## wpiea2021120-print-pdf

## Source details

**Canonical URL:** [wpiea2021120-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021120-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021120-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021120-print-pdf.pdf.json)

---

### Data sources and construction
- Pandemic coverage:
  - Focus on five major epidemics since 2000: SARS (2003), H1N1 (2009), MERS (2012), Ebola (2014) and Zika (2016).
  - H1N1 had over 6 ½ million cases across 148 countries (about 1.2 cases per thousand people) and over 19,000 deaths.
  - Other events were regionally concentrated: SARS and MERS in Asia, Ebola in Africa, and Zika in the Americas.
  - Mortality rates (deaths/confirmed cases): MERS and Ebola most severe, followed by SARS, H1N1 and Zika.
  - Pandemic event indicator: constructed as a (0,1) dummy taking value 1 for countries declared by WHO to be affected; yields a total of 225 pandemic events.
- Inequality measures:
  - Gini coefficient from the Standardized World Income Inequality Database (SWIID - version 8.3).
  - SWIID provides market (pre-taxes and transfers) and net (post-taxes and transfers) income inequality for 177 countries from 1960 to the present.
  - Redistribution measured as market Gini minus net Gini.
- Fiscal and health expenditure data:
  - Fiscal balances from IMF’s World Economic Outlook (WEO) database; fiscal balance measured as general government net lending or borrowing (in percent of GDP); positive values indicate surpluses. (WEO release used: April 2020.)
  - Health expenditures (in percent of GDP) from World Bank’s World Development Indicators; data cover 180 countries from 2000 to the present and include current and capital health expenditures.

### Empirical approach and robustness
- Empirical strategy:
  - Local projections following Jordà (2005) estimate impulse response functions via:
    - y_{i,t+k} = α_i^k + γ_t^k + β_k D_{i,t} + θ_k X_{i,t} + ε_{i,t+k}
    - y is market or net Gini or redistribution; D is pandemic dummy; X includes four lags of dependent variable, pandemic dummy lags, country-specific time trends; horizons k=0,..,5.
- Robustness checks:
  - ADL, IV approach, AIPW estimation.
  - Alternative controls (development proxies, demographics, trade/financial globalization).
  - Continuous pandemic intensity measure, alternative inequality measures, and placebo tests.

### Key empirical findings on pandemic impacts
- Average effects (21st century pandemics):
  - Both market and net Gini increase following a pandemic.
  - Increase in Gini is about 0.3 after three years and is both statistically and economically significant.
  - Redistribution increases for the first two years and then declines toward zero; these redistribution effects are not statistically significant.
  - Effects on the Gini tend to level off past five years.
- Early evidence from COVID-19:
  - Infection and mortality distribution:
    - In some localities, the poor have been more prone to infection (e.g., richer zip codes in New York City less likely to test positive).
    - In the United States, African Americans account for 25% of deaths from COVID-19 though they make up a little under 13% of the US population.
    - Infection rates higher in U.S. counties with higher shares of African Americans and Hispanics and with higher income inequality.
  - Labor-market and income channels:
    - The poorest 20% of the population are in jobs that can be done from home in less than 20% of cases.
    - In Italy, low-education and low-income service workers were more likely to stop working and suffer immediate income falls; relaxations benefited highly educated and white-collar workers.
    - Survey and transaction data from the U.K. and Spain show increases in market income inequality driven by job losses and wage cuts for low-income workers.
  - Longer-term risks:
    - Nearly 40% of the global workforce is employed in sectors with high risk of worker displacement, many in informal employment with limited access to health services and social protection.
    - Global remittance flows are expected to fall 20% in 2020 (compared with a 5% fall during the 2009 financial crisis).

### Fiscal response, austerity, and distributional outcomes
- Average fiscal response:
  - Fiscal balance weakens after a pandemic; five years after start of pandemic, fiscal balance (as percent of GDP) is about 2 ½ percentage points lower than at the outset.
  - Government total health expenditures increase for four years after a pandemic before returning to normal; peak increase about 0.3 percentage points of GDP (driven by current health expenditures); capital health expenditures increase not statistically significant.
- Heterogeneity and nonlinear specification:
  - Extended specification allows pandemic impact to vary with country-specific response z (normalized to zero mean and unit variance) using weighting function F(z) = exp(−γ z) / [1 + exp(−γ z)] with γ=1.5.
  - Episodes segmented into: (i) low vs. high redistribution; (ii) mild vs. strong fiscal-balance response (surpluses/small deficits vs. larger deficits); (iii) small vs. large health-expenditure response.
  - Discretionary fiscal shocks isolated by regressing fiscal balance (or health expenditures) on GDP growth and country fixed effects; residuals interpreted as discretionary shocks.
- Main results on austerity vs. supportive fiscal response:
  - In austerity episodes, pandemics raise inequality substantially: five years on, the Gini increases by about 0.6 points (about twice the average impact).
  - When the fiscal response is strongly supportive, inequality barely increases after pandemics; this holds across measures of redistribution, fiscal balance response, and health-expenditure response.
  - The impact on the net Gini under austerity is roughly similar across the three measures and quantitatively large.

### Main findings on inequality and income transfers toward 2030
- Labor income shares and transfers:
  - "Labor income shares will also decrease, by more than 3 percentage points globally, implying a transfer of income from workers to profit earners of approximately $40 trillion by 2030."
- Austerity magnifies inequality:
  - "The increase in the Gini coefficient is nearly thrice as large in episodes of higher government fiscal balance (i.e. smaller fiscal deficits), lower government health expenditures, and lower redistribution than in cases where there was no turn to austerity."

### Evidence on fiscal policy potency and historical experience
- Post-2010 European experience documented cuts in health budgets and reductions in hospital investment; notable declines reported for Spain and Italy (Thomson, 2020).
- OECD (2016) found common measures in EU countries to balance health budgets included "reducing wages in public hospitals, postponing staff replacement and delaying investment in hospital infrastructure."
- Examples of fiscal policy potency during COVID-19:
  - Surico, Känzig and Hacioglu (2020) for the United Kingdom.
  - Aspachs et al. (2020) document public transfers were "very effective in offsetting most, though not all, of the increase in wage inequality in the early months of the COVID-19 crisis."
  - Balasubramanian et al. (2021) discuss effectiveness of electronic direct benefit transfers in India, with concerns that many informal workers may fall outside electronic safety nets (Furceri et al., 2020).

### Policy implications and recommendations
- Tradeoffs and prudence of fiscal support:
  - Two common arguments for aggressive debt reduction are addressed:
    - Reducing debt lowers perceived financial-crisis risk, but countries with strong fiscal records gain little insurance by modest debt reduction; moving from 120 percent of GDP to 100 percent yields only small crisis-risk reductions.
    - Fiscal consolidations could be expansionary, but expansionary austerity is rare; typically consolidations are followed by output drops.
  - Empirical magnitudes from prior literature:
    - On average, a consolidation of 1 percent of GDP increases the long-term unemployment rate by 0.6 percentage point and raises the Gini coefficient by 1.5 percent within five years.
- Recommended stance:
  - Case for strong fiscal support in many economies despite high debt-to-GDP ratios, given limited insurance gains from debt reduction and high social costs of premature austerity.
  - Central banks in several advanced and many emerging and middle-income economies can facilitate fiscal responses by directly or indirectly financing some debt buildup; low long-term interest rates and the ability to extend bond maturities moderate debt-service burdens.
  - Low-income developing countries need private creditor assistance and additional concessional financing from the official sector to alleviate financing constraints.
- Cautionary examples:
  - Premature fiscal consolidation after the Global Financial Crisis (starting in 2010) is cited as partly responsible for tepid recovery and failure to reduce debt-to-GDP ratios.
  - A turn to austerity may have affected governments’ health expenditures in the run-up to COVID-19.
- Risks:
  - Austerity following pandemics tends to raise inequality substantially; supportive fiscal policies can prevent or greatly reduce pandemic-induced increases in inequality.
  - Absent adequate fiscal support and financing, developing countries risk a lost decade of growth.
- Recommended fiscal policy design:
  - Anchor fiscal plans in a credible medium-term framework.
  - Orient public expenditures toward productive investments in digital and green infrastructure.
  - Maintain support until recovery is assured and target measures to help the most vulnerable.

### Data and sample statistics (preserved exactly)
- Impulse-response samples:
  - 177 countries (1960-2019) for Gini/redistribution.
  - 185 countries (1980-2019) for fiscal balance.
  - 180 countries (2000-2017) for health expenditures.
- TABLE 1: Total # of Episodes: 225.
- TABLE 2 (selected rows):
  - Gini Market (SWIID 8.3): Obs. 5,472; Mean 45.39; Std. Dev. 6.59; No. of Countries 177.
  - Gini Net (SWIID 8.3): Obs. 5,472; Mean 38.38; Std. Dev. 8.73; No. of Countries 177.
  - Redistribution (SWIID 8.3): Obs. 5,472; Mean 7.01; Std. Dev. 6.82; No. of Countries 177.
  - General government fiscal balance (WEO): Obs. 5,242; Mean -2.69; Std. Dev. 14.88; No. of Countries 185.
  - Capital Health Expenditures (WDI): Obs. 3,208; Mean 0.24; Std. Dev. 0.46; No. of Countries 180.
  - Current Health Expenditures (WDI): Obs. 3,208; Mean 6.20; Std. Dev. 2.58; No. of Countries 180.
  - Total Health Expenditures (WDI): Obs. 3,208; Mean 6.44; Std. Dev. 2.68; No. of Countries 180.

*wpiea2021120-print-pdf*

### Section II describes the data on pandemics, inequality and fiscal variables. Section III

### wpiea2021120-print-pdf - Section II describes the data on pandemics, inequality and fiscal variables. Section III

### Data sources and construction
- Pandemic coverage:
  - Focus on five major epidemics since 2000: SARS (2003), H1N1 (2009), MERS (2012), Ebola (2014) and Zika (2016).
  - H1N1 had over 6 ½ million cases across 148 countries (about 1.2 cases per thousand people) and over 19,000 deaths.
  - Other events were regionally concentrated: SARS and MERS in Asia, Ebola in Africa, and Zika in the Americas.
  - Mortality rates (deaths/confirmed cases): MERS and Ebola most severe, followed by SARS, H1N1 and Zika.
  - Pandemic event indicator: constructed as a (0,1) dummy taking value 1 for countries declared by WHO to be affected; yields a total of 225 pandemic events.
- Inequality measures:
  - Gini coefficient from the Standardized World Income Inequality Database (SWIID - version 8.3).
  - SWIID provides market (pre-taxes and transfers) and net (post-taxes and transfers) income inequality for 177 countries from 1960 to the present.
  - Redistribution measured as market Gini minus net Gini.
- Fiscal and health expenditure data:
  - Fiscal balances from IMF’s World Economic Outlook (WEO) database; fiscal balance measured as general government net lending or borrowing (in percent of GDP); positive values indicate surpluses. (WEO release used: April 2020.)
  - Health expenditures (in percent of GDP) from World Bank’s World Development Indicators; data cover 180 countries from 2000 to the present and include current and capital health expenditures.

### Evidence on pandemics and inequality (literature and empirical approach)
- Historical-pandemic findings:
  - Historical episodes give mixed outcomes: Black Death associated with reductions in inequality (real wages rise), while other plagues and recurrent cholera outbreaks often exacerbated poor outcomes; 1918 Spanish Flu evidence indicates income inequality rose in more-affected Italian towns.
  - Institutional characteristics and the ability of the wealthy to protect assets influence distributional outcomes.
- Empirical strategy for 21st-century pandemics:
  - Local projections following Jordà (2005) estimate impulse response functions via:
    - y_{i,t+k} = α_i^k + γ_t^k + β_k D_{i,t} + θ_k X_{i,t} + ε_{i,t+k}
    - y is market or net Gini or redistribution; D is pandemic dummy; X includes four lags of dependent variable, pandemic dummy lags, country-specific time trends; horizons k=0,..,5.
  - Robustness checks include ADL, IV approach, AIPW estimation, alternative controls (development proxies, demographics, trade/financial globalization), continuous pandemic intensity measure, alternative inequality measures, and placebo tests.

### Key empirical findings on pandemic impacts
- Average effects (21st century pandemics):
  - Both market and net Gini increase following a pandemic.
  - Increase in Gini is about 0.3 after three years and is both statistically and economically significant.
  - Redistribution increases for the first two years and then declines toward zero; these redistribution effects are not statistically significant.
  - Effects on the Gini tend to level off past five years.
- Early evidence from COVID-19:
  - Infection and mortality distribution:
    - In some localities, the poor have been more prone to infection (e.g., richer zip codes in New York City less likely to test positive).
    - In the United States, African Americans account for 25% of deaths from COVID-19 though they make up a little under 13% of the US population.
    - Infection rates higher in U.S. counties with higher shares of African Americans and Hispanics and with higher income inequality.
  - Labor-market and income channels:
    - The poorest 20% of the population are in jobs that can be done from home in less than 20% of cases.
    - In Italy, low-education and low-income service workers were more likely to stop working and suffer immediate income falls; relaxations benefited highly educated and white-collar workers.
    - Survey and transaction data from the U.K. and Spain show increases in market income inequality driven by job losses and wage cuts for low-income workers.
  - Longer-term risks:
    - Nearly 40% of the global workforce is employed in sectors with high risk of worker displacement, many in informal employment with limited access to health services and social protection.
    - Global remittance flows are expected to fall 20% in 2020 (compared with a 5% fall during the 2009 financial crisis).

### Fiscal response, austerity, and distributional outcomes
- Average fiscal response to pandemics:
  - Fiscal balance weakens after a pandemic; five years after start of pandemic, fiscal balance (as percent of GDP) is about 2 ½ percentage points lower than at the outset.
  - Government total health expenditures increase for four years after a pandemic before returning to normal; peak increase about 0.3 percentage points of GDP (driven by current health expenditures); capital health expenditures increase not statistically significant.
- Heterogeneity by fiscal stance — nonlinear model:
  - Extended specification allows pandemic impact to vary with country-specific response z (normalized to zero mean and unit variance) using weighting function F(z) = exp(−γ z) / [1 + exp(−γ z)] with γ=1.5.
  - Episodes segmented into: (i) low vs. high redistribution; (ii) mild vs. strong fiscal-balance response (surpluses/small deficits vs. larger deficits); (iii) small vs. large health-expenditure response.
  - Discretionary fiscal shocks isolated by regressing fiscal balance (or health expenditures) on GDP growth and country fixed effects; residuals interpreted as discretionary shocks.
- Main results on austerity vs. supportive fiscal response:
  - In austerity episodes, pandemics raise inequality substantially: five years on, the Gini increases by about 0.6 points (about twice the average impact).
  - When the fiscal response is strongly supportive, inequality barely increases after pandemics; this holds across measures of redistribution, fiscal balance response, and health-expenditure response.
  - The impact on the net Gini under austerity is roughly similar across the three measures and quantitatively large.

### Policy implications and considerations
- Tradeoffs and prudence of fiscal support:
  - Two common arguments for aggressive debt reduction are addressed:
    - Reducing debt lowers perceived financial-crisis risk, but countries with strong fiscal records gain little insurance by modest debt reduction; moving from 120 percent of GDP to 100 percent yields only small crisis-risk reductions.
    - Fiscal consolidations could be expansionary, but expansionary austerity is rare; typically consolidations are followed by output drops.
  - Empirical magnitudes from prior literature:
    - On average, a consolidation of 1 percent of GDP increases the long-term unemployment rate by 0.6 percentage point and raises the Gini coefficient by 1.5 percent within five years.
- Policy stance recommendations and constraints:
  - There is a case for strong fiscal support in many economies despite high debt-to-GDP ratios, given limited insurance gains from debt reduction and the high social costs of premature austerity.
  - Central banks in several advanced and many emerging and middle-income economies can facilitate fiscal responses by directly or indirectly financing some debt buildup; low long-term interest rates and the ability to extend bond maturities moderate debt-service burdens.
  - Low-income developing countries face tighter constraints; alleviating financing constraints may require private creditor assistance and additional concessional financing from the official sector.
  - Cautionary examples:
    - Premature fiscal consolidation after the Global Financial Crisis (starting in 2010) is cited as partly responsible for tepid recovery and failure to reduce debt-to-GDP ratios.
    - A turn to austerity may have affected governments’ health expenditures in the run-up to COVID-19.
- Risks of austerity in the pandemic context:
  - Austerity following pandemics tends to raise inequality substantially; supportive fiscal policies can prevent or greatly reduce pandemic-induced increases in inequality.
  - Absent adequate fiscal support and financing, developing countries risk a lost decade of growth.

*wpiea2021120-print-pdf*

### 2030. Labor income shares will also decrease, by more than 3 percentage points globally, implying a transfer of

### wpiea2021120-print-pdf - 2030. Labor income shares will also decrease, by more than 3 percentage points globally, implying a transfer of

### Main findings on inequality and income transfers
- "Labor income shares will also decrease, by more than 3 percentage points globally, implying a transfer of income from workers to profit earners of approximately $40 trillion by 2030."
- The rise in inequality in the aftermath of major epidemics over the last two decades has been higher in episodes of greater austerity:
  - "The increase in the Gini coefficient is nearly thrice as large in episodes of higher government fiscal balance (i.e. smaller fiscal deficits), lower government health expenditures, and lower redistribution than in cases where there was no turn to austerity."

### Evidence on fiscal policy, health spending, and past crises
- Post-2010 European experience documented cuts in health budgets and reductions in hospital investment; notable declines reported for Spain and Italy (Thomson, 2020).
- OECD (2016) found common measures in EU countries to balance health budgets included "reducing wages in public hospitals, postponing staff replacement and delaying investment in hospital infrastructure."
- Empirical impulse-response evidence:
  - Estimates use samples of up to 185 countries for fiscal balance (1980-2019), 180 countries for health expenditures (2000-2017), and 177 countries for Gini/redistribution (1960-2019). (See figure notes for estimation equations and controls.)
- Examples of fiscal policy potency during COVID-19:
  - Surico, Känzig and Hacioglu (2020) for the United Kingdom.
  - Aspachs et al. (2020) document public transfers were "very effective in offsetting most, though not all, of the increase in wage inequality in the early months of the COVID-19 crisis."
  - Balasubramanian et al. (2021) discuss effectiveness of electronic direct benefit transfers in India, with concerns that many informal workers may fall outside electronic safety nets (Furceri et al., 2020).

### Policy recommendations and rationale
- Against a premature return to austerity, countries would do better by:
  - Anchoring fiscal plans in a credible medium-term framework.
  - Orienting public expenditures toward productive investments in digital and green infrastructure (Gaspar 2020).
- Advantages of these steps:
  - "By building market confidence in fiscal sustainability and boosting growth, respectively, these two steps can bring down the debt-to-GDP ratio over time in a more durable way than sharp fiscal consolidations, which risk causing such an immediate fall in output and keeping the debt-to-GDP ratio unchanged or even raising it."
- IMF and other observers urge caution: maintain support until recovery is assured and target measures to help the most vulnerable (Georgieva, 2020).
- For low-income countries, maintaining debt sustainability likely requires further support from private creditors (citing G20 Leaders’ Declaration, November 21, 2020).

### Conclusions and open questions for future work
- Descriptive evidence suggests COVID-19 is likely to increase inequality absent supportive policy actions, including fiscal and monetary policies that are more supportive than in the past.
- The paper highlights the importance of preserving fiscal support to check the rise of excessive inequality, given other drivers such as increased automation.
- Suggested avenues for future research:
  - Whether the composition of fiscal consolidation (spending cuts vs. tax increases) matters for distributional impacts.
  - How distributional impacts differ across country groups (advanced vs. emerging markets and low-income countries).

### Key statistics and data points preserved from the source
- "$40 trillion by 2030."
- "more than 3 percentage points globally" (decline in labor income shares).
- "The increase in the Gini coefficient is nearly thrice as large" (in episodes of greater austerity).
- Sample and data statistics (from tables and figure notes):
  - Impulse-response samples: 177 countries (1960-2019) for Gini/redistribution; 185 countries (1980-2019) for fiscal balance; 180 countries (2000-2017) for health expenditures.
  - TABLE 1: Total # of Episodes: 225.
  - TABLE 2 (selected rows):
    - Gini Market (SWIID 8.3): Obs. 5,472; Mean 45.39; Std. Dev. 6.59; No. of Countries 177.
    - Gini Net (SWIID 8.3): Obs. 5,472; Mean 38.38; Std. Dev. 8.73; No. of Countries 177.
    - Redistribution (SWIID 8.3): Obs. 5,472; Mean 7.01; Std. Dev. 6.82; No. of Countries 177.
    - General government fiscal balance (WEO): Obs. 5,242; Mean -2.69; Std. Dev. 14.88; No. of Countries 185.
    - Capital Health Expenditures (WDI): Obs. 3,208; Mean 0.24; Std. Dev. 0.46; No. of Countries 180.
    - Current Health Expenditures (WDI): Obs. 3,208; Mean 6.20; Std. Dev. 2.58; No. of Countries 180.
    - Total Health Expenditures (WDI): Obs. 3,208; Mean 6.44; Std. Dev. 2.68; No. of Countries 180.

*Source: Excerpts from the provided IMF working paper content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021120-print-pdf.pdf_