0.513714 (0.382)
  - 4th Quartile * Surprises (Retail): -0.440691 (0.350)
- Observations: 5,861 (cols 1–2); 5,839 (cols 3–4)
- R-squared: CO cols: 0.278; Retail cols: 0.449 (col 3), 0.452 (col 4)
- Country FE: Yes; Time FE: Yes; Country-Time Trend: Yes
- No. of countries: 44 (all columns)

Notes: Table reports results for equation 3. Dependent variable is change in CO per capita (cols 1–2) and change in retail and recreational mobility (cols 3–4). Forecast error in vaccine rollout is interacted with stringency quartiles and new cases quartiles as specified. Variables are lagged 1 day. All regressions control for listed variables. Standard errors clustered at the country level. *** p<0.01, ** p<0.05, * p<0.1

*Annex Tables 2–6 — selected regression coefficients and statistics — source PDF content.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021248-print-pdf.pdf_