## wpiea2021181-print-pdf

## Source details

**Canonical URL:** [wpiea2021181-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021181-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021181-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021181-print-pdf.pdf.json)

---

### Poverty and inequality: headcount and poverty gap
- Using headcount ratio at $3.20 per day (instead of $1.90 a day), pandemics lead to an increase of 1.1 percentage points in the poverty rate five years later.
- Poverty gap (WDI measure) increases following a pandemic for the two measures corresponding to headcount ratios of $1.90/day and $3.20/day, with the effect statistically significant for both in the second year of the crisis.
- The increase in poverty gap persists over the medium term.
- Pandemics both increase the share of population in poverty and intensify the hardships of those in poverty.
- Baseline regressions use net Gini indices from SWIID (Solt, 2020); inequality continues to increase in the time horizon considered with the effect becoming statistically significant in the medium term.
- After five years, inequality increases by 1.7 percent (net Gini).
- Market inequality (pre-tax and pre-transfer) rises by about 1.0 percent after five years (statistically significant in the medium term).
- Impulse responses using WDI income shares show:
  - Share of income for bottom 10 and 20 percent falls after pandemics (statistically significant in the near term for bottom 10 and 20 percent).
  - Share of income for top 10 and 20 percent increases after pandemics (statistically significant in the near term and in the second year for top 10 and 20 percent).
  - These effects persist over the medium term.

### Fiscal Policy: fiscal impulse and macroeconomic outcomes
- Fiscal impulse defined as the negative of the annual change in structural primary balance (share of potential GDP); structural primary balance data from IMF WEO.
- Countries split into “high” versus “low” fiscal impulse relative to the median in the pandemic year; pandemic shock interacted with high/low dummy (equation II).
- Output effects after three years following the pandemic shock:
  - Low fiscal impulse countries: output declines by 3.4 percent after three years.
  - High fiscal impulse countries: output declines by 1.5 percent after three years.
  - Baseline case: 2.2 percent drop after three years.
- Negative output effect statistically significant for four of six time-horizons for low-support countries; not statistically significant for any time horizon for high-support countries.
- Output returns close to pre-pandemic levels by the fifth year for high-support countries; remains below pre-pandemic levels for low-support countries.
- Unemployment:
  - Rises higher and does not return to pre-pandemic levels within five years for low fiscal support countries.
  - Smaller effects and return to pre-pandemic levels for high fiscal support countries.
  - Effects statistically significant for low-support group, not for high-support group.
- Poverty rate:
  - Low fiscal support countries: increases by 0.6 percentage points after the first year (statistically significant).
  - High fiscal support countries: increases by 0.2 percentage points after the first year (statistically insignificant).
  - For high fiscal support countries, poverty rate falls below pre-pandemic levels over time.
- Inequality (Gini):
  - Low fiscal support countries: increases by 1.5 percent after five years (statistically significant).
  - High fiscal support countries: increases by 0.7 percent after five years (statistically insignificant).
- Robustness checks: relative magnitudes for output, unemployment, poverty, and inequality for low versus high fiscal support using structural primary balance hold across a range of alternative specifications.
- Related evidence:
  - Ma et al. (2020): countries with higher government expenditures, especially on health care, have a higher output growth bounce-back.
  - Robustness checks performed dividing fiscal indicator by previous year’s GDP (Ma et al. approach).
  - Furceri et al. (2021b): extent of fiscal consolidation following past pandemics determines increase in inequality; extreme austerity associated with Gini increase three times as large as more supportive fiscal policy.

### Informality
- Informality data from Medina and Schneider (2020); countries split into “high” and “low” informality using values for the year before the pandemic (equation II).
- Output:
  - High informality countries: output falls by 2.6 percent after three years (lowest point).
  - Low informality countries: output falls by 2.2 percent after three years.
  - Effects statistically significant for both groups.
  - Output decline is higher for low informal group in the fourth and fifth year in some specifications.
- Unemployment:
  - Rises more for countries with high informality; effect statistically significant for high informality and not for low informality.
- Poverty:
  - High informality countries: poverty rate increases by 1.3 percentage points after the first year (near-term statistically significant).
  - Low informality countries: poverty rate increases by 0.7 percentage points after the first year.
  - Five years after the pandemic:
    - High informality countries: poverty rate remains 1 percentage points above pre-pandemic levels.
    - Low informality countries: poverty rate remains 0.4 percentage point above pre-pandemic levels.
- Inequality:
  - High informality countries: 0.5 percentage point increase after two years.
  - Low informality countries: 0.2 percentage point increase after two years.
  - After five years, rise in inequality slightly higher for high informality; effect statistically significant for both groups.
- Robustness: results for unemployment, poverty, and inequality generally robust across alternative specifications; some robustness cases show relative magnitudes for output by informality may not hold.

### Social Expenditure (family benefits and other components)
- OECD social expenditure data used; baseline uses family benefits component (public spending on family benefits).
- Decadal average of social expenditures used to determine “high” and “low” family benefits; regressions limited to around 30 countries.
- During Covid-19, increased social protection spending averaged 0.6 percent of GDP in the first three quarters of 2020 and estimated to have reduced the rise of extreme global poverty by about 10 million people (IMF 2021b).
- Findings for family benefits:
  - Countries with lower family benefits associated with relatively higher unemployment rise, higher poverty, and higher inequality (robust across specifications).
  - Output drops similar for low and high family benefits groups; robustness checks show relative magnitude contingent on specification.
  - Unemployment rise stronger for low family benefits group, though effect not statistically significant for either group.
  - Poverty rate rise: 0.15 percentage points after two years higher for countries with low family benefits; effects statistically significant for high group and not for low group.
  - Inequality increase after five years:
    - Low family benefits: 2.2 percent increase (statistically significant).
    - High family benefits: 1.5 percent increase (statistically significant).
- Figure 9 (summary for three-year horizon) using total social expenditures and components:
  - Negative output effect after three years is higher for countries with lower spending across total and components (with exception of family benefits).
  - Unemployment rate higher for countries with lower benefits across items (not statistically significant for either group).
  - Rise in poverty rate higher for countries with lower benefits compared to higher benefits.
  - Rise in inequality is higher for countries with lower family benefits and lower health spending, but lower for countries with lower total social expenditure and lower active labor market policies.

### Health Expenditure per Capita
- Health expenditure per capita from WDI: current health expenditure per capita in current international dollars (PPP).
- Output:
  - Low health expenditure countries: output falling by 3.8 percent after three years (statistically significant).
  - High health expenditure countries: output falling by 3.4 percent after three years (statistically significant).
  - Negative output-effect stronger for high expenditure group in years four and five.
- Unemployment:
  - Low health expenditure group: rise of 1.7 percentage points after four years (effects statistically significant).
  - High health expenditure group: rise of 0.8 percentage points after four years.
- Poverty:
  - Low health expenditure group: poverty rate increases by 1.0 percentage points within the first year and remains 0.7 percentage points above pre-pandemic levels after five years (near-term effect statistically significant).
  - High health expenditure group: poverty rate increases by 0.4 percentage points after the first year and falls slightly below pre-pandemic levels after five years (effects on poverty not statistically significant for high group).
- Inequality:
  - Inequality increases higher for low health expenditure countries (details and exact magnitudes for inequality continuation referenced in source).
- Comparative point preserved exactly: 1.2 percent after five years compared to 0.6 percent increase for the high expenditure group.

### Key empirical findings
- Using a sample of 55 countries over 1990-2019 and employing the local projection methodology, past pandemics had lasting damaging scarring effects with detrimental impacts on output, unemployment, poverty, and inequality.
- A relatively stronger fiscal response helped to mitigate the negative output effects.
- A relatively stronger fiscal response and better initial conditions in informality, family benefits, and health expenditure per capita helped to alleviate negative effects on unemployment, poverty, and inequality.
- Results on output are not robust across all specifications and further work is needed to understand the channels better.
- Application of empirical estimates from past pandemics on inequality suggests significant scarring effects of Covid-19; Emmerling et al. (2021) project an increase in poverty of about 75 million people worldwide.
- Certain Covid-19 specific mechanisms amplify adverse outcomes in weaker initial-condition countries:
  - Social distancing led to reliance on online schooling; UN (2020) reports the largest disruption of education systems in history, affecting nearly 1.6 billion learners in more than 190 countries.
  - Higher informality is associated with lower internet access and lower account ownership at financial institutions or with mobile-money-service providers, exacerbating loss of human capital, poverty, and inequality.
  - Countries with lower social expenditure have less internet and financial access, further worsening poverty and inequality outcomes.

### Robustness checks (alternative formulations employed) and main robustness findings
- Robustness specifications include:
  - (i) dropping all control variables, except output;
  - (ii) dropping all control variables, including output;
  - (iii) including all control variables;
  - (iv) including all control variables except credit;
  - (v) including additional control variables (two lags of public-to-GDP ratio; two lags of ICRG indicators);
  - (vi) including sample from 2000;
  - (vii) adding four lags, instead of two lags as specified in the baseline and augmented regressions;
  - (viii) splitting the sample using different percentiles (median; bottom 25th percentile versus rest; top 75th percentile versus rest);
  - (ix) using contemporaneous values of control variables (in addition to two lags);
  - (x) using contemporaneous values of informality and health expenditure per capita (alternative to using lagged values one year before the pandemic);
  - (xi) using alternative indicators for fiscal (divide negative change in structural primary balance by previous year nominal GDP; use changes in structural overall fiscal balances; cyclically adjusted overall and primary balances; overall and primary fiscal balances);
  - (xii) using alternative definition of health: lagged current health expenditure as a share of GDP and general government health expenditure (as a share of GDP).
- Main robustness findings:
  - Output decline in the baseline is found to be more pronounced in some cases.
  - The unemployment results come out statistically significant in some cases, compared to the baseline where it narrowly misses statistical significance.
  - Relative magnitudes highlighted using fiscal policy are found to be robust using the main/baseline definition of fiscal impulse in the paper; findings on the role of discretionary fiscal policies remain intact.
  - Results for structural features (informality, family benefits, health) are generally robust for unemployment, poverty, and inequality. Exceptions: relative magnitudes for output using informality, family benefits, and health expenditure per capita do not always hold; impact of family benefits on output is inconclusive in robustness checks.

### Policy implications and recommendations
- Adequate fiscal support, higher health spending, and targeted family benefits should be part of the policy package to limit long-lasting scarring effects of pandemics, especially for vulnerable groups and to mitigate persistent loss of human capital from Covid-19.
- Higher policy support and complementary policies might be required in countries with high informality to reduce larger negative impacts on unemployment, poverty, and inequality.
- Delechat et al. (2020) policy recommendations to address informality include:
  - (i) improve access to and quality of education;
  - (ii) avoid tax system design that inadvertently increases incentives for individuals and firms to remain in the informal sector;
  - (iii) enhance financial inclusion by promoting access to formal financial services;
  - (iv) pursue a range of structural policies (e.g., in labor market regulations, competition policies, digitalization) to increase incentives and lower the cost of formalization.

### Data, sample, and estimation
- Sample: 55 countries over the period 1990-2019.
- Pandemic episodes: taken from Ma et al. (2020) for the last five pandemics.
- Key data sources:
  - Outcome variables: World Development Indicators (WDI).
  - Inequality (Gini indices): Standardized World Income Inequality Database (SWIID).
  - Pandemic episodes: Ma et al. (2020).
  - Fiscal impulse: IMF World Economic Outlook (WEO) — defined as the negative of change in structural primary balance (percentage points of potential GDP).
  - Informal sector: Medina and Schneider (2020) — percent of GDP.
  - Social expenditure and its components: OECD.
  - Current health expenditure per capita (PPP current intl. $): WDI.
  - Banking crisis indicator: Laeven and Valencia (2013)/(2020).
  - Institutional quality: PRS Group, International Country Risk Guide (ICRG).
- Estimation approach:
  - Impulse response functions estimated by local projections.
  - Equations include: a pandemic dummy (and two lags), two lags of the dependent variable, two lags of output, and country- and time-fixed effects.
  - Output and unemployment equations additionally control for two lags of: log of income per capita, trade-to-GDP ratio, private credit-to-GDP ratio, and banking crisis (Laeven and Valencia).
  - Standard errors: clustered at the country level.
  - Confidence bands reported at 90 percent.
  - Timing convention: x-axis denotes years before and after pandemic where t=0 is the year of the pandemic.

### Summary quantitative points (preserved exactly as in source)
- 1.2 percent after five years compared to 0.6 percent increase for the high expenditure group.
- Sample: 55 countries over 1990-2019.
- Emmerling et al. (2021) project an increase in poverty of about 75 million people worldwide.
- UN (2020) reports nearly 1.6 billion learners affected in more than 190 countries.

*Source: wpiea2021181-print-pdf - https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021181-print-pdf.pdf*

### 0.9 percentage points above pre-pandemic levels (bottom-left chart of Figure 3). As

### wpiea2021181-print-pdf - 0.9 percentage points above pre-pandemic levels (bottom-left chart of Figure 3). As

### Poverty and inequality: headcount and poverty gap
- Using headcount ratio at $3.20 per day (instead of $1.90 a day), pandemics lead to an increase of 1.1 percentage points in the poverty rate five years later.
- Poverty gap (WDI measure) increases following a pandemic for the two measures corresponding to headcount ratios of $1.90/day and $3.20/day, with the effect statistically significant for both in the second year of the crisis.
- The increase in poverty gap persists over the medium term.
- Pandemics both increase the share of population in poverty and intensify the hardships of those in poverty.
- Baseline regressions use net Gini indices from SWIID (Solt, 2020); inequality continues to increase in the time horizon considered with the effect becoming statistically significant in the medium term.
- After five years, inequality increases by 1.7 percent (net Gini).
- Market inequality (pre-tax and pre-transfer) rises by about 1.0 percent after five years (statistically significant in the medium term).
- Impulse responses using WDI income shares show:
  - Share of income for bottom 10 and 20 percent falls after pandemics (statistically significant in the near term for bottom 10 and 20 percent).
  - Share of income for top 10 and 20 percent increases after pandemics (statistically significant in the near term and in the second year for top 10 and 20 percent).
  - These effects persist over the medium term.

### Fiscal Policy: fiscal impulse and macroeconomic outcomes
- Fiscal impulse defined as the negative of the annual change in structural primary balance (share of potential GDP); structural primary balance data from IMF WEO.
- Countries split into “high” versus “low” fiscal impulse relative to the median in the pandemic year; pandemic shock interacted with high/low dummy (equation II).
- Output effects after three years following the pandemic shock:
  - Low fiscal impulse countries: output declines by 3.4 percent after three years.
  - High fiscal impulse countries: output declines by 1.5 percent after three years.
  - Baseline case: 2.2 percent drop after three years.
- Negative output effect statistically significant for four of six time-horizons for low-support countries; not statistically significant for any time horizon for high-support countries.
- Output returns close to pre-pandemic levels by the fifth year for high-support countries; remains below pre-pandemic levels for low-support countries.
- Unemployment:
  - Rises higher and does not return to pre-pandemic levels within five years for low fiscal support countries.
  - Smaller effects and return to pre-pandemic levels for high fiscal support countries.
  - Effects statistically significant for low-support group, not for high-support group.
- Poverty rate:
  - Low fiscal support countries: increases by 0.6 percentage points after the first year (statistically significant).
  - High fiscal support countries: increases by 0.2 percentage points after the first year (statistically insignificant).
  - For high fiscal support countries, poverty rate falls below pre-pandemic levels over time.
- Inequality (Gini):
  - Low fiscal support countries: increases by 1.5 percent after five years (statistically significant).
  - High fiscal support countries: increases by 0.7 percent after five years (statistically insignificant).
- Robustness checks: relative magnitudes for output, unemployment, poverty, and inequality for low versus high fiscal support using structural primary balance hold across a range of alternative specifications.
- Related evidence:
  - Ma et al. (2020): countries with higher government expenditures, especially on health care, have a higher output growth bounce-back.
  - Robustness checks performed dividing fiscal indicator by previous year’s GDP (Ma et al. approach).
  - Furceri et al. (2021b): extent of fiscal consolidation following past pandemics determines increase in inequality; extreme austerity associated with Gini increase three times as large as more supportive fiscal policy.

### Informality
- Informality data from Medina and Schneider (2020); countries split into “high” and “low” informality using values for the year before the pandemic (equation II).
- Output:
  - High informality countries: output falls by 2.6 percent after three years (lowest point).
  - Low informality countries: output falls by 2.2 percent after three years.
  - Effects statistically significant for both groups.
  - Output decline is higher for low informal group in the fourth and fifth year in some specifications.
- Unemployment:
  - Rises more for countries with high informality; effect statistically significant for high informality and not for low informality.
- Poverty:
  - High informality countries: poverty rate increases by 1.3 percentage points after the first year (near-term statistically significant).
  - Low informality countries: poverty rate increases by 0.7 percentage points after the first year.
  - Five years after the pandemic:
    - High informality countries: poverty rate remains 1 percentage points above pre-pandemic levels.
    - Low informality countries: poverty rate remains 0.4 percentage point above pre-pandemic levels.
- Inequality:
  - High informality countries: 0.5 percentage point increase after two years.
  - Low informality countries: 0.2 percentage point increase after two years.
  - After five years, rise in inequality slightly higher for high informality; effect statistically significant for both groups.
- Robustness: results for unemployment, poverty, and inequality generally robust across alternative specifications; some robustness cases show relative magnitudes for output by informality may not hold.

### Social Expenditure (family benefits and other components)
- OECD social expenditure data used; baseline uses family benefits component (public spending on family benefits).
- Decadal average of social expenditures used to determine “high” and “low” family benefits; regressions limited to around 30 countries.
- During Covid-19, increased social protection spending averaged 0.6 percent of GDP in the first three quarters of 2020 and estimated to have reduced the rise of extreme global poverty by about 10 million people (IMF 2021b).
- Findings for family benefits:
  - Countries with lower family benefits associated with relatively higher unemployment rise, higher poverty, and higher inequality (robust across specifications).
  - Output drops similar for low and high family benefits groups; robustness checks show relative magnitude contingent on specification.
  - Unemployment rise stronger for low family benefits group, though effect not statistically significant for either group.
  - Poverty rate rise: 0.15 percentage points after two years higher for countries with low family benefits; effects statistically significant for high group and not for low group.
  - Inequality increase after five years:
    - Low family benefits: 2.2 percent increase (statistically significant).
    - High family benefits: 1.5 percent increase (statistically significant).
- Figure 9 (summary for three-year horizon) using total social expenditures and components:
  - Negative output effect after three years is higher for countries with lower spending across total and components (with exception of family benefits).
  - Unemployment rate higher for countries with lower benefits across items (not statistically significant for either group).
  - Rise in poverty rate higher for countries with lower benefits compared to higher benefits.
  - Rise in inequality is higher for countries with lower family benefits and lower health spending, but lower for countries with lower total social expenditure and lower active labor market policies.

### Health Expenditure per Capita
- Health expenditure per capita from WDI: current health expenditure per capita in current international dollars (PPP).
- Output:
  - Low health expenditure countries: output falling by 3.8 percent after three years (statistically significant).
  - High health expenditure countries: output falling by 3.4 percent after three years (statistically significant).
  - Negative output-effect stronger for high expenditure group in years four and five.
- Unemployment:
  - Low health expenditure group: rise of 1.7 percentage points after four years (effects statistically significant).
  - High health expenditure group: rise of 0.8 percentage points after four years.
- Poverty:
  - Low health expenditure group: poverty rate increases by 1.0 percentage points within the first year and remains 0.7 percentage points above pre-pandemic levels after five years (near-term effect statistically significant).
  - High health expenditure group: poverty rate increases by 0.4 percentage points after the first year and falls slightly below pre-pandemic levels after five years (effects on poverty not statistically significant for high group).
- Inequality:
  - Inequality increases higher for low health expenditure countries (details and exact magnitudes for inequality continuation referenced in source).

*Source: wpiea2021181-print-pdf - https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021181-print-pdf.pdf*

### 1.2 percent after five years compared to 0.6 percent increase for the high expenditure group

### wpiea2021181-print-pdf - 1.2 percent after five years compared to 0.6 percent increase for the high expenditure group

### Key empirical findings
- Using a sample of 55 countries over 1990-2019 and employing the local projection methodology, past pandemics had lasting damaging scarring effects with detrimental impacts on output, unemployment, poverty, and inequality.
- A relatively stronger fiscal response helped to mitigate the negative output effects.
- A relatively stronger fiscal response and better initial conditions in informality, family benefits, and health expenditure per capita helped to alleviate negative effects on unemployment, poverty, and inequality.
- Results on output are not robust across all specifications and further work is needed to understand the channels better.
- Application of empirical estimates from past pandemics on inequality suggests significant scarring effects of Covid-19; Emmerling et al. (2021) project an increase in poverty of about 75 million people worldwide.
- Certain Covid-19 specific mechanisms amplify adverse outcomes in weaker initial-condition countries:
  - Social distancing led to reliance on online schooling; UN (2020) reports the largest disruption of education systems in history, affecting nearly 1.6 billion learners in more than 190 countries.
  - Higher informality is associated with lower internet access and lower account ownership at financial institutions or with mobile-money-service providers, exacerbating loss of human capital, poverty, and inequality.
  - Countries with lower social expenditure have less internet and financial access, further worsening poverty and inequality outcomes.

### Robustness checks (alternative formulations employed)
- The robustness checks include the following specifications:
  - (i) dropping all control variables, except output;
  - (ii) dropping all control variables, including output;
  - (iii) including all control variables;
  - (iv) including all control variables except credit;
  - (v) including additional control variables. As part of (v):
    - regressions controlling for two lags of public-to-GDP ratio;
    - regressions controlling for two lags of ICRG indicators;
  - (vi) including sample from 2000;
  - (vii) adding four lags, instead of two lags as specified in the baseline and augmented regressions;
  - (viii) splitting the sample using different percentiles. The augmented regressions incorporate policy and structural features by splitting the sample in the median. Also considered:
    - splitting using bottom 25th percentiles versus the rest (fiscal support, health expenditures per capita);
    - splitting using top 75th percentiles versus the rest (informality);
  - (ix) using contemporaneous values of control variables (in addition to two lags);
  - (x) using contemporaneous values of informality and health expenditure per capita (alternative to using lagged values one year before the pandemic);
  - (xi) using alternative indicators for fiscal. The main regressions use the fiscal impulse, which is the negative of the annual change in structural primary balance (share of potential GDP, expressed in percentage points). Alternative fiscal specifications:
    - divide the negative of the change in primary structural balance (in level terms) by the nominal GDP of the previous year;
    - use changes in structural overall fiscal balances, cyclically adjusted overall and primary fiscal balances, and overall and primary fiscal balances (instead of changes in structural primary balances);
  - (xii) using alternative definition of health: lagged current health expenditure as a share of GDP (instead of per capita) and general government health expenditure (as a share of GDP).

### Main findings from robustness checks
- (i) The output decline in the baseline is found to be more pronounced in some cases.
- (ii) The unemployment results come out statistically significant in some cases, compared to the baseline where it narrowly misses statistical significance.
- (iii) The relative magnitudes highlighted using fiscal policy are found to be robust using the main/baseline definition of fiscal impulse in the paper. In many cases the output decline is more pronounced for the relatively lower fiscal support group. The results hold when using overall structural balances and cyclically adjusted overall and primary balances, except poverty for cyclically adjusted balances. With the exception of output when primary balance is used and inequality when overall balance is used, the results do not hold when overall and primary fiscal balances (net lending/borrowing) are used. The paper emphasizes interest in discretionary fiscal support, so findings on the role of discretionary fiscal policies remain intact.
- (iv) The results for structural features (informality, family benefits, health) are generally robust for unemployment, poverty, and inequality. Exceptions:
  - The relative magnitudes for output using informality, family benefits, and health expenditure per capita do not always hold.
  - The impact of family benefits on output is inconclusive in robustness checks, but total social expenditure and other components show higher output decline for relatively lower support groups after three years.

### Policy implications and recommendations
- Adequate fiscal support, higher health spending, and targeted family benefits should be part of the policy package to limit long-lasting scarring effects of pandemics, especially for vulnerable groups and to mitigate persistent loss of human capital from Covid-19.
- Higher policy support and complementary policies might be required in countries with high informality to reduce larger negative impacts on unemployment, poverty, and inequality.
- Delechat et al. (2020) policy recommendations to address informality (listed in the source) include:
  - (i) improve access to and quality of education;
  - (ii) avoid tax system design that inadvertently increases incentives for individuals and firms to remain in the informal sector;
  - (iii) enhance financial inclusion by promoting access to formal financial services;
  - (iv) pursue a range of structural policies (e.g., in labor market regulations, competition policies, digitalization) to increase incentives and lower the cost of formalization.

### Summary quantitative points (preserved exactly as in source)
- 1.2 percent after five years compared to 0.6 percent increase for the high expenditure group.
- Sample: 55 countries over 1990-2019.
- Emmerling et al. (2021) project an increase in poverty of about 75 million people worldwide.
- UN (2020) reports nearly 1.6 billion learners affected in more than 190 countries.

*Source: IMF Working Paper content from wpiea2021181-print-pdf*

### 2017. Disease Control Priorities, vol. 9, World Bank: Washington DC, 3ed.

### wpiea2021181-print-pdf - 2017. Disease Control Priorities, vol. 9, World Bank: Washington DC, 3ed.

### Data, sample, and estimation
- Sample: 55 countries over the period 1990-2019.
- Pandemic episodes: taken from Ma et al. (2020) for the last five pandemics.
- Key data sources:
  - Outcome variables: World Development Indicators (WDI).
  - Inequality (Gini indices): Standardized World Income Inequality Database (SWIID).
  - Pandemic episodes: Ma et al. (2020).
  - Fiscal impulse: IMF World Economic Outlook (WEO) — defined as the negative of change in structural primary balance (percentage points of potential GDP).
  - Informal sector: Medina and Schneider (2020) — percent of GDP; covers activities hidden from official authorities for monetary, regulatory, and institutional reasons.
  - Social expenditure and its components: OECD.
  - Current health expenditure per capita (PPP current intl. $): WDI.
  - Banking crisis indicator: Laeven and Valencia (2013)/(2020) database.
  - Institutional quality: PRS Group, International Country Risk Guide (ICRG).
- Estimation approach:
  - Impulse response functions estimated by local projections.
  - Equations include: a pandemic dummy (and two lags), two lags of the dependent variable, two lags of output, and country- and time-fixed effects.
  - Output and unemployment equations additionally control for two lags of: log of income per capita, trade-to-GDP ratio, private credit-to-GDP ratio, and banking crisis (Laeven and Valencia).
  - Standard errors: clustered at the country level.
  - Confidence bands reported at 90 percent.
  - Timing convention: x-axis denotes years before and after pandemic where t=0 is the year of the pandemic.

### Main empirical findings
- Output, unemployment, poverty, and inequality responses to pandemics:
  - Estimated impulse responses for Output (%), Unemployment (percentage points), Poverty (percentage points), and Inequality (%) are presented up to t=5 years after pandemic (t measured relative to t=0).
  - Poverty refers to WDI’s poverty headcount ratio at $1.90 a day (2011 PPP; as a share of population); other poverty indicators also shown: headcount at $3.20/day and poverty gap measures.
  - Income-share responses are reported for deciles and quintiles (e.g., Lowest 10%, Lowest 20%, Highest 10%, Highest 20%) in percentage point changes.
- Heterogeneity by region and sample:
  - Figures present macroeconomic variables before and after pandemics for the entire sample and Latin America separately.
  - Structural features (current health expenditure per capita, social expenditure, family benefits, informality) are compared across regions and decades (e.g., 1991-1999, 2000-2009, 2010-2017).
- Statistical significance:
  - Filled bars in the three-year-ahead bar charts denote variables that are statistically significant at the 90 percent level for at least one year in the t=0 to t=5 horizon.

### Role of fiscal support (fiscal impulse)
- Fiscal impulse defined as the negative of the annual change in structural primary balance as a share of potential GDP.
- Comparative results:
  - Impulse responses are shown for High versus Low fiscal impulse (countries above versus below/equal to the median fiscal impulse in the pandemic year).
  - Outcomes reported include Output (%), Unemployment (ppt), Poverty (ppt), and Inequality (%).
- Three-year differential responses:
  - The differential effect three years after pandemic shock is reported via coefficients for Output (%), Poverty (ppt), Inequality (%), and Unemployment (ppt) comparing High versus Low fiscal impulse.

### Role of informality
- Informal sector measure: percent of GDP from Medina and Schneider (2020).
- Comparative results:
  - Impulse responses are shown for High versus Low informality (countries above versus below/equal to the median informal sector share in the year before pandemic).
  - Outcomes include Output (%), Unemployment (ppt), Poverty (ppt), and Inequality (%).
- Three-year differential responses:
  - Coefficients three years after pandemic shock presented for Informality comparing High versus Low.

### Role of social expenditure and components (family benefits, ALMP, health)
- Social expenditure and components sourced from OECD; family benefits defined as public spending on family benefits (cash transfers, services, tax-based support) per OECD.
- Comparative results:
  - Impulse responses are shown for High versus Low family benefits (median taken over decadal average), and for High versus Low total social expenditure and ALMP (Active Labor Market Programmes).
  - Outcomes include Output (%), Unemployment (ppt), Poverty (ppt), and Inequality (%).
- Three-year differential responses:
  - Bars show coefficients three years after pandemic shock for Total, Health, Family benefits, and ALMP on Output (%), Unemployment (ppt), Poverty (ppt), and Inequality (%).
  - Filled bars indicate statistical significance at the 90 percent level for at least one year in t=0 to t=5.

### Role of health spending per capita
- Current health expenditure per capita measured in PPP current international dollars (WDI).
- Comparative results:
  - Impulse responses shown for High versus Low health expenditure per capita (above versus below/equal to median in the year before pandemic).
  - Outcomes include Output (%), Unemployment (ppt), Poverty (ppt), and Inequality (%).
- Three-year differential responses:
  - Health spending per capita included among the set of covariates whose high/low medians are used to split samples for three-year-ahead coefficient comparisons.

### Cross-sectional associations: informality and social expenditure with development indicators
- Social Expenditure (% of GDP) regressed graphically against development indicators:
  - Positive associations displayed between Social Expenditure (% of GDP) and:
    - Individuals using the Internet (% of population): y = 0.46x + 76.89, R² = 0.11.
    - Account ownership at a financial institution or with a mobile-money-service provider (% ages 15+): y = 1.53x + 59.07, R² = 0.36.
    - Account ownership, female (% ages 15+): y = 1.70x + 54.30, R² = 0.36.
    - Account ownership, male (% ages 15+): y = 1.34x + 64.20, R² = 0.35.
- Informality (% GDP) regressed graphically against development indicators:
  - Negative associations displayed between Informality (% GDP) and:
    - Individuals using the Internet (% of population): y = -1.60x + 98.71, R² = 0.36.
    - Account ownership at a financial institution or with a mobile-money-service provider (% ages 15+): y = -1.46x + 95.67, R² = 0.30.
    - Account ownership (% ages 15+): y = -1.39x + 112.22, R² = 0.15.
    - School enrollment, secondary (% gross): y = -1.47x + 77.82, R² = 0.27.
    - School enrollment, tertiary (% gross): graphical relationship shown (no explicit equation printed for tertiary in supplied excerpt).

### Figures and visual summaries (methodological notes)
- Impulse response figures:
  - Solid line: estimated response.
  - Dotted lines: 90 percent confidence bands using standard errors clustered at the country level.
  - x-axis: time in years relative to pandemic (t=0 is the pandemic year).
  - y-axis: change in the variable of interest at time t, compared to the year before pandemic.
- Poverty figures:
  - Headcount ratios at $1.90/day and $3.20/day, and poverty gaps at these thresholds, presented in percentage point changes.
- Income share figures:
  - Changes in income shares for Lowest 10%, Lowest 20%, Highest 10%, Highest 20% reported in percentage points.
- Figures comparing High versus Low groups:
  - High/Low splits use medians (except family benefits where decadal averages are used).
  - High/Low comparisons presented for Fiscal impulse, Informality, Family benefits, Total social expenditure, Health spending per capita, and combinations thereof.

*Italic: wpiea2021181-print-pdf - 2017. Disease Control Priorities, vol. 9, World Bank: Washington DC, 3ed.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021181-print-pdf.pdf_