## 1. Economic Effects of Natural Disasters

## Source details

**Canonical URL:** [1. Economic Effects of Natural Disasters](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023261-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023261-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023261-print-pdf.pdf.json)

---

### Exposure and types of disasters
- Droughts affect, on average, 1.6 percent of the population annually in sub-Saharan Africa (SSA) since 1990.
- This level of exposure is thirty-three times higher than in advanced economies (AEs).
- Storms and floods impact around 0.5 percent of the population annually in emerging markets and developing economies (EMDEs).
- Island countries in the Northern Pacific and the Caribbeans are particularly vulnerable to storms; Southeastern Africa (e.g., Madagascar and Mozambique) is frequently hit by cyclones.
- The frequency and intensity of these disasters are expected to increase as temperature continues to rise (IPCC 2023).

### Macroeconomic and fiscal impacts (key empirical findings)
- Real GDP growth:
  - Lowered by more than one percentage point in EMDEs in the year of a drought or a severe storm.
  - The impact of storms appears to arise mostly in low-income countries (LICs).
  - Droughts and storms did not have significant impact in AEs.
  - No significant growth impact of floods in EMDEs was found.
- Fiscal revenue:
  - Total real revenue declined by 4.5 percentage points—leading to a 0.5 percent decline in government revenue as a share of GDP—in a drought year in EMDEs.
  - No statistically significant association between fiscal revenue and storms or floods was found.
- Public expenditure:
  - Does not change meaningfully in response to disasters in EMDEs.
  - Absolute level of expenditure declined slightly during a disaster year, producing only a slight increase in expenditure as a share of GDP.
  - LICs, which experience larger growth declines, tend to have higher spending as a share of GDP in the aftermath of storms.
- Primary balance and debt:
  - Primary balance tends to worsen during drought and storm years, raising public debt in EMDEs.
  - In AEs, these disasters have no significant impact on the primary balance or public debt.

### Mechanisms and heterogeneity
- Two main channels by which climate change-induced disasters affect fiscal positions:
  - Slower economic activity erodes the tax base, leading to a decline in revenue.
  - Disaster responses (social protection, reconstruction) could elevate fiscal expenditure, but actual expenditure responses differ across countries and may be muted by limited fiscal space.
- Destruction of public infrastructure can further impair revenue mobilization capacity.
- Heterogeneity examined across:
  - Country groups: AEs vs EMDEs; within EMDEs: LICs vs emerging markets (EMs); special emphasis on SSA.
  - Disaster types: droughts, storms, floods.
  - Institutional and economic factors: access to development assistance, levels of trade openness, and quality of governance.

### Descriptive statistics and trends
- Extreme disasters are recorded every year globally; droughts are more common than extreme storms and floods.
- The number of droughts has increased in recent years, topping the early 2000s high drought incidence.
- Regional averages show:
  - Revenue as a share of GDP is lower in SSA compared to other-EMDEs and AEs.
  - Public expenditure as a share of GDP is smaller in SSA relative to other regions.

---

### Data Definitions, Measurement, and Empirical Approach

### Data definitions and measurement
- Droughts:
  - Measured with the Palmer Drought Severity Index (PDSI), which ranges from -10 (driest) to +10 (wettest) and is concentrated around [-4, +4].
  - PDSI raw data at 0.5° × 0.5° resolution are aggregated to first administrative divisions and then to country level weighted by nightlight-based proxies for economic activity.
  - A country-year is considered facing drought when the aggregated index is below –2 during crop growing seasons.
  - Crop growing seasons are identified by the uptake in the Normalized Difference Vegetation Index (NDVI); three consecutive months with the highest increase in NDVI mark the beginning of the crop growing season.
  - Formal indicator: Dit = I{[∑NTLp × (1/3 ∑PDSI i,p,t,m for m=1..3 )p] < −2}, with NTLp as provincial nightlight weight.
- Storms and floods:
  - Sourced from EM-DAT (Emergency Events Database), focusing on meteorological and hydrological disasters with human-cost focus.
  - An extreme flood or storm is defined when (death toll) + 0.3 × (affected population) makes up at least 1 percent of the total population:
    - Dit = I{ [ (Death toll)it + 0.3 (Affected Population)it ] / Populationit > 0.01 }.
  - EM-DAT inclusion criteria: 10 or more people dead; 100 or more people affected; declaration of a state of emergency; or a call for international assistance.
- Macro and auxiliary data:
  - GDP, revenue, expenditure, debt, and current account from IMF World Economic Outlook (WEO) database.
  - Terms of trade (TOT), agricultural value added, and population from World Bank WDI.
  - Governance index (government effectiveness) ranges from 0 (lowest rank) to 100 (highest rank); countries grouped into “High Governance” and “Low Governance” using median value of 50 as threshold.

### Panel regression (baseline and heterogeneity tests)
- Baseline specification (global sample):
  - Y_it (real GDP growth, revenue, expenditure, primary fiscal balance) regressed on country fixed effects (γ_i), lagged dependent variable (π Y_{i,t−1}), disaster dummy D_it, and controls X_it (year-on-year inflation, TOT level, current account deficit, agricultural value added as share of GDP, debt to GDP ratio, total population, GDP per capita (US$), and ODA in US$ per capita).
  - Model: Y_it = γ_i + π Y_{i,t−1} + α D_it + X'_it β + e_it. α indicates the instantaneous impact of disasters.
- Regional differential impacts:
  - Interaction specification with AE and EMDE indicators to obtain α_AEs and α_EMDEs.
- Heterogeneity by ODA access, trade openness, and governance:
  - Interaction specification: Y_it = γ_i + π Y_{i,t−1} + α_1 (D_it × C_it) + α_2 (D_it × (1−C_it)) + X'_it β + e_it, with C_it = 1 for countries above median in the given characteristic.
- Additional subgroup analysis:
  - EMDE-only regressions to compare SSA versus other-EMDEs and LICs versus EMs.

### Local projections (dynamic responses)
- Specification for horizon h:
  - Y_{i,t+h} = π Y_{i,t−1} + Σ_{p=0}^P α_p^h D_{i,t−p} + Σ_{q=0}^Q X'_{t−q} β_q^h + γ_{i,h} + δ_{t,h}, with γ_{i,h} and δ_{t,h} country and time fixed effects varying by horizon.
  - Coefficients α_p^h trace dynamic responses; lags p and q set to one.

### Key econometric notes
- Revenue, expenditure, and primary balance defined as percentage of GDP unless otherwise noted.
- Absolute changes use year-on-year change of real revenue and expenditure (not as shares of GDP) to disentangle numerator-denominator co-movement.
- Robustness checks: additional year fixed effects; dynamic panel model with two years lag (Arellano and Bond 1991); regressions accounting for historical shock. Results in Appendix B consistent with main findings.

---

### 3. Empirical Findings: Instantaneous Impacts

### Instantaneous growth impacts (summary)
- Droughts reduce growth by 1 percentage point globally.
- High-intensity storms decrease growth by 1.6 percentage point globally.
- Floods have limited and localized impact on growth (weaker effect).
- Differential impacts:
  - Droughts reduce growth by 1.4 percentage point in EMDEs relative to AEs.
  - Storms reduce growth by 1.8 percentage point in EMDEs relative to AEs.
  - Storms in AEs show a surprising positive growth impact (potential post-disaster cleanup/reconstruction effects).
- Regional and income-group findings:
  - Growth of SSA countries is similarly dragged as elsewhere; gap between SSA and other-EMDEs is negligible.
  - LICs are impacted more by storms than EMs; droughts and floods affect LICs and EMs similarly.
- Example: Cyclone Idai (March 2019) caused US$ 1.4 billion damage in Mozambique; reconstruction cost estimated at US$ 2.9 billion (PDNA 2019).

### Instantaneous fiscal impacts (summary)
- Droughts:
  - Globally, droughts reduced revenue by 0.5 percent of GDP.
  - In EMDEs, droughts reduced revenue by 0.7 percent of GDP; AEs do not experience any decrease.
  - Absolute magnitude (real US$) decline in revenue due to droughts: about 3.4 percentage points globally and 4.5 percentage points in EMDEs (Table A.1).
  - Public expenditure does not surge in response to droughts; expenditure as percent of GDP remains unchanged after droughts.
  - Primary deficit increases by as much as 1.5 percent of GDP in EMDEs following droughts.
  - Public debt increases by 1.4 percent of GDP in EMDEs following droughts.
- Storms:
  - Fiscal impact is muted for both EMDEs and AEs.
  - Declines in absolute revenue and expenditure marginal; significant for AEs only (Table A.1).
  - Slight increase in revenue and expenditure as percent of GDP observed globally and in EMDEs (as GDP shrinks).
  - In AEs, expenditure as a share of GDP declined by 1 percent after storms.
  - Impact on primary balance and debt negligible for both groups.
- Floods:
  - Fiscal expenditure and revenue do not change much after floods except in AEs where revenue appears to decline.
  - Massive flooding episodes affecting more than 30 percent of a country’s population led to a decrease in public spending by 4.7 percentage points in AEs (Table A.1); as a share of GDP this change is negligible but adds to debt.
  - In EMDEs, neither revenue nor expenditure responds to floods.

### Differential impacts (highlights from Table 2)
- Table 2 reports differential impacts across growth, revenue, expenditure, primary balance, and debt (values shown as point estimates):

- Drought — SSA relative to other-EMDEs:
  - Growth: -0.03
  - Revenue: -0.77
  - Expenditure: -0.44
  - Primary Balance: -0.20
  - Debt: -0.33
  - Observations: 2921 (revenue sample 2881; expenditure 2880; primary balance 2873; debt 2832)
  - R-squared examples: Growth 0.084; Revenue 0.494; Expenditure 0.572; Primary Balance 0.343; Debt 0.829

- Drought — LICs relative to EMs:
  - Growth: 0.02
  - Revenue: -0.28
  - Expenditure: -0.12
  - Primary Balance: -0.06
  - Debt: 0.18
  - Observations: 1549 (revenue 1526; expenditure 1522; primary balance 1520; debt 1498)
  - R-squared examples: Growth 0.027; Revenue 0.465; Expenditure 0.590; Primary Balance 0.187; Debt 0.837

- Storm — SSA relative to other-EMDEs:
  - Growth: 0.33
  - Revenue: 1.36
  - Expenditure: 2.09*** 
  - Primary Balance: -0.97
  - Debt: -1.56

- Storm — LICs relative to EMs:
  - Growth: -2.19**
  - Revenue: 0.50
  - Expenditure: 0.73*
  - Primary Balance: -0.21
  - Debt: 0.41

- Flood — SSA relative to other-EMDEs:
  - Growth: -0.03
  - Revenue: -0.73
  - Expenditure: -0.22
  - Primary Balance: 0.03
  - Debt: -4.07

- Flood — LICs relative to EMs:
  - Growth: 0.15
  - Revenue: -0.55
  - Expenditure: -0.15
  - Primary Balance: 0.16
  - Debt: -2.26

- Notes: Standard errors are reported in parentheses in Table 2 in the source. EMs = Emerging Markets, EMDEs = Emerging Markets and Developing Economies, LICs = Low-Income Countries, SSA = Sub-Saharan Africa.

---

### 4. Persistence, Mitigating Factors, and Debt Dynamics

### Mitigating and aggravating factors
- Official Development Assistance (ODA):
  - EMDEs with larger ODA inflows attenuate adverse growth and fiscal impacts.
  - Countries with limited access to ODA experienced greater growth decreases of 1.9–2.4 percentage points after droughts and storms; countries with better access saw almost negligible reductions (Table A.2).
  - Absence of ODA forces spending from public budgets, increasing non-grant public expenditure, deficits, and debt.
  - Role of ODA for floods is unclear; both revenue and expenditure decline for countries with low ODA.
- Trade openness:
  - Countries reliant on import-export markets experienced larger adverse impacts from climate shocks.
  - Droughts decrease agricultural output and drag growth for export-reliant countries, resulting in higher fiscal deficits and debt (Table A.3).
  - Storms may impose larger growth drags on more export-oriented economies due to logistics and transportation disruptions, leading to higher public expenditure for reconstruction/imports.
  - Floods show no differential growth impact by trade openness, but revenue and expenditure of low-exposure countries decline after floods.
- Governance quality:
  - Better regulatory quality helps contain growth reduction effects of droughts.
  - Countries with below-average regulatory quality faced larger growth reductions and fiscal deterioration after droughts.
  - For storms, growth reduced regardless of regulatory quality; magnitude of decline greater for those with better regulatory quality, while fiscal impact is muted for both groups.
  - For floods, gaps in fiscal impact are minimal; growth of those with better regulatory quality seems to decline strongly (Table A.4).

### Persistence of impacts (local projections)
- Growth persistence:
  - Negative impact on growth is short-lived.
  - Droughts and storms cause negative growth knock-on effects in EMDEs that dissipate within a year.
  - Initial growth impact more adverse in EMDEs but EMDEs recover more quickly and return to higher growth subsequently.
  - Floods have negligible initial impacts; in subsequent years they seem beneficial for growth, especially in AEs, with a positive medium-term impact in both EMDEs and AEs.
- Fiscal persistence:
  - Fiscal impacts tend to last longer than growth impacts and are heterogeneous by country group and disaster type.
  - For EMDEs:
    - Droughts reduce revenue and raise expenditure in the medium term, resulting in persistent primary deficit.
    - Floods depress both revenue and expenditure, leading to a small primary balance surplus.
    - Storms exert a small positive impact on revenue but a much larger reduction in expenditure, giving rise to a sizeable increase in primary surplus.
  - For AEs:
    - Drought has a more positive impact on revenue and expenditure.
    - Flood and storm have more negative impacts compared to EMDEs.
    - On average, natural disasters’ impact on AEs’ primary balance is not statistically significant.
- Debt dynamics:
  - Medium-term debt-to-GDP trajectory of EMDEs does not necessarily worsen after disasters.
  - For droughts and storms, debt-to-GDP ratio starts to decline after three years; for floods, reduction is almost immediate in the first year.
  - Drivers of debt dynamics:
    - Post-disaster GDP rebound raises the denominator and can reduce debt-to-GDP ratio.
    - Primary balance worsens after droughts but improves after floods and storms.
    - Controlled inflation and monetary policy lowering domestic interest rates can reduce interest payments and improve debt dynamics.
  - Empirically, average debt-to-GDP shows signs of decline following droughts, storms, and floods, but heterogeneity across countries is large.

---

### Policy Implications and Conclusions

### Key conclusions
- The impact of natural disasters on economic growth and fiscal positions is heterogeneous across disaster types and countries.
- Droughts and severe storms reduce growth; floods tend to have mild macroeconomic impacts.
- Negative economic consequences are usually larger for EMDEs than for AEs, but among EMDEs the adverse impact on LICs is not always worse than on EMs.
- The typical expected fiscal response—reduced revenue as economic activity shrinks and elevated expenditure as government responds—does not always materialize.

### Role of fiscal space and transmission channels
- The interplay between economic growth, changes in the tax base, and availability of fiscal space is crucial in determining fiscal outcomes after disasters.
- A muted expenditure response in EMDEs signals the absence of fiscal space, highlighting the tension between the need to react and constraints of financial resources.
- Transmission channels from disasters to fiscal positions vary across countries and are related to other macro variables; a data-driven understanding of these channels is essential.

### Strengthening monitoring, analysis, and preparedness
- Enhancing disaster monitoring and impact analysis is a crucial first step toward preserving macro-fiscal stability.
- A data-driven approach should identify the disaster’s type and severity and map its transmission channels to fiscal positions, recognizing cross-country heterogeneity.

### Building fiscal resilience
- Building fiscal buffers and maintaining ample fiscal space are essential for effective response to natural disaster shocks.
- The documented discord between the need to help and the observed lack of expenditure response underscores the necessity of fiscal capacity that promotes resilience to disasters.
- Traditional policy tools—revenue mobilization, expenditure control, and public financial management—remain highly relevant to strengthen disaster resilience.

### International support and timing
- International support for disaster preparedness and adaptation is instrumental in attenuating the effects of climate-related disasters.
- While natural disasters’ negative impact on growth is short-lived, the fiscal impact tends to linger.
- In countries with limited fiscal space, faster mobilization of international support is needed to help alleviate fiscal burdens and contain scarring effects.

- Policy recommendation highlighted:
  - Integrating disaster response in the budget process and building fiscal resilience (fiscal buffers and self-insurance) can help countries better respond to climate change-induced disasters.
  - Reference made to IMF note “How to Manage the Fiscal Costs of Natural Disasters” (2018) for details on managing fiscal costs, building fiscal buffer, and utilizing it effectively.

*Source: IMF WORKING PAPERS — Fiscal Impacts of Climate Disasters in Developing Countries (Sections 1, 3, and 5).*

### 1. Economic Effects of Natural Disasters ...............................................................................

### 1. Economic Effects of Natural Disasters

### Exposure and types of disasters
- Droughts affect, on average, 1.6 percent of the population annually in sub-Saharan Africa (SSA) since 1990.
- This level of exposure is thirty-three times higher than in advanced economies (AEs).
- Storms and floods impact around 0.5 percent of the population annually in emerging markets and developing economies (EMDEs).
- Island countries in the Northern Pacific and the Caribbeans are particularly vulnerable to storms; Southeastern Africa (e.g., Madagascar and Mozambique) is frequently hit by cyclones.
- The frequency and intensity of these disasters are expected to increase as temperature continues to rise (IPCC 2023).

### Macroeconomic and fiscal impacts (key empirical findings)
- Real GDP growth is lowered by more than one percentage point in EMDEs in the year of a drought or a severe storm.
  - The impact of storms appears to arise mostly in low-income countries (LICs).
  - Droughts and storms did not have significant impact in AEs.
  - No significant growth impact of floods in EMDEs was found.
- Fiscal revenue (total real revenue) declined by 4.5 percentage points—leading to a 0.5 percent decline in government revenue as a share of GDP—in a drought year in EMDEs.
  - No statistically significant association between fiscal revenue and storms or floods was found.
- Public expenditure does not change meaningfully in response to disasters in EMDEs.
  - Absolute level of expenditure declined slightly during a disaster year, producing only a slight increase in expenditure as a share of GDP.
  - LICs, which experience larger growth declines, tend to have higher spending as a share of GDP in the aftermath of storms.
- Primary balance tends to worsen during drought and storm years, raising public debt in EMDEs.
  - In AEs, these disasters have no significant impact on the primary balance or public debt.

### Mechanisms and heterogeneity
- Two main channels by which climate change-induced disasters affect fiscal positions:
  - Slower economic activity erodes the tax base, leading to a decline in revenue.
  - Disaster responses (social protection, reconstruction) could elevate fiscal expenditure, but actual expenditure responses differ across countries and may be muted by limited fiscal space.
- Destruction of public infrastructure can further impair revenue mobilization capacity.
- Heterogeneity examined across:
  - Country groups: AEs vs EMDEs; within EMDEs: LICs vs emerging markets (EMs); special emphasis on SSA.
  - Disaster types: droughts, storms, floods.
  - Institutional and economic factors: access to development assistance, levels of trade openness, and quality of governance.

### Data definitions and measurement
- Droughts:
  - Measured with the Palmer Drought Severity Index (PDSI), which ranges from -10 (driest) to +10 (wettest) and is concentrated around [-4, +4].
  - PDSI raw data at 0.5° × 0.5° resolution are aggregated to first administrative divisions and then to country level weighted by nightlight-based proxies for economic activity.
  - A country-year is considered facing drought when the aggregated index is below –2 during crop growing seasons.
  - Crop growing seasons are identified by the uptake in the Normalized Difference Vegetation Index (NDVI); three consecutive months with the highest increase in NDVI mark the beginning of the crop growing season.
  - Formal indicator (as defined in the text): Dit = I{[∑NTLp × (1/3 ∑PDSI i,p,t,m for m=1..3 )p] < −2}, with NTLp as provincial nightlight weight.
- Storms and floods:
  - Sourced from EM-DAT (Emergency Events Database), focusing on meteorological and hydrological disasters with human-cost focus.
  - An extreme flood or storm is defined when (death toll) + 0.3 × (affected population) makes up at least 1 percent of the total population:
    - Dit = I{ [ (Death toll)it + 0.3 (Affected Population)it ] / Populationit > 0.01 }.
  - EM-DAT inclusion criteria: 10 or more people dead; 100 or more people affected; declaration of a state of emergency; or a call for international assistance.
- Macro and auxiliary data:
  - GDP, revenue, expenditure, debt, and current account from IMF World Economic Outlook (WEO) database.
  - Terms of trade (TOT), agricultural value added, and population from World Bank WDI.
  - Governance index (government effectiveness) ranges from 0 (lowest rank) to 100 (highest rank); countries grouped into “High Governance” and “Low Governance” using median value of 50 as threshold.

### Descriptive statistics and trends
- Extreme disasters are recorded every year globally; droughts are more common than extreme storms and floods.
- The number of droughts has increased in recent years, topping the early 2000s high drought incidence.
- Regional averages show:
  - Revenue as a share of GDP is lower in SSA compared to other-EMDEs and AEs.
  - Public expenditure as a share of GDP is smaller in SSA relative to other regions.

### Literature and contribution
- Prior studies document negative impacts of cyclones, hurricanes, rainfall changes, and droughts on economic growth and sectoral outcomes (e.g., Hsiang 2010; Acevedo 2016; Bakkensen and Barrage 2018; Desbureaux and Rodella 2019; Kotz et al. 2022; Mendelsohn 2009; De Winne and Peersman 2021).
- Studies on fiscal impacts report decreased revenue, increased expenditure, and higher debt following weather or disaster shocks (e.g., Melecky and Raddatz 2011; Akyapi et al. 2022; Duenwald et al. 2022).
- This study distinguishes disaster types, expands country coverage to 164 countries over three decades, focuses on EMDEs with emphasis on SSA, and employs local projections to assess persistence of fiscal impacts.
- The analysis focuses on natural disasters (human-cost centered as in EM-DAT) rather than extreme weather events.

### Policy implication highlighted
- Integrating disaster response in the budget process and building fiscal resilience (fiscal buffers and self-insurance) can help countries better respond to climate change-induced disasters.
  - Reference made to IMF note “How to Manage the Fiscal Costs of Natural Disasters” (2018) for details on managing fiscal costs, building fiscal buffer, and utilizing it effectively.

*Source: IMF WORKING PAPERS — Fiscal Impacts of Climate Disasters in Developing Countries, Section 1. Economic Effects of Natural Disasters.*

### 3. Empirical Approach

### 3. Empirical Approach

### Panel Regression: baseline and heterogeneity tests
- Baseline specification (global sample):
  - Y_it (real GDP growth, revenue, expenditure, primary fiscal balance) is regressed on:
    - country fixed effects (γ_i)
    - lagged dependent variable (π Y_{i,t−1})
    - disaster dummy D_it (drought, extreme flood, or extreme storm analyzed separately)
    - vector of controls X_it: year-on-year inflation, level of terms of trade (TOT), current account deficit, value added of agriculture as a share of GDP, debt to GDP ratio, total population, GDP per capita (US$), and official development assistance (ODA) in US$ per capita
  - Model: Y_it = γ_i + π Y_{i,t−1} + α D_it + X'_it β + e_it
  - Interpretation: α indicates the (causal) instantaneous impact of disasters.

- Regional differential impacts (AEs vs EMDEs):
  - Interaction specification:
    - Y_it = γ_i + π Y_{i,t−1} + α_AEs (D_it * AE_i) + α_EMDEs (D_it * EMDE_i) + X'_it β + e_it
  - Coefficients α_AEs and α_EMDEs measure disaster impacts in AEs and EMDEs respectively.

- Heterogeneity by ODA access, trade openness, and governance:
  - Interaction specification:
    - Y_it = γ_i + π Y_{i,t−1} + α_1 (D_it × C_it) + α_2 (D_it × (1−C_it)) + X'_it β + e_it
  - C_it = 1 for countries above median in ODA access, trade openness, or governance; 0 otherwise.
  - α_1 captures impacts for higher-C countries; α_2 captures impacts for lower-C countries.

- Additional subgroup analysis:
  - EMDE-only regressions to compare SSA versus other-EMDEs:
    - Y_it = γ_i + π Y_{i,t−1} + α_SSA (D_it * SSA_i) + X'_it β + e_it
  - Similar analysis for LICs versus EMs.

### Local projections: dynamic responses
- Local projection specification for horizon h (h = 0,1,2,...):
  - Y_{i,t+h} = π Y_{i,t−1} + Σ_{p=0}^P α_p^h D_{i,t−p} + Σ_{q=0}^Q X'_{t−q} β_q^h + γ_{i,h} + δ_{t,h}
  - γ_{i,h} and δ_{t,h} are country and time fixed effects varying by horizon.
  - Coefficients α_p^h trace dynamic responses to disasters.
  - Lags p and q set to one in this paper.

### Key econometric notes
- Revenue, expenditure, and primary balance are defined as a percentage of GDP unless otherwise noted.
- When analyzing absolute changes, year-on-year change of real revenue and expenditure (not as shares of GDP) is used (e.g., Table A.1) to disentangle numerator-denominator co-movement.
- Robustness checks: regressions with additional year fixed effects; dynamic panel model with two years lag of the dependent variable (Arellano and Bond 1991); regressions accounting for historical shock (last year). Results in Appendix B consistent with main findings.

### Instantaneous growth impacts (summary of empirical findings)
- Droughts reduce growth by 1 percentage point globally.
- High-intensity storms decrease growth by 1.6 percentage point globally.
- Floods have limited and localized impact on growth (weaker effect).
- Differential impacts:
  - Droughts reduce growth by 1.4 percentage point in EMDEs relative to AEs.
  - Storms reduce growth by 1.8 percentage point in EMDEs relative to AEs.
  - Storms in AEs show a surprising positive growth impact (potential post-disaster cleanup/reconstruction effects).
- Regional and income-group findings:
  - Growth of SSA countries is similarly dragged as elsewhere; gap between SSA and other-EMDEs is negligible.
  - LICs are impacted more by storms than EMs; droughts and floods affect LICs and EMs similarly.
- Example: Cyclone Idai (March 2019) caused US$ 1.4 billion damage in Mozambique; reconstruction cost estimated at US$ 2.9 billion (PDNA 2019).

### Instantaneous fiscal impacts (summary of empirical findings)
- Droughts:
  - Globally, droughts reduced revenue by 0.5 percent of GDP.
  - In EMDEs, droughts reduced revenue by 0.7 percent of GDP; AEs do not experience any decrease.
  - Absolute magnitude (real US$) decline in revenue due to droughts: about 3.4 percentage points globally and 4.5 percentage points in EMDEs (Table A.1).
  - Public expenditure does not surge in response to droughts; expenditure as percent of GDP remains unchanged after droughts.
  - Primary deficit increases by as much as 1.5 percent of GDP in EMDEs following droughts.
  - Public debt increases by 1.4 percent of GDP in EMDEs following droughts.
- Storms:
  - Fiscal impact is muted for both EMDEs and AEs.
  - Declines in absolute revenue and expenditure marginal; significant for AEs only (Table A.1).
  - Slight increase in revenue and expenditure as percent of GDP observed globally and in EMDEs (as GDP shrinks).
  - In AEs, expenditure as a share of GDP declined by 1 percent after storms.
  - Impact on primary balance and debt negligible for both groups.
- Floods:
  - Fiscal expenditure and revenue do not change much after floods except in AEs where revenue appears to decline.
  - Massive flooding episodes affecting more than 30 percent of a country’s population led to a decrease in public spending by 4.7 percentage points in AEs (Table A.1); as a share of GDP this change is negligible but adds to debt.
  - In EMDEs, neither revenue nor expenditure responds to floods.

### Differential impacts (Table 2 highlights)
- Table 2 reports differential impacts (columns: (1) growth [percentage point change], (2) revenue [% of GDP], (3) expenditure [% of GDP], (4) primary balance [% of GDP], (5) debt [% of GDP]):

- Drought:
  - SSA relative to other-EMDEs:
    - Growth: -0.03
    - Revenue: -0.77
    - Expenditure: -0.44
    - Primary Balance: -0.20
    - Debt: -0.33
    - Observations: 2921 (revenue sample 2881; expenditure 2880; primary balance 2873; debt 2832)
    - R-squared examples: Growth 0.084; Revenue 0.494; Expenditure 0.572; Primary Balance 0.343; Debt 0.829
  - LICs relative to EMs:
    - Growth: 0.02
    - Revenue: -0.28
    - Expenditure: -0.12
    - Primary Balance: -0.06
    - Debt: 0.18
    - Observations: 1549 (revenue 1526; expenditure 1522; primary balance 1520; debt 1498)
    - R-squared examples: Growth 0.027; Revenue 0.465; Expenditure 0.590; Primary Balance 0.187; Debt 0.837

- Storm:
  - SSA relative to other-EMDEs:
    - Growth: 0.33
    - Revenue: 1.36
    - Expenditure: 2.09*** (statistical significance indicated in table)
    - Primary Balance: -0.97
    - Debt: -1.56
  - LICs relative to EMs:
    - Growth: -2.19** (statistical significance indicated)
    - Revenue: 0.50
    - Expenditure: 0.73*
    - Primary Balance: -0.21
    - Debt: 0.41

- Flood:
  - SSA relative to other-EMDEs:
    - Growth: -0.03
    - Revenue: -0.73
    - Expenditure: -0.22
    - Primary Balance: 0.03
    - Debt: -4.07
  - LICs relative to EMs:
    - Growth: 0.15
    - Revenue: -0.55
    - Expenditure: -0.15
    - Primary Balance: 0.16
    - Debt: -2.26

- Notes: Standard errors are reported in parentheses in Table 2 in the source. EMs = Emerging Markets, EMDEs = Emerging Markets and Developing Economies, LICs = Low-Income Countries, SSA = Sub-Saharan Africa.

### Mitigating and aggravating factors
- Official Development Assistance (ODA):
  - EMDEs with larger ODA inflows attenuate adverse growth and fiscal impacts.
  - Countries with limited access to ODA experienced greater growth decreases of 1.9–2.4 percentage points after droughts and storms; countries with better access saw almost negligible reductions (Table A.2).
  - Absence of ODA forces spending from public budgets, increasing non-grant public expenditure, deficits, and debt.
  - Role of ODA for floods is unclear; both revenue and expenditure decline for countries with low ODA.

- Trade openness:
  - Countries reliant on import-export markets experienced larger adverse impacts from climate shocks.
  - Droughts decrease agricultural output and drag growth for export-reliant countries, resulting in higher fiscal deficits and debt (Table A.3).
  - Storms may impose larger growth drags on more export-oriented economies due to logistics and transportation disruptions, leading to higher public expenditure for reconstruction/imports.
  - Floods show no differential growth impact by trade openness, but revenue and expenditure of low-exposure countries decline after floods.

- Governance quality:
  - Better regulatory quality helps contain growth reduction effects of droughts.
  - Countries with below-average regulatory quality faced larger growth reductions and fiscal deterioration after droughts.
  - For storms, growth reduced regardless of regulatory quality; magnitude of decline greater for those with better regulatory quality, while fiscal impact is muted for both groups.
  - For floods, gaps in fiscal impact are minimal; growth of those with better regulatory quality seems to decline strongly (Table A.4).

### Persistence of impacts (local projections)
- Growth persistence:
  - Negative impact on growth is short-lived.
  - Droughts and storms cause negative growth knock-on effects in EMDEs that dissipate within a year.
  - Initial growth impact more adverse in EMDEs but EMDEs recover more quickly and return to higher growth subsequently.
  - Floods have negligible initial impacts; in subsequent years they seem beneficial for growth, especially in AEs, with a positive medium-term impact in both EMDEs and AEs.

- Fiscal persistence:
  - Fiscal impacts tend to last longer than growth impacts and are heterogeneous by country group and disaster type.
  - For EMDEs:
    - Droughts reduce revenue and raise expenditure in the medium term, resulting in persistent primary deficit.
    - Floods depress both revenue and expenditure, leading to a small primary balance surplus.
    - Storms exert a small positive impact on revenue but a much larger reduction in expenditure, giving rise to a sizeable increase in primary surplus.
  - For AEs:
    - Drought has a more positive impact on revenue and expenditure.
    - Flood and storm have more negative impacts compared to EMDEs.
    - On average, natural disasters’ impact on AEs’ primary balance is not statistically significant.

- Debt dynamics:
  - Medium-term debt-to-GDP trajectory of EMDEs does not necessarily worsen after disasters.
  - For droughts and storms, debt-to-GDP ratio starts to decline after three years; for floods, reduction is almost immediate in the first year.
  - Drivers of debt dynamics:
    - Post-disaster GDP rebound raises the denominator and can reduce debt-to-GDP ratio.
    - Primary balance worsens after droughts but improves after floods and storms.
    - Controlled inflation and monetary policy lowering domestic interest rates can reduce interest payments and improve debt dynamics.
  - Empirically, average debt-to-GDP shows signs of decline following droughts, storms, and floods, but heterogeneity across countries is large.

*Source: wpiea2023261-print-pdf - 3. Empirical Approach*

### 5. Conclusions

### 5. Conclusions

### Heterogeneous macroeconomic and fiscal impacts
- The impact of natural disasters on economic growth and fiscal positions is heterogeneous across disaster types and countries.
- Droughts and severe storms reduce growth; floods tend to have mild macroeconomic impacts.
- Negative economic consequences are usually larger for EMDEs than for AEs, but among EMDEs the adverse impact on LICs is not always worse than on EMs.
- The typical expected fiscal response—reduced revenue as economic activity shrinks and elevated expenditure as government responds—does not always materialize.

### Role of fiscal space and transmission channels
- The interplay between economic growth, changes in the tax base, and availability of fiscal space is crucial in determining fiscal outcomes after disasters.
- A muted expenditure response in EMDEs signals the absence of fiscal space, highlighting the tension between the need to react and constraints of financial resources.
- Transmission channels from disasters to fiscal positions vary across countries and are related to other macro variables; a data-driven understanding of these channels is essential.

### Strengthening monitoring, analysis, and preparedness
- Enhancing disaster monitoring and impact analysis is a crucial first step toward preserving macro-fiscal stability.
- A data-driven approach should identify the disaster’s type and severity and map its transmission channels to fiscal positions, recognizing cross-country heterogeneity.

### Building fiscal resilience
- Building fiscal buffers and maintaining ample fiscal space are essential for effective response to natural disaster shocks.
- The documented discord between the need to help and the observed lack of expenditure response underscores the necessity of fiscal capacity that promotes resilience to disasters.
- Traditional policy tools—revenue mobilization, expenditure control, and public financial management—remain highly relevant to strengthen disaster resilience.

### International support and timing
- International support for disaster preparedness and adaptation is instrumental in attenuating the effects of climate-related disasters.
- While natural disasters’ negative impact on growth is short-lived, the fiscal impact tends to linger.
- In countries with limited fiscal space, faster mobilization of international support is needed to help alleviate fiscal burdens and contain scarring effects.

*IMF Working Paper — Fiscal Impacts of Climate Disasters in Developing Countries (Section 5: Conclusions).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023261-print-pdf.pdf_