## wpiea2025046-print-pdf

## Source details

**Canonical URL:** [wpiea2025046-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025046-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025046-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025046-print-pdf.pdf.json)

---

### Scope and objectives
- Focus: impact of natural disasters on macroeconomic outcomes, emphasizing output growth and its components.
- Motivation: natural disasters are a key channel through which climate change affects economic activity; understanding impacts informs adaptation, preparedness, recovery, reconstruction, and international policy coordination.
- Timeframe of empirical data: 1980 and 2019.

### Contributions of the paper
- Extends literature by examining impacts on output components: consumption, investment, exports, imports, and government expenditure.
- Analyzes heterogeneity by three country groups: advanced economies (AEs), non-small-island emerging markets and developing economies (NSI EMDEs), and small-island emerging markets and developing economies (SI EMDEs).
- Evaluates the role of structural and cyclical country characteristics (pre-disaster fiscal space, adaptive capacity, income group, small-island status) and disaster attributes (magnitude of physical damage, disaster type: floods, storms, droughts, other).

### Data and definitions
- Disaster data source: EM-DAT (International Disaster Database, Université Catholique de Louvain).
- Baseline sample restrictions:
  - Years after 1980.
  - Large natural disasters: single-year disasters with total damages exceeding 1 percent of GDP.
  - “Single-year” = starts and ends in the same calendar year; multi-year disasters excluded in baseline.
  - “Non-overlapping” = no other large disasters in the preceding or subsequent two years.
  - Winsorize and drop observations at top and bottom 1 percent of output growth distribution (real annual output growth above 18 percent or below -13.1 percent).
  - Drop data in 2020 and years after COVID-19 pandemic.
- Sample result after cleaning and aggregation: 190 large, single-year, non-overlapping disasters (1980–2019).
- EM-DAT frequency highlights (Total 15,594, 1960–2022):
  - Flood 5,615 (36.01 percent)
  - Storm 4,300 (27.57 percent)
  - Earthquake 1,258 (8.07 percent)
  - Epidemic 1,492 (9.57 percent)
  - Drought 759 (4.87 percent)
  - Wildfire 443 (2.84 percent)
- Sample of Large, Single-year Natural Disasters, 1980-2019 (Total 190):
  - Storm 87 (44.57 percent)
  - Flood 43 (23.37 percent)
  - Earthquake 34 (17.93 percent)
  - Drought 10 (5.43 percent)
- Physical damage as percent of GDP (sample of 190 disasters; summary statistics):
  - AEs (Number of disasters 18): Mean 5.84, Min 1.03, p25 1.46, Median 2.18, p75 2.98, Max 65.73
  - Non-Small-Island EMDEs (Number of disasters 116): Mean 6.79, Min 1.01, p25 1.59, Median 2.74, p75 7.27, Max 127.025
  - Small-Island EMDEs (Number of disasters 56): Mean 17.02, Min 1.097, p25 2.81, Median 5.54, p75 18.28, Max 148.38
  - All countries (190): Mean 9.511, Min 1.011, p25 1.769, Median 2.984, p75 8.148, Max 148.385
- Additional data sources: WEO for macro series (WEO codes: NGDP_RPCH; NI_RPCH; NCP_RPCH; NX_RPCH; NM_RPCH), IMF Commodity Terms of Trade database, ND-GAIN (adaptive capacity), INFORM-RISK (robustness).

### Empirical specification and estimation approach
- Baseline: local-projection specification (à la Jordà, 2005), horizon h = 0, 1, 2 (three-year horizon).
- Key regression features:
  - Dependent variable y_{i,t+h} = growth (real output or components) at t+h.
  - ND_t dummy = 1 if a large, single-year, non-overlapping natural disaster at year t.
  - Controls: commodity price shocks (i,t+h−1), lagged dependent variable y_{i,t−1}, country fixed effects, year fixed effects.
- Interaction specification: ND_t * Characteristics_{i,t−1} to assess heterogeneous effects.

### Key empirical findings — aggregate impacts on output growth
- All countries (179 countries with commodity shock data, ~31-year span):
  - Output growth drops by about 1.3 percent in the year of the disaster (t).
  - Output growth recovers by about 0.8 percent in t+1.
  - Impact at t+2 not statistically significant.
  - Interpretation: temporary impact on growth (returns to baseline at t+2) but permanent level loss because recovery does not fully offset initial decline.
- Commodity price shocks have highly significant effects on growth and fiscal variables.

### Heterogeneity across country groups and components
- Advanced Economies (AEs):
  - Disasters have insignificant impact on aggregate real GDP growth.
  - Government expenditure rises significantly in the disaster year (about 1.8 percent), offsetting declines in private investment.
  - Selected coefficients (Advanced Economies, Table 4):
    - REAL GDP GROWTH Natural Disaster (T): -0.637 [0.675]; (T+1): -0.180 [0.608]; (T+2): 0.051 [0.397]
    - REAL GOVEXP GROWTH Natural Disaster(T): 1.798* [0.938] at T
    - REAL INVESTMENT GROWTH Natural Disaster(T): -3.295* [1.677] at T
- Non-Small-Island EMDEs (NSI EMDEs):
  - Significant negative GDP growth in disaster year.
  - Investment falls substantially in disaster year; government expenditure response is limited.
  - Selected coefficients (Table 5):
    - REAL GDP GROWTH Natural Disaster (T): -1.109*** [0.352] at T; 0.630* [0.363] at T+1; 0.349 [0.360] at T+2
    - REAL INVESTMENT GROWTH Natural Disaster (T): -4.913* [2.637] at T; 7.905** [3.609] at T+1
    - REAL GOVEXP GROWTH Natural Disaster (T): -1.154 [1.226] at T (not statistically significant)
- Small-Island EMDEs (SI EMDEs):
  - Significant negative GDP growth in disaster year and larger recovery at t+1.
  - Exports often bear the bulk of the impact (vulnerability of ports and tourism dependence).
  - Selected coefficients (Table 6):
    - REAL GDP GROWTH Natural Disaster (T): -1.510*** [0.541] at T; 1.398*** [0.407] at T+1
    - REAL GOVEXP GROWTH Natural Disaster (T): -2.283 [2.099] at T; 3.342 [2.631] at T+1
    - REAL EXPORT GROWTH Natural Disaster (T): -5.647 [5.079] at T (not statistically significant)
  - Note: component coverage limited (15–19 SI EMDEs); component findings may not generalize to all small islands.

### Channels and mechanisms
- Transmission channels identified:
  - Destruction of physical assets causing immediate negative effect on output.
  - Disruption to economic activity (notably exports in SI EMDEs).
  - Reconstruction efforts that can raise growth in subsequent year(s).
- Specific mechanisms:
  - Declines in private investment are a major negative channel, especially in NSI EMDEs.
  - Export disruption is a key channel in SI EMDEs, linked to damaged ports and tourism losses.
  - Government expenditure response varies with pre-disaster fiscal space and institutional capacity; larger fiscal space associated with stronger expenditure response.

### Factors driving heterogeneous effects (Section VI empirical highlights)
- Disaster characteristics:
  - Physical damage (percent of GDP): larger damage associated with lower output growth at t and with larger subsequent government expenditure.
  - Disaster type: storms and droughts associated with higher output growth at t+1 relative to floods and other disasters (stronger recovery).
- Country characteristics:
  - Small-island status: ND(t)*small island: -1.324** [0.670] at t (small islands suffer about 1.3 percent lower output growth at disaster year).
  - Income group: ND(t)*LIDC: 1.503* [0.797] at t (LIDCs display higher growth conditional on other factors).
  - Adaptive capacity (ND-GAIN): Adaptive capacity (t-1) 4.634 [3.922] at t; 10.487** [4.820] at t+1; 9.040* [5.082] at t+2.
  - Pre-disaster fiscal balance:
    - ND(t)*Fiscal Balance(t-1): 0.070 [0.111] at t; 0.131* [0.079] at t+1; 0.155** [0.066] at t+2.
    - Fiscal Balance(t-1) strongly associated with GOVEXP growth: 0.416*** [0.109] at t; 0.264*** [0.069] at t+1; 0.133*** [0.042] at t+2.
- Selected interaction coefficients (Table 8 & Table 9, selected):
  - ND(t)*physical damage: -0.031*** [0.007] at t for output growth.
  - ND(t)*storm: 1.602* [0.844] at t+1 for output growth.
  - ND(t)*drought: 2.259** [1.038] at t+1 for output growth.
  - ND(t)*AE: 8.116** [3.766] for GOVEXP growth at t.
  - ND(t)*physical damage: 0.084** [0.037] at t+1 for GOVEXP growth.
  - Commodity price shocks (selected): (t-1) 6.160*** [1.886]; (t) 8.101*** [2.201]; (t+1) 8.998*** [2.080] for GDP growth; and large positive coefficients for GOVEXP growth ((t-1) 14.346** [7.066]; (t) 28.866*** [7.243]; (t+1) 28.437*** [8.003]).

### Robustness checks and alternative specifications
- Excluding commodity price shocks: results very similar to baseline (AEs: insignificant; EMDEs: negative in disaster year with partial recovery).
- Including overlapping and multi-year disasters: impacts remain similar; including overlapping disasters attenuates NSI EMDEs’ recovery at t+1.
  - Large disasters in EM-DAT by length (Total 418): duration 0: 392 (93.78 percent); 1: 15 (3.59 percent); others up to 9 years.
- Alternative disaster selection by share of population affected (large = affecting at least 5 percent of population): sample of 155 large, single-year non-overlapping disasters (including 50 droughts, 38 floods, 47 storms); results similar to baseline.
- Panel quantile regressions (e.g., q = 10 percent):
  - At the 10th-percent conditional quantile, the disaster-year coefficient is negative and statistically significant and slightly more negative than the mean-estimate, indicating increased downside risk (fatter left tail) in the disaster year; subsequent years show possible tightening of the left-tail.

### Conclusion and policy-relevant implications (Section VII highlights)
- Main empirical conclusions:
  - Natural disasters temporarily reduce GDP growth in EMDEs but not in AEs in the disaster year; government expenditure in AEs typically responds quickly and offsets negative effects.
  - Government expenditure in EMDEs is slower to rise after disasters and does not fully compensate for negative output effects.
  - NSI EMDEs: investment is more adversely affected.
  - SI EMDEs: exports (notably tourism) are more affected.
  - Output growth effects are temporary (return to baseline by t+2), but output level effects are permanent because recovery does not fully offset the initial decline.
  - Countries with larger pre-disaster fiscal space can implement greater government expenditure responses and on average achieve higher output growth after disasters.
- Policy-relevant implications emerging from empirical evidence:
  - Pre-disaster fiscal space matters for post-disaster government response and output recovery.
  - Strengthening adaptive capacity and addressing country characteristics (income group, small-island status) can influence heterogeneous outcomes.
  - Managing exposure to commodity price shocks is important because such shocks materially affect both GDP growth and fiscal response.

*Source: wpiea2025046-print-pdf (EM-DAT and IMF staff calculations, as presented in the supplied document).*

### References .............................................................................................................

### wpiea2025046-print-pdf - References .............................................................................................................

### Scope and objectives
- Focus: impact of natural disasters on macroeconomic outcomes, emphasizing output growth and its components.
- Motivation: natural disasters are a key channel through which climate change affects economic activity; understanding impacts informs adaptation, preparedness, recovery, reconstruction, and international policy coordination.
- Timeframe of empirical data: 1980 and 2019.

### Contributions of the paper
- Extends literature by examining impacts on output components: consumption, investment, exports, imports, and government expenditure.
- Analyzes heterogeneity by three country groups: advanced economies (AEs), non-small-island emerging markets and developing economies (NSI EMDEs), and small-island emerging markets and developing economies (SI EMDEs).
- Evaluates the role of structural and cyclical country characteristics (pre-disaster fiscal space, adaptive capacity, income group, small-island status) and disaster attributes (magnitude of physical damage, disaster type: floods, storms, droughts, other).

### Data and definitions
- Focused empirical analysis on large natural disasters: single-year disasters with total damages exceeding 1 percent of GDP.
- Historical and large natural disaster events paired with economic variables for 1980–2019.

### Key empirical findings
- Average aggregate output growth effects:
  - Output growth on average drops by around 1.3 percent in the year of the disaster relative to non-disaster control countries.
  - Output growth recovers in the year immediately following the disaster by about 0.8 percent higher than in the control group.
  - In later subsequent years, no statistical difference in output growth between disaster countries and the control group.
  - Interpretation: temporary impacts on growth rates, but a permanent loss in output level because post-disaster growth recovery does not fully offset the initial decline.
- Heterogeneity across country groups:
  - Advanced economies (AEs): government expenditure rises immediately in the year of the disaster, largely offsetting declines in private investment and mitigating negative effects on output growth; output growth in AEs does not appear to be significantly affected by natural disasters.
  - EMDEs overall: limited government expenditure response following disasters, insufficient to fully compensate negative effects on output growth.
  - NSI EMDEs: investment is mostly adversely affected by natural disasters.
  - SI EMDEs: exports bear the bulk of the impact, reflecting vulnerability of export infrastructure (e.g., ports) and dependence on tourism.
- Role of fiscal space and disaster magnitude:
  - Countries with larger fiscal space (proxied by pre-disaster fiscal balance) tend to increase government expenditure more after disasters.
  - Larger disasters (measured by damage to physical assets) on average prompt a larger government expenditure response one year later.

### Channels and mechanisms highlighted
- Transmission channels include declines in private investment, export disruptions (notably in SI EMDEs), and variable government expenditure responses depending on fiscal space and institutional capacity.
- Export impacts in SI EMDEs linked to damage/disruption of ports and tourism-dependence.

### Relation to existing literature
- Mixed findings in prior studies: examples include persistent effects on GDP per capita (Lian and others, 2022), negligible effects after controlling for political events (Cavallo and others, 2013), and variable disaster-type effects (Cevik and Jalles, 2023a).
- Supporting findings in literature: greater output declines in developing and smaller economies (Noy, 2009); mitigation from disaster preparedness and lower public debt (Bayoumi and others, 2021); worse outcomes in fragile, conflict-affected states (Jaramillo et al., 2023); corruption linked to higher disaster-related deaths (Cevik and Jalles, 2023b); pre-disaster institutions aiding long-term recovery (Barone and Mocetti, 2014).
- Paper also connects to literature on natural disasters’ effects on inflation, fiscal outcomes, and exchange rates.

### Organization of the paper
- Section II: data.
- Section III: econometric approach.
- Section IV: empirical results by country group.
- Section V: robustness checks.

*Source: wpiea2025046-print-pdf - References (I. Introduction, pages 5–6).*

### Section VI examines potential ex-ante country and disaster characteristics associated with macroeconomic

### wpiea2025046-print-pdf - Section VI examines potential ex-ante country and disaster characteristics associated with macroeconomic

### Data: natural disasters and other series
- Source of disaster data: The International Disaster Database (EM-DAT) hosted by Université Catholique de Louvain. EM-DAT records over 15,500 natural disasters between 1960 until 2022.
- Natural disaster types in EM-DAT include: earthquake, mass movement (dry), volcanic activity, extreme temperature, fog, storm, flood, landslide, wave action, drought, glacial lake outburst, wildfire, epidemic, insect infestation, and animal accident.
- EM-DAT fields used: country, disaster type, start year and end year, number of deaths, number of people affected, and estimated damages (damage to property, crops, and livestock in US dollars; asset damages).
- Baseline sample restrictions:
  - Years after 1980.
  - Focus on non-overlapping, single-year, large natural disasters. “Large” = damages exceeding 1 percent of national GDP.
  - “Single-year” = starts and ends in the same calendar year; multi-year disasters (e.g., droughts spanning years) excluded in baseline.
  - “Non-overlapping” = no other large disasters in the preceding or subsequent two years.
  - Winsorize and drop observations at top and bottom 1 percent of output growth distribution (real annual output growth above 18 percent or below -13.1 percent).
  - Drop data in 2020 and years after COVID-19 pandemic.
- Sample result after cleaning and aggregation of simultaneous large disasters: 190 large, single-year, non-overlapping disasters (1980–2019).
- Table 1 (EM-DAT, 1960-2022) frequency highlights (Total 15,594):
  - Flood 5,615 (36.01 percent)
  - Storm 4,300 (27.57 percent)
  - Earthquake 1,258 (8.07 percent)
  - Epidemic 1,492 (9.57 percent)
  - Drought 759 (4.87 percent)
  - Wildfire 443 (2.84 percent)
  - Others and small-frequency types reported in EM-DAT.
- Table 2 (Sample of Large, Single-year Natural Disasters, 1980-2019; Total 190):
  - Storm 87 (44.57 percent)
  - Flood 43 (23.37 percent)
  - Earthquake 34 (17.93 percent)
  - Drought 10 (5.43 percent)
- Table 3 (Physical damage as percent of GDP; summary statistics for sample of 190 disasters):
  - AEs (Number of disasters 18): Mean 5.84, Min 1.03, p25 1.46, Median 2.18, p75 2.98, Max 65.73
  - Non-Small-Island EMDEs (Number of disasters 116): Mean 6.79, Min 1.01, p25 1.59, Median 2.74, p75 7.27, Max 127.025
  - Small-Island EMDEs (Number of disasters 56): Mean 17.02, Min 1.097, p25 2.81, Median 5.54, p75 18.28, Max 148.38
  - All countries (190): Mean 9.511, Min 1.011, p25 1.769, Median 2.984, p75 8.148, Max 148.385
- Other data sources:
  - Macroeconomic series (real output, real investment, real private consumption, real exports, real imports, GDP, government revenue and expenditure) from World Economic Outlook (WEO). WEO codes: NGDP_RPCH; NI_RPCH; NCP_RPCH; NX_RPCH; NM_RPCH.
  - Commodity export price shocks from IMF’s Commodity Terms of Trade database (index constructed from 40 commodities; formula provided in text).
  - Adaptive capacity indexes: ND-GAIN (available 1995–2020 for 176 countries; for 1980–1994 the 1995 value is assigned) and INFORM-RISK (2013–2022; used as robustness check).

### Empirical specification and estimation approach
- Baseline local-projection specification (à la Jordà, 2005), horizon h = 0, 1, 2 (three-year horizon). Regression equation (1) in text:
  - Dependent variable y_{i,t+h} = growth (real output or components) at t+h.
  - ND_t dummy = 1 if a large, single-year, non-overlapping natural disaster at year t.
  - Controls: commodity price shocks (i,t+h−1), lagged dependent variable y_{i,t−1}, country fixed effects, year fixed effects.
- Interaction specification (equation (2)) adds interactions: ND_t * Characteristics_{i,t−1} to assess heterogeneous effects by disaster and country characteristics.
- Conceptual channels: destruction (asset damage), disruption to economic activity (negative for GDP growth), and reconstruction efforts (positive for GDP growth).

### Aggregate impacts on output growth (all countries and by income group)
- All countries (179 countries with commodity shock data, ~31-year span):
  - Average impact: output growth drops by about 1.3 percent in the year of the disaster (t) and recovers by about 0.8 percent in t+1. Impact at t+2 not statistically significant.
  - Interpretation: temporary impact on growth (returns to baseline at t+2) but permanent level loss as recovery does not fully offset the initial decline.
  - Commodity price shocks have highly significant effects (discussed in Section VI).
- By income group:
  - Advanced Economies (AEs): disasters have insignificant impact on aggregate real GDP growth.
    - Mechanism: government expenditure rises significantly in the disaster year (about 1.8 percent), offsetting decline in investment.
    - Table 4 (selected coefficients, Advanced Economies; observations and country counts preserved):
      - REAL GDP GROWTH Natural Disaster (T): -0.637 [0.675] (T), -0.180 [0.608] (T+1), 0.051 [0.397] (T+2)
      - REAL GOVEXP GROWTH Natural Disaster(T): 1.798* [0.938] at T
      - REAL INVESTMENT GROWTH Natural Disaster(T): -3.295* [1.677] at T
  - Non-Small-Island EMDEs (NSI EMDEs): significant negative GDP growth in disaster year.
    - Table 5 (selected coefficients):
      - REAL GDP GROWTH Natural Disaster (T): -1.109*** [0.352] at T; 0.630* [0.363] at T+1; 0.349 [0.360] at T+2
      - REAL INVESTMENT GROWTH Natural Disaster (T): -4.913* [2.637] at T; 7.905** [3.609] at T+1
      - REAL GOVEXP GROWTH Natural Disaster (T): -1.154 [1.226] at T (drop not statistically significant)
    - Mechanism: investment falls substantially in disaster year; government expenditure does not sufficiently offset fall.
  - Small-Island EMDEs (SI EMDEs): significant negative GDP growth in disaster year and larger recovery at t+1.
    - Table 6 (selected coefficients):
      - REAL GDP GROWTH Natural Disaster (T): -1.510*** [0.541] at T; 1.398*** [0.407] at T+1
      - REAL GOVEXP GROWTH Natural Disaster (T): -2.283 [2.099] at T; 3.342 [2.631] at T+1
      - REAL EXPORT GROWTH Natural Disaster (T): -5.647 [5.079] at T (not statistically significant)
      - Note: data coverage for components limited (15–19 SI EMDEs), so component findings may not generalize to all small islands.
    - Mechanism: exports (often tourism) decline and government expenditure response lags; investment and imports may rise immediately in some cases.

### Robustness checks and alternative specifications
- Robustness Check 1: Excluding export commodity price shocks — results very similar to baseline. AEs: insignificant effects; EMDEs: negative in disaster year with partial recovery.
- Robustness Check 2: Including overlapping and multi-year disasters — impacts remain similar to baseline; when including overlapping disasters, NSI EMDEs show no longer statistically significant recovery at t+1.
  - Table 7 (Large disasters in EM-DAT by length; Total 418): duration 0: 392 (93.78 percent); 1: 15 (3.59 percent); others up to 9 years.
- Robustness Check 3: Alternative selection of disasters by share of population affected (large = affecting at least 5 percent of population). Sample of 155 large, single-year non-overlapping disasters (including 50 droughts, 38 floods, 47 storms). Results similar to baseline.
- Alternative specification: Panel quantile regressions (e.g., q = 10 percent) to assess downside risk.
  - Findings: At the 10th-percent conditional quantile, the disaster-year coefficient is negative and statistically significant and slightly more negative than mean-estimate (local projection). This implies natural disasters increase downside risk (fatter left tail) in the disaster year; in subsequent years the left-tail may tighten.

### Section VI: Factors driving heterogeneous effects (empirical highlights)
- Disaster characteristics considered:
  - Economic damage (percent of GDP): ambiguous theoretical sign; empirically associated with larger disruptions at t but also larger reconstruction needs.
  - Disaster type: droughts, storms, floods (others include earthquakes, volcanos, wildfires).
- Country characteristics considered:
  - Income group (AEs, EMMIEs, LIDCs), small-island status, adaptive capacity (ND-GAIN), pre-disaster fiscal space (pre-disaster overall fiscal balance in percent of GDP).
- Table 8 (Heterogeneous Growth Impacts of Natural Disasters; selected results, standard errors in brackets):
  - Commodity price shocks significant: Commodity price shocks (t-1) 6.160*** [1.886]; (t) 8.101*** [2.201]; (t+1) 8.998*** [2.080].
  - GDP growth (t-1) coefficients: 0.299*** [0.027] (t), 0.106*** [0.023] (t+1), 0.070*** [0.022] (t+2).
  - Natural Disaster (ND) interactions:
    - ND(t)*physical damage: -0.031*** [0.007] at t; 0.018 [0.018] at t+1; -0.032 [0.021] at t+2.
    - ND(t)*storm: 0.040 [0.705] at t; 1.602* [0.844] at t+1.
    - ND(t)*drought: -0.489 [0.995] at t; 2.259** [1.038] at t+1.
    - ND(t)*small island: -1.324** [0.670] at t.
    - ND(t)*LIDC: 1.503* [0.797] at t.
    - ND(t)*Fiscal Balance(t-1): 0.070 [0.111] at t; 0.131* [0.079] at t+1; 0.155** [0.066] at t+2.
  - Adaptive capacity (t-1): Adaptive capacity (t-1) 4.634 [3.922] at t; 10.487** [4.820] at t+1; 9.040* [5.082] at t+2.
  - Sample: Observations 4,321 (t), 4,151 (t+1), 3,984 (t+2); Number of countries 172.
- Table 9 (Heterogeneous Impacts on Government Expenditure Growth; selected results):
  - Commodity price shocks large and significant for GOVEXP growth: (t-1) 14.346** [7.066]; (t) 28.866*** [7.243]; (t+1) 28.437*** [8.003].
  - Fiscal Balance(t-1) strong positive association with GOVEXP growth: 0.416*** [0.109] at t; 0.264*** [0.069] at t+1; 0.133*** [0.042] at t+2.
  - ND(t)*physical damage: 0.084** [0.037] at t+1 (larger physical damage associated with higher GOVEXP growth at t+1).
  - ND(t)*AE: 8.116** [3.766] for GOVEXP growth at t (AEs respond with higher govexp growth).
  - ND(t)*Fiscal Balance(t-1): 0.796*** [0.304] for GOVEXP growth at t+2 (interaction significant).
- Synthesis of Tables 8 and 9 (text summary preserved):
  - Physical damage: larger physical damage causes significantly lower output growth in the year of disaster (t=0).
  - Disaster types: single-year storms and droughts are associated with higher output growth than floods and other disasters at t+1, indicating stronger recovery after storms and droughts.
  - Larger physical damage associated with significantly higher government expenditure growth at t+1.
  - LIDCs display higher growth than EMMIEs and AEs when holding other factors constant (possible role of grants and donor funding).
  - AEs have higher government expenditure growth than EMMIEs and LIDCs (about 8.1 percent), holding other factors constant.
  - Small islands suffer lower output growth at year of disaster (about 1.3 percent lower) compared to non-small islands.
  - Better adaptive capacity (smaller ND-GAIN index) associated with higher growth after disasters, though not always statistically significant.
  - Higher pre-disaster fiscal balance is associated with higher output growth (statistically significant at t+1 and t+2) and enables larger government expenditure response when disasters occur.
  - Commodity price shocks have highly significant impacts on GDP growth and government expenditure growth.

### Conclusion (Section VII highlights)
- Main findings:
  - Natural disasters temporarily reduce GDP growth in EMDEs but not in AEs in the disaster year; government expenditure in AEs typically responds quickly and offsets negative effects.
  - Government expenditure in EMDEs is slower to rise after disasters and does not fully compensate for negative output effects.
  - In NSI EMDEs, investment is more adversely affected; in SI EMDEs, exports (notably tourism) are more affected.
  - Output growth effects are temporary (return to baseline by t+2), but output level effects are permanent because recovery does not fully offset the initial decline.
  - Countries with larger pre-disaster fiscal space can implement greater government expenditure responses and on average achieve higher output growth after disasters.
- Policy-relevant implications emerging from empirical evidence:
  - Pre-disaster fiscal space matters for post-disaster government response and output recovery.
  - Adaptive capacity and country characteristics (income group, small-island status) influence heterogeneous outcomes.
  - Commodity price shocks materially affect both GDP growth and fiscal response.

*Source: EM-DAT and IMF staff calculations, as presented in the supplied document.*

### References

### References

### Commodity prices, terms of trade, and external debt
- Arezki, Rabah and Markus Brückner. 2012. "Commodity Windfalls, Democracy and External Debt," Economic Journal, vol. 122(561), pp 848-866. 
- Gruss, B. and S. Kebhaj. 2019. “Commodity terms of trade: A new database.” IMF Working Paper No. 19/21.  

### Natural disasters, growth, and scarring effects
- Barone, G., and S. Mocetti, 2014, "Natural Disasters, Growth and Institutions: A Tale of Two Earthquakes," Journal of Urban Economics, 84, 52-66.  
- Cavallo, E., S. Galiani, I. Noy, and J. Pantano, 2013, “Catastrophic Natural Disasters and Economic Growth,” Review of Economic and Statistics 95, 1549–1561.  
- Fomby, Thomas, Yuki Ikeda and Norman V. Loayza, 2013. "The Growth Aftermath Of Natural Disasters," Journal of Applied Econometrics, vol. 28(3), pages 412-434. 
- Lian, Weicheng and Jose Ramon Moran & Raadhika Vishvesh, 2022. "Natural Disasters and Scarring Effects," IMF Working Papers 2022/253,  
- Noy, Ilan, 2009. "The macroeconomic consequences of disasters," Journal of Development Economics, Elsevier, vol. 88(2), pages 221-231 

### Fiscal and macro-fiscal aftermath
- Noy, Ilan and Nualsri, Aekkanush, 2011. "Fiscal storms: public spending and revenues in the aftermath of natural disasters," Environment and Development Economics, vol. 16(1), pages 113-128 
- Gerling, Kerstin (2017), The Macro-Fiscal Aftermath of Weather-Related Disasters: Do Loss Dimensions Matter? IMF Working Paper 2017/235. 

### Climate shocks, inflation, exchange rates, and price dynamics
- Cevik, Serhan and João Tovar Jalles. 2023a. "Eye of the Storm: The Impact of Climate Shocks on Inflation and Growth," IMF Working Papers 2023/087 
- Kabundi, Alain, Montfort Mlachila, and Jiaxiong Yao. 2022. “How Persistent are Climate-Related Price Shocks? Implications for Monetary Policy,” IMF Working Paper 22/207.  
- Hale, Galina. 2022. Climate risks and exchange rates, mimeo. 
- Feng, Alan & Haishi Li and Yulin Wang, 2023. "We Are All in the Same Boat: Cross-Border Spillovers of Climate Shocks through International Trade and Supply Chain," CESifo Working Paper Series 10402, CESifo.  

### Risk, vulnerability indices, and risk management
- Disaster Risk Management Knowledge Centre (DRMKC). 2022. INFORM Risk Index. European Commission. https://drmkc.jrc.ec.europa.eu/inform-index/  INFORM-Risk  

### Corruption, governance, and disaster outcomes
- Cevik Serhan and João Tovar Jalles. 2023b. Corruption Kills: Global Evidence from Natural Disasters.” IMF Working Paper No. 2023/220 

### Methodology and econometrics
- Jordà, Òscar, 2005. "Estimation and Inference of Impulse Responses by Local Projections," American Economic Review, vol. 95(1), pages 161-182, March.  

### Climate assessment reports and policy notes
- IPCC. 2014. Intergovernmental Panel on Climate Change’s Fifth Assessment Report  
- IPCC. 2023. Intergovernmental Panel on Climate Change’s Sixth Assessment Report  
- Jaramillo, Laura, Aliona Cebotari, Yoro Diallo, Rhea Gupta, Yugo Koshima, Chandana Kularatne, Daniel Jeong Dae Lee, Sidra Rehman, Kalin Tintchev, and Fang Yang. 2023. “Climate Challenges in Fragile and Conflict-Affected States.” IMF Staff Climate Note 2023/001, International Monetary Fund, Washington, DC. 

### Cross-border risk sharing and macroeconomic recovery
- von Peter, Goetz,, Sebastian von Dahlen and Sweta Saxena. 2024. "Unmitigated disasters? Risk sharing and macroeconomic recovery in a large international panel," Journal of International Economics, Elsevier, vol. 149(C).  

### Working papers on growth at risk from disasters
- Bayoumi, Tamim and Quayyum, Saad and Das, Sibabrata, 2021. Growth at Risk from Natural Disasters. IMF Working Paper No. 2021/234  

*Understanding the Macroeconomic Effects of Natural Disasters — Working Paper No. WP/2025/046*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025046-print-pdf.pdf_