## wpiea2021013-print-pdf

## Source details

**Canonical URL:** [wpiea2021013-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021013-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021013-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021013-print-pdf.pdf.json)

---

### Main contributions and findings
- Demonstrates that foreign climatic disasters and foreign climate risk exposures affect aggregate stock market valuations at the country and country-sector level and have macrofinancial implications.
- Shows heterogeneous sectoral stock returns in response to foreign climatic disasters; identification strengthened by placebo tests on non-major trading partners and by examining tradability.
- Finds that financial sector capitalization and international trade guarantees influence the impact of foreign climate shocks on domestic financial sector health.
- Argues that climate change adaptation (enhancing resilience of factories, roads, and ports) yields significantly positive externalities on other countries by reducing negative spillovers through global trade and supply-chain networks.
- Suggests that optimal global adaptation likely requires collective action in a multilateral framework because of documented cross-country spillovers.

### Policy relevance and recommendations
- Climate change adaptation programs can produce substantial cross-border monetary benefits and may improve debt sustainability if partly or fully funded by trade partners.
- Rationale to address climate change adaptation in a multilateral framework because helping trade partners build resilience also enhances home countries’ climate resilience.
- Framework complements central banks’ and financial regulators’ analytical agendas (for example, NGFS) on climate–financial stability links; applicable to transition risk by replacing climatic disasters with climate transition shocks and incorporating sectoral carbon intensity in bilateral trade relationships.

### Data, sample, and key variable construction
- Sample: 68 countries (34 advanced, 34 emerging markets and developing economies).
- Sample period: 1970 to 2018. Note: UN Comtrade trade data available until 2018; climate disaster and stock price data used by the GFSR cover 1970–2019, hence study sample ends in 2018.
- Trade and national accounts:
  - Country-bilateral total trade from United Nations Comtrade Database.
  - Annual GDP from United Nations National Account Database.
- Value-added and gross output:
  - Value-added to gross output ratio (푉퐴푆푖,푦) from Johnson and Noguera (1970–2004) and OECD AAMNE (2005–2016); missing values approximated by category-year averages.
  - Gross output computed as 푌푖,푦 = 푉퐴푖,푦 / 푉퐴푆푖,푦; home sales computed as 푥푖푖,푦 = 푌푖,푦 − ∑_{j≠i} 푥_{j i, y}.
  - Total expenditure: 푋_{n, y} = ∑_{i=1}^N 푥_{n i, y}.
- Trade-share measures:
  - Expenditure share: 휋_{n i, y} = 푥_{n i, y} / 푋_{n, y}, with ∑_{i=1}^N 휋_{n i, y} = 1 for a buying country.
  - Output share: 푆_{n i, y} = 푥_{n i, y} / 푌_{i, y}, with ∑_{n=1}^N 푆_{n i, y} = 1 for a selling country.
- Climate and disaster data:
  - Disaster data: Emergency Events Database (EM-DAT).
  - Disaster selection: require exact start date; restricted to affected population rate > "0.5 percent" or damage > "0.05 percent of GDP".
  - Disaster damage metrics: total deaths, total people affected, total monetary loss.
  - Examples: Hurricane Katrina (2005) monetary damage "$125 billion"; 2011 Thai floods monetary damage relative to host GDP "10.1 percent".
  - Sample averages: average disaster causes "$783 million" monetary damage in current USD and "113" deaths, and affects "1.36 million" people. On average, monetary damage is "0.01 percent of the hit country’s GDP".
- Climate risk index:
  - Verisk Maplecroft Climate Change Exposure Index (raw 0 highest risk to 10 lowest risk); hazard index = 10 − raw index, then normalized to mean "0" and standard deviation "1".
  - Fixed country climate risk to value in "2018". Verisk Maplecroft data availability "2013–19".
- Financial and sectoral data:
  - Stock market variables and three-month government bond yields from Refinitiv Datastream.
  - Sector tradability from WIOD 2016 release (covers "2000–16"), using "2000" as benchmark year.
- Data treatment: all variables winsorized at top and bottom "1 percent".

### Modeling assumptions and identification
- Bilateral trade responses to foreign climate shocks follow gravity model predictions: downstream countries import less from upstream foreign countries hit by disasters; upstream countries export less to downstream countries hit by disasters.
- Assumptions:
  - Globalization treated as exogenous to climate change.
  - Expenditure shares (π_{ni,y}) not affected by climatic disasters in downstream country n in year y.
  - Output shares (S_{ni,y}) not affected by climatic disasters in downstream country i in year y.
  - Upstream disasters reduce upstream supplies proportionally to annual output shares; downstream disasters reduce purchases proportionally to annual expenditure shares.
- Identification strengthened with placebo tests on non-major trading partners and examination of tradability.
- Limitation noted: deliberate decoupling or large departures from gravity model could lead the model to exaggerate negative impacts of foreign climate risks.

### Construction of exposure measures and normalization
- Upstream exposure: U_{n,y} = ∑_{i=1}^N S_{n i, y} Damage_{i,y}.
- Downstream exposure: D_{n,y} = ∑_{j=1}^N π_{j n, y} Damage_{j,y}.
- Normalized exposures:
  - Nor_U_{n,y} = ∑ S_{n i, y} Damage_{i,y} / ∑ Damage_{i,y}.
  - Nor_D_{n,y} = ∑ π_{j n, y} Damage_{j,y} / ∑ Damage_{j,y}.
  - Cross-country sums of normalized exposures equal total world disaster damages.
- Benchmarking:
  - Trade shares fixed at benchmark year y0; benchmark set to "t0 = 1971".
  - Relative exposure ratios for country groups defined, e.g., Rel_U_{A,y0,y} = ∑_{n∈A} Nor_U_{n,y0,y} / ∑_{n∈A} Nor_U_{n,y0,y0}.

### Global trade trends and exposure evolution (key statistics)
- World trade-to-world GDP ratio:
  - "0.07 in 1970"; rose sharply in the 1970s, flattened in the 1980s, returned to strong upward trajectory in the 1990s and first half of the 2000s.
  - The 2007-2009 Great Recession broke the trend: the ratio fell by "20 percent" (the “Great Trade Collapse”).
  - The ratio bounced back after the Great Recession but never reached pre-recession peak of "about 0.25".
  - "As of 2018, the world trade-to-world GDP ratio was about 0.20".
- Service trade: UN Comtrade covers only goods; service trade accounts for "about 22 percent of world total trade as of 2018".
- Exposure evolution for low home-country shock group (bottom 10 percent by home-country annual deaths or hazard index):
  - Upstream exposure to total climatic disaster-related deaths increased by "about 25 times".
  - Downstream exposure to total climatic disaster-related deaths increased by "about 35 times".
  - Upstream exposure to climate risks increased by "about 8 percent"; downstream exposure increased by "about 4 percent".
  - Global exposures for the low-shock group started to fall approximately at the time of the Great Recession.
- Exposure evolution for high home-country shock group (top 10 percent by home-country annual deaths or hazard index):
  - Upstream exposure to global climatic disasters dropped by "about 8 percent".
  - Downstream exposure dropped by "about 6 percent".
  - Upstream exposure to global climate risks dropped by "about 20 percent"; downstream by "about 10 percent".
  - Around the Great Recession, high home-country shock countries began to increase their exposures.
- Annual average changes during 1971–2018 (country-level):
  - Country a (low-shock) experienced "0.34" and "0.54" annual average increase in upstream and downstream exposures of global climate disaster-related deaths, and "0.15" and "0.06" percentage points increase in upstream and downstream exposures of climate risks.
  - Country b (high-shock) experienced "0.07" and "0.08" annual average decline in upstream and downstream exposures of global climate disaster-related deaths, and "0.24" and "0.17" percentage points increase in upstream and downstream exposures of climate risks.
- Interpretation: globalization increased exposure of previously low-risk countries via trade linkages and reduced exposure of previously high-risk countries; trade reallocates disaster incidences across countries without changing world total damages.

### Global spillover index and drivers decomposition
- Spillover index:
  - Spillover_up_{y0,y} = sd_n(Nor_U_{n,y0,y}) / sd_n(Nor_U_{n,y,y}); analogous for downstream.
  - Index equals "1" in benchmark year y = y0; rising index indicates reduced cross-country dispersion due to trade openness.
- Empirical relationship: time series of spillover indexes closely follow world trade-to-world GDP ratio; deepening globalization increases cross-border spillovers and reduces dispersion of exposures.
- Decomposition (1971–2018):
  - Openness to trade alone explains "87.4 percent" (upstream) and "83.7 percent" (downstream) of the decline in cross-country dispersion in exposures to global climate disaster damages.
  - Changes in geographical distribution of climate disasters alone explain "35.9 percent" (upstream) and "29.7 percent" (downstream) of the decline.
  - From "1971 to 2008", cross-country dispersion in global climate disaster damage exposures declined by "6.3%" for upstream shocks and by "10.7%" for downstream shocks, relative to 1971.

### Event study: equity market responses to foreign climatic disasters (design and aggregate results)
- Linking disasters to most-affected partners:
  - For disaster d hitting country i(d):
    - Largest export destination: n_D(i(d)) = argmax_{n≠i} S_{n i(d), y(d)}.
    - Largest import origin: n_U(i(d)) = argmax_{n≠i} π_{i(d) n, y(d)}.
  - On average, a country sells "about 3.1%" of its gross output to its largest exporting destination and spends "about 2.7 percent" of its total expenditure on its largest importing source.
- Event window and return measurement:
  - Expected returns estimated on a window starting "12 months" before disaster and ending "one month" before disaster.
  - Event window: from "21" trading days before disaster to "60" trading days after disaster.
  - x-day cumulative abnormal return (CAR) computed and averaged across disasters; 95 percent confidence intervals reported.
- Aggregate market-level findings:
  - Aggregate market-level cumulative abnormal returns to upstream and downstream climatic disasters (from -21 trading days to +40 trading days) were both "about -0.5%" and were marginally significant at "95 percent" confidence interval around "35" trading days after disaster start dates.
  - Magnitude comparable to home-country disaster’s average impact on stock market (about "-1 percent") reported in the GFSR.

### Sectoral heterogeneity, tradability, and CAR magnitudes
- Sector responses:
  - Tradable sectors respond more strongly to foreign climate shocks; non-tradable sectors show little response.
  - Examples of sector CARs:
    - Chemicals: "-0.8 percent" for upstream disasters and "-1 percent" for downstream disasters.
    - Automobiles: "-1.8 percent" for upstream disasters and "-1.5 percent" for downstream disasters.
    - Media and telecommunication sectors do not respond significantly to foreign disasters.
- Placebo tests: the disaster-hit country’s 35th largest (median) exporting or importing partner does not show significant stock market response to the disaster, supporting importance of top trade linkages.
- Regression evidence (pooled and interaction results):
  - Pooled level coefficients (without interactions):
    - nor_up_damage coefficient: -47.20** (standard error 22.43).
    - nor_down_damage coefficient: -32.67* (standard error 16.37).
  - With tradability interactions:
    - nor_up_damage * TDIM_s = -246.5*** (standard error 56.28).
    - nor_down_damage * TDEX_s = -83.74*** (standard error 18.96).
  - Level effects become insignificant when interactions included—tradable sectors drive negative CAR effects.
- Implied magnitudes (for a 0.1 percent of home-country GDP increase in exposure):
  - Upstream exposure: predicted CAR reductions of:
    - "15.4 percent" in sector with highest importing tradability (chemicals).
    - "4.1 percent" in sector with median importing tradability (food and beverages).
    - "0.98 percent" in sector with lowest importing tradability (real estate).
  - Downstream exposure: predicted CAR reductions of:
    - "9.5 percent" in sector with highest exporting tradability (automobile).
    - "2.2 percent" in sector with median exporting tradability (media).
    - "0.92 percent" in sector with lowest exporting tradability (real estate).

### Regression estimates: upstream and downstream disaster impact on market CARs
- Upstream disasters:
  - An increase in exposure to foreign upstream climatic disaster damage by 0.1 percent of the home country’s GDP is associated with a "5.9 percent" decline in the market-wide trading day 40’s cumulative abnormal return.
  - Tradable sectors (automobile, basic materials, chemicals, food and beverages, food producers, industrial goods, industrial producers) show negative and significant responses; some sectors (automobile and chemicals) see more than "10 percent" decline for a 0.1 percent of GDP foreign upstream damage.
  - Most non-tradable sectors not significantly affected.
- Downstream disasters:
  - An increase in exposures to downstream foreign climatic disaster damage by 0.1 percent of home-country GDP is associated with a "3.7 percent" decline in trading day 40’s market-level cumulative abnormal return.
  - Typical tradable sectors respond more negatively and significantly than most non-tradable sectors.
- Pooled partial result (text truncated in source): reported partial result starts "An increase in upstream and downstream exposure by 0.1 percent of home-country GDP is associated with 4.7 percent and" (full downstream pooled estimate truncated).

### Financial sector heterogeneity and institutional buffers
- Financial-sector valuation:
  - Home-country climatic disasters reduce valuation of financial sector stocks; foreign disasters undermine financial stability indirectly via tradable-sector channel.
- Bank capitalization:
  - One percentage point increase in bank regulatory capital-to-risk-weighted assets ratio is associated with about a "0.2 percent" increase in the cumulative abnormal return from upstream and downstream disasters.
- Factoring activity:
  - factoring_to_gdp (%) coefficient ~ 0.00108** and 0.00105** in regressions — positive association holding normalized damage fixed.
- Implication: stronger bank capitalization and higher factoring activity mitigate negative CAR responses in the financial sector to foreign climatic disasters.

### Equity pricing of foreign climate physical risk (P/E analysis)
- Cross-section (year fixed at "2018") pooled results:
  - foreign_exp (upstream pooled) coefficient: -43.04*** (standard error 15.11).
  - foreign_exp (downstream pooled) coefficient: -43.94** (standard error 20.06).
  - One standard deviation increase in upstream or downstream exposures corresponds to about 0.05 standard deviation decline in P/E; inter-quartile increase associated with P/E reductions of about "3.0" (upstream) and "3.7" (downstream).
- Sector heterogeneity:
  - Industrial producers (tradable) show strong negative correlations; real estate (non-tradable) shows no significant correlation.
- Interaction with tradability:
  - Adding TDIM_s * U_i removes level significance of foreign_exp — tradable sectors drive negative association.
  - Examples for upstream U_i (Table 8, Column 7):
    - 50th percentile importing tradability (food and beverages): one standard deviation increase in U_i → "0.0488" standard deviation decline in P/E.
    - 25th percentile (travel and leisure): "0.0286" standard deviation decline.
    - 75th percentile (industrial producers): "0.0742" standard deviation decline.
  - For downstream D_i (Table 8, Column 8):
    - 25th percentile exporting tradability (insurance): "0.0075" standard deviation decline.
    - 50th percentile (media): "0.0169" standard deviation decline.
    - 75th percentile (industrial producers): "0.1066" standard deviation decline.
- Placebo tests:
  - Placebo upstream/downstream foreign exposures (Ũ_i, D̃_i) are not significantly correlated with home P/E ratios; interactions of placebo exposures with tradability also not significant — openness alone does not explain negative P/E association.

### Key conclusions, implications, and caveats
- Foreign climatic disasters generate negative asset-price spillovers to aggregate markets and especially to tradable sectors in partner countries via trade and supply-chain linkages.
- Tradable-sector exposure is the main propagation channel; non-tradable sectors show little or no direct asset-price sensitivity to foreign climate risk in these results.
- Bank capitalization and factoring markets can cushion the financial-sector valuation impact from foreign climatic disasters.
- Policy implications:
  - Enhancing resilience through adaptation is a common international responsibility because trade transmits climate risk across borders.
  - Collective international policy action complements domestic adaptation and cross-border economic incentives.
  - Analytical framework applicable to transition risk, other crises (for example, COVID-19), and other globalization channels (multinational production, remittances, tourism).
- Caveats and further research:
  - Financial stability implications depend on country-specific factors (size of tradable sectors, bank exposures); quantifying country-specific stability effects requires further calibrated modeling.
  - Study links each disaster to the single largest affected partner and provides a partial-equilibrium upper-bound estimate of foreign spillovers; aggregation across all trading partners is left to future work.
  - More granular data could test additional global supply chain characteristics (for example, input-specificity) and firm- or region-level spillovers.

*Source: wpiea2021013-print-pdf - 0.05 percent of GDP.*

### 0.05 percent of GDP.

### wpiea2021013-print-pdf - 0.05 percent of GDP.

### Main contributions and findings
- Demonstrates that foreign climatic disasters and foreign climate risk exposures affect aggregate stock market valuations at the country and country-sector level and have macrofinancial implications.
- Shows heterogeneous sectoral stock returns in response to foreign climatic disasters; identification strengthened by placebo tests on non-major trading partners and by examining tradability.
- Finds that financial sector capitalization and international trade guarantees influence the impact of foreign climate shocks on domestic financial sector health.
- Argues that climate change adaptation (enhancing resilience of factories, roads, and ports) yields significantly positive externalities on other countries by reducing negative spillovers through global trade and supply-chain networks.
- Suggests that optimal global adaptation likely requires collective action in a multilateral framework because of documented cross-country spillovers.

### Policy relevance and recommendations
- Climate change adaptation programs can produce substantial cross-border monetary benefits and may improve debt sustainability if partly or fully funded by trade partners.
- There is a rationale to address climate change adaptation in a multilateral framework because helping trade partners build resilience also enhances home countries’ climate resilience.
- The paper complements central banks’ and financial regulators’ analytical agendas (for example, NGFS) on climate–financial stability links and notes the conceptual framework is applicable to transition risk analysis by replacing climatic disasters with climate transition shocks and incorporating sectoral carbon intensity in bilateral trade relationships.

### Relation to prior literature and empirical context
- Builds on literature showing extreme climatic events lower GDP and have persistent negative effects:
  - Hsiang (2010): climatic disasters lower overall GDP with severe negative effects on agriculture, retail, and tourism.
  - Anttila-Hughes and Hsiang (2013): the average typhoon reduces household income by 6.6 percent in the short run; losses persist for several years.
  - Hsiang and Jina (2014): negative and persistent effect of climatic disasters on GDP growth rates.
  - Strobl (2011): a hurricane in the US on average lowers the annual growth rate of per capita income in the local county by 0.45 percent.
  - Dell, Jones, and Olken (2012) and Burke, Hsiang, and Miguel (2015): higher temperatures on average negatively affect economic growth, especially for low-income countries.
  - Kahn, Mohaddes, Ng, Pesaran, Raissi, and Yang (2019): climate change reduces growth in real GDP per capita in the long run.
- Asset-price impacts:
  - Chapter 5 of the IMF Global Financial Stability Report (GFSR April 2020): following a large climatic disaster, domestic banking and insurance sectors lose on average about 1–2 percent in stock market valuation.
  - Firm- and sector-level studies (e.g., Addoum, Ng, and Ortiz-Bobea (2018); Hong, Li, and Xu (2019)) show temperature and drought indices predict sector returns and firm earnings in some industries.
- Extends trade-propagation literature by focusing on cross-border spillovers via international trade and input-output linkages, differentiating from works that study within-country or firm-level transmission.

### Data, sample, and key variable construction
- Sample: same set of 68 countries as in the GFSR; 34 countries are advanced, and 34 countries are emerging markets and developing economies.
- Sample period: 1970 to 2018.
- Uses country-bilateral total trade from the United Nations Comtrade Database and annual GDP from the United Nations National Account Database.
- Value-added to gross output ratio (푉퐴푆푖,푦) sourced from Johnson and Noguera (1970–2004) and OECD AAMNE (2005–2016); missing values approximated by category-year averages.
- Gross output computed as 푌푖,푦 = 푉퐴푖,푦 / 푉퐴푆푖,푦; home sales computed as 푥푖푖,푦 = 푌푖,푦 − ∑푗≠푖 푥푗푖,푦.
- Total expenditure: 푋푛,푦 = ∑푖=1푁 푥푛푖,푦.
- Defines two key trade-share measures:
  - Expenditure share: 휋푛푖,푦 = 푥푛푖,푦 / 푋푛,푦, with ∑푖=1푁 휋푛푖,푦 = 1 for a given buying country.
  - Output share: 푆푛푖,푦 = 푥푛푖,푦 / 푌푖,푦, with ∑푛=1푁 푆푛푖,푦 = 1 for a given selling country.
- Note: international trade data from UN Comtrade are available until 2018; climate disaster and stock price data used by the GFSR cover 1970–2019, hence this study’s sample ends in 2018.

### Methodology and identification
- Assumes bilateral trade responses to foreign climate shocks follow predictions of the gravity model of international trade; this allows downstream countries to import less from upstream foreign countries hit by climatic disasters and upstream countries to export less to downstream countries hit by disasters.
- Acknowledges limitation: if a country deliberately decouples from a foreign country due to perceived higher climate risks or adjusts trade in ways that drastically depart from the gravity model, the model may exaggerate the negative impact of foreign climate risks.
- Identification strengthened through placebo tests on non-major trading partners and examination of tradability to explain sectoral heterogeneity.

*Source: wpiea2021013-print-pdf - 0.05 percent of GDP.*

### 0.07 in 1970, but it rose sharply in the 1970s, flattened in the 1980s, and returned to a strong

### wpiea2021013-print-pdf - 0.07 in 1970, but it rose sharply in the 1970s, flattened in the 1980s, and returned to a strong

### Global trade trends and context
- World trade-to-world GDP ratio:
  - "0.07 in 1970"
  - Rose sharply in the 1970s, flattened in the 1980s, returned to strong upward trajectory in the 1990s and first half of the 2000s.
  - The 2007-2009 Great Recession broke the trend: the ratio fell by "20 percent" (the “Great Trade Collapse”).
  - The ratio bounced back after the Great Recession but never reached pre-recession peak of "about 0.25".
  - "As of 2018, the world trade-to-world GDP ratio was about 0.20".
- Service trade coverage and assumption:
  - UN Comtrade covers only goods; service trade accounts for "about 22 percent of world total trade as of 2018".
  - Paper assumes expenditure and output shares of goods trade represent shares of total trade.

### Key modeling assumptions and data sources
- Assumptions on trade shares and disasters:
  - Globalization treated as exogenous to climate change.
  - Expenditure shares (π_{ni,y}) not affected by climatic disasters in downstream country n in year y.
  - Output shares (S_{ni,y}) not affected by climatic disasters in downstream country i in year y.
  - Upstream disasters reduce upstream supplies proportionally to annual output shares; downstream disasters reduce purchases proportionally to annual expenditure shares.
- Climate disaster and risk data:
  - Disaster data: Emergency Events Database (EM-DAT).
  - Disaster selection: require exact start date; further restricted to affected population rate > "0.5 percent" or damage > "0.05 percent of GDP".
  - Disaster damage metrics: total deaths, total people affected, total monetary loss.
  - Notable disasters:
    - Hurricane Katrina (2005) caused largest monetary damage: "$125 billion".
    - 2011 Thai floods caused largest monetary damage relative to host GDP: "10.1 percent".
  - Sample averages:
    - Average disaster causes "$783 million" monetary damage in current USD and "113" deaths, and affects "1.36 million" people.
    - On average, monetary damage is "0.01 percent of the hit country’s GDP".
  - Climate risk index: Verisk Maplecroft Climate Change Exposure Index.
    - Raw scale 0 (highest risk) to 10 (lowest risk); paper constructs hazard index by subtracting raw index from 10, then normalizes to mean "0" and standard deviation "1".
    - Fixed country climate risk to value in "2018".
    - Verisk Maplecroft data availability: "2013–19".
- Financial and sectoral data:
  - Stock market variables: stock index returns, price-to-earnings ratios, earnings per share — from Refinitiv Datastream.
  - Three-month government bond yields from Refinitiv Datastream.
  - Sector tradability: computed from WIOD 2016 release (covers "2000–16"), using "2000" as benchmark year to approximate sector tradability for all sample years.
- Data treatment:
  - All variables winsorized at top and bottom "1 percent".

### Construction of upstream and downstream exposure measures
- Upstream exposure (U_{n,y}):
  - U_{n,y} = ∑_{i=1}^N S_{ni,y} Damage_{i,y}
  - Interpreted as weighted sum of damages in countries that sell to n, weights = output shares S_{ni,y}.
- Downstream exposure (D_{n,y}):
  - D_{n,y} = ∑_{j=1}^N π_{jn,y} Damage_{j,y}
  - Interpreted as weighted sum of damages in countries that buy from n, weights = expenditure shares π_{jn,y}.
- Normalized exposures:
  - Nor_U_{n,y} = ∑ S_{ni,y} Damage_{i,y} / ∑ Damage_{i,y}
  - Nor_D_{n,y} = ∑ π_{jn,y} Damage_{j,y} / ∑ Damage_{j,y}
  - Cross-country sums of normalized exposures equal total world disaster damages.
- Benchmarking and relative exposure measures:
  - To isolate trade-share effects, define exposures using trade shares fixed at benchmark year y0.
  - Benchmark set to "t0 = 1971" (more countries have trade data starting 1971).
  - Relative exposure ratios for country groups:
    - Rel_U_{A,y0,y} = ∑_{n∈A} Nor_U_{n,y0,y} / ∑_{n∈A} Nor_U_{n,y0,y0}
    - Rel_D_{B,y0,y} analogous for downstream group B.

### Empirical findings on exposure evolution and globalization effects
- Low home-country shock group (country a; bottom 10 percent by home-country annual deaths or hazard index):
  - Upstream exposure to total climatic disaster-related deaths increased by "about 25 times".
  - Downstream exposure to total climatic disaster-related deaths increased by "about 35 times".
  - Upstream exposure to climate risks increased by "about 8 percent"; downstream exposure increased by "about 4 percent".
  - Global exposures for the low-shock group started to fall approximately at the time of the Great Recession.
- High home-country shock group (country b; top 10 percent by home-country annual deaths or hazard index):
  - Upstream exposure to global climatic disasters dropped by "about 8 percent".
  - Downstream exposure dropped by "about 6 percent".
  - Upstream exposure to global climate risks dropped by "about 20 percent"; downstream by "about 10 percent".
  - Around the Great Recession, high home-country shock countries began to increase their exposures.
- Annual average changes during sample (1971–2018) — country-level summary:
  - Country a experienced "0.34" and "0.54" annual average increase in upstream and downstream exposures of global climate disaster-related deaths, and "0.15" and "0.06" percentage points increase in upstream and downstream exposures of climate risks.
  - Country b experienced "0.07" and "0.08" annual average decline in upstream and downstream exposures of global climate disaster-related deaths, and "0.24" and "0.17" percentage points increase in upstream and downstream exposures of climate risks.
- Interpretation:
  - Globalization increased exposure of previously low-risk countries via trade linkages; reduced exposure of previously high-risk countries.
  - International trade changes allocation of disaster incidences across countries but does not change world total damages.

### Global spillover index and decomposition of drivers
- Global spillover index definition:
  - Spillover_up_{y0,y} = sd_n(Nor_U_{n,y0,y}) / sd_n(Nor_U_{n,y,y})
  - Spillover_down_{y0,y} analogous for downstream exposures.
  - Index equals "1" in benchmark year y = y0; rising index indicates reduced cross-country dispersion due to trade openness.
- Empirical relationship:
  - Time series of spillover indexes closely follow world trade-to-world GDP ratio.
  - Deepening globalization increases cross-border spillovers of climatic disasters and climate risks, reducing dispersion of exposures across countries.
- Decomposition of decline in cross-country dispersion (1971–2018):
  - Openness to trade alone explains "87.4 percent" (upstream) and "83.7 percent" (downstream) of the decline in cross-country dispersion in exposures to global climate disaster damages.
  - Changes in geographical distribution of climate disasters alone explain "35.9 percent" (upstream) and "29.7 percent" (downstream) of the decline.
  - From "1971 to 2008", cross-country dispersion in global climate disaster damage exposures declined by "6.3%" for upstream shocks and by "10.7%" for downstream shocks, relative to 1971.

### Event study: equity market responses to foreign climatic disasters
- Event linking approach:
  - For each disaster d hitting country i(d):
    - Link to largest export destination n_D(i(d)) = argmax_n≠i S_{n i(d), y(d)} — downstream country most affected by upstream disaster.
    - Link to largest import origin n_U(i(d)) = argmax_n≠i π_{i(d) n, y(d)} — upstream country most affected by downstream disaster.
  - On average, a country sells "about 3.1%" of its gross output to its largest exporting destination.
  - On average, a country spends "about 2.7 percent" of its total expenditure on its largest importing source.
- Event study design:
  - Excess returns defined relative to three-month government bond yield in the country.
  - Expected returns estimated on a window starting "12 months" before disaster and ending "one month" before disaster.
  - Event window: from "21" trading days before disaster to "60" trading days after disaster.
  - x-day cumulative abnormal return (CAR) computed and averaged across disasters; 95 percent confidence intervals reported.
- Main event-study findings:
  - Aggregate market-level cumulative abnormal returns to upstream and downstream climatic disasters (from -21 trading days to +40 trading days) were both "about -0.5%" and were marginally significant at "95 percent" confidence interval around "35" trading days after disaster start dates.
  - Magnitude comparable to home-country disaster’s average impact on stock market (about "-1 percent") reported in the GFSR.
- Sectoral heterogeneity:
  - Tradable sectors respond more strongly to foreign climate shocks; non-tradable sectors show little response.
  - Examples of sector CARs:
    - Chemicals: "-0.8 percent" for upstream disasters and "-1 percent" for downstream disasters.
    - Automobiles: "-1.8 percent" for upstream disasters and "-1.5 percent" for downstream disasters.
    - Media and telecommunication sectors do not respond significantly to foreign disasters.
- Placebo tests:
  - The disaster-hit country’s 35th largest (median) exporting or importing partner does not show significant stock market response to the disaster, supporting importance of trade linkages.
- Confounding channels and interpretation:
  - Upstream disaster on downstream country interpreted as negative supply shock to downstream.
  - Downstream disaster on upstream country interpreted as negative demand shock to upstream.
  - Offsetting channels (e.g., rebuilding demand raising upstream sales; domestic suppliers gaining market share) likely bias estimates upward; observed negative total effects for tradable sectors strengthen the proposed channels.
- Scope note:
  - Study links each disaster to the single largest affected partner (largest export destination or import origin) and provides a partial-equilibrium upper-bound estimate of foreign spillovers; aggregation across all trading partners left to future work.

### Methodological and data details relevant for replication
- Sector tradability measures:
  - Exporting tradability for sector s: TDEX_s = EX_s / VA_s (world total exports to world total value added).
  - Market-level exporting tradability: TDEX_mkt = EX_world / VA_world.
  - Importing tradability for sector s: TDIM_s = IIM_s / VA_s, where IIM_s = world total imports used as intermediate input for sector s.
  - Market-level importing tradability: TDIM_mkt = IIM_world / VA_world.
- Financial sector heterogeneity variables:
  - Total factoring volume-to-GDP ratio and bank regulatory capital-to-risk-weighted assets ratio from World Bank Global Financial Development Database used to explain cross-country heterogeneity in financial-sector responses.
- Winsorization:
  - To reduce outlier influence, all variables winsorized at top and bottom "1 percent".

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021013-print-pdf.pdf*

### Section III shows the cross-border spillovers of climatic disasters might reduce the country-level

### wpiea2021013-print-pdf - Section III shows the cross-border spillovers of climatic disasters might reduce the country-level

### Methodology and damage measures
- Upstream monetary disaster loss for country i is denoted Damage_i(d).
- Loss in sales from i to n_D is measured with S_{n_D i(d), y(d)} Damage_i(d).
- Normalized upstream damage:
  - nor_up_damage_{n_D(d), y(d)} = S_{n_D i(d), y(d)} Damage_i(d) / GDP_{n_D, y(d)}
- Downstream loss in purchases by i from n_U equals π_{i(d), n_U y(d)} Damage_i(d).
- Normalized downstream damage:
  - nor_down_damage_{n_U(d), y(d)} = π_{i(d), n_U y(d)} Damage_i(d) / GDP_{n_U, y(d)}
- Trading day 40’s cumulative abnormal return is defined as the sum of abnormal returns from 21 trading days before the disaster to 40 trading days after the disaster.

### Regression specification and estimation details
- Sector-level regression (upstream specification):
  - CAR_{n_D(d),40 s} = α_{1 s} nor_upstr_damage_{n_D(d), y(d)} + δ_{n_D s} + γ_{y s} + ε_{d s}
- Trading day 40’s cumulative abnormal return in the home country, sector s (or aggregate market) is regressed on normalized upstream damage.
- Controls:
  - Home-country fixed effect (δ_{n_D s}) to reflect cross-country differences in hedging abilities against foreign climate shocks.
  - Year fixed effect (γ_{y s}).
- Standard errors:
  - Two-way clustering on the stock market and year level.
- Regressions are run sector by sector, allowing α_{1 s} to vary across sectors.
- Downstream specification replaces n_D with n_U and normalized upstream damage with normalized downstream damage.

### Key empirical results — upstream climatic disasters
- The home-country stock market index return is negatively associated with the size of the upstream disaster.
- Exact estimated associations reported:
  - An increase in exposure to foreign upstream climatic disaster damage by 0.1 percent of the home country’s GDP is associated with a 5.9 percent decline in the market-wide trading day 40’s cumulative abnormal return.
- Sector heterogeneity (Table 4 results summarized):
  - Most tradable sectors show negative and significant responses: automobile, basic materials, chemicals, food and beverages, food producers, industrial goods, and industrial producers.
  - An increase in foreign upstream damage by 0.1 percent of home-country GDP is associated with a more than 10 percent decline in the stock market valuations in automobile and chemical sectors.
  - Most non-tradable sectors are not significantly affected by upstream damages.

### Key empirical results — downstream climatic disasters
- The correlation between cumulative abnormal return in the home-country market and the size of downstream disasters is significant and negative.
- Exact estimated association reported:
  - An increase in exposures to downstream foreign climatic disaster damage by 0.1 percent of home-country GDP is associated with a 3.7 percent decline in trading day 40’s market-level cumulative abnormal return.
- Sectoral pattern:
  - Typical tradable sectors respond more negatively and more significantly to the magnitude of downstream disasters than most non-tradable sectors.

### Tradability and cross-border propagation (pooled evidence)
- Pooled regression controlling for home country, year, and sector fixed effects:
  - CAR_{n_D(d),40 s} = α nor_upstr_damage_{n_D(d), y(d)} + δ_{n_D} + γ_{y} + ζ_{s} + ε_{d s}
- Table 6, Columns 1–2 (pooled averages) show:
  - For the average sector, the cumulative abnormal return is negatively correlated with the size of the upstream and downstream disasters.
  - Reported partial result: "An increase in upstream and downstream exposure by 0.1 percent of home-country GDP is associated with 4.7 percent and" (text truncated in source).

*Source: wpiea2021013-print-pdf - Section III (IMF).*

### 3.3 percent declines in market cumulative abnormal return on trading day 40.

### wpiea2021013-print-pdf - 3.3 percent declines in market cumulative abnormal return on trading day 40.

### Empirical design and regressions
- Use of importing tradability measure TDIM_s (and exporting tradability TDEX_s for downstream damage regressions).
- Panel regression for sector-level 40 trading day cumulative abnormal return (CAR_nD(d),40_s) on:
  - level of normalized upstream damage nor_upstr_damage_nD(d),y(d),
  - interaction nor_upstr_damage_nD(d),y(d) * TDIM_s,
  - country (δ_nD), year (γ_y), and sector (ζ_s) fixed effects.
- Equivalent specification for financial-sector analysis replaces CAR_nD(d),40 with CAR_nD(d),40_FIN and uses regulatory_to_assets (bank regulatory capital-to-risk-weighted assets) and factoring_to_gdp as institutional regressors (lagged one year).
- Cross-sectional analysis of P/E ratios:
  - First remove components explained by r_i,y (three-month government bond yield), EXPFE_i,y (mean annual growth of earnings per share over past five years), and ERP_i,y (std. dev. of annual growth of earnings per share over past five years).
  - Residual P/E (RPE_i,s) regressed on upstream (U_i) and downstream (D_i) foreign climate risk exposures, sector fixed effects, and interactions with sector tradability (TDIM_s or TDEX_s).
- Placebo tests construct Ũ_i and D̃_i by setting all country climate risks to 1/(N−1) to test whether openness to trade alone drives results.

### Main empirical findings — tradability and CAR responses
- Without interaction terms, level effects of normalized upstream and downstream damage are statistically significant negative for the pooled sample (Columns 1–2, Table 6):
  - nor_up_damage coefficient: -47.20** (standard error 22.43).
  - nor_down_damage coefficient: -32.67* (standard error 16.37).
- Once interactions are included:
  - nor_up_damage * TDIM_s is -246.5*** (standard error 56.28) (Column 3, Table 6).
  - nor_down_damage * TDEX_s is -83.74*** (standard error 18.96) (Column 4, Table 6).
  - Level effects of upstream/downstream damage become insignificant when interaction terms are included — implying tradable sectors drive negative CAR effects.
- Magnitude examples implied by estimated coefficients:
  - An increase in exposure to upstream foreign climatic disaster by 0.1 percent of home-country GDP is predicted to reduce cumulative abnormal returns by:
    - 15.4 percent in the sector with the highest importing tradability (chemicals).
    - 4.1 percent in the sector with the median importing tradability (food and beverages).
    - 0.98 percent in the sector with the lowest importing tradability (real estate).
  - An increase in exposure to downstream foreign climatic disaster by 0.1 percent of home-country GDP is predicted to reduce cumulative abnormal returns by:
    - 9.5 percent in the sector with the highest exporting tradability (automobile).
    - 2.2 percent in the sector with the median exporting tradability (media).
    - 0.92 percent in the sector with the lowest exporting tradability (real estate).

### Financial sector heterogeneity and institutional buffers
- Home-country climatic disasters reduce valuation of the financial sector stocks; foreign disasters undermine financial stability indirectly via tradable sectors.
- Regression linking financial-sector CAR to banking capitalization finds:
  - One percentage point increase in the bank regulatory capital-to-risk-weighted assets ratio is associated with about a 0.2 percent increase in the cumulative abnormal return from upstream and downstream disasters (Columns 2 and 4 of Table 7).
- Factoring-to-GDP is also positively associated with CAR in the financial sector:
  - factoring_to_gdp (%) coefficient ~ 0.00108** (Column 1, Table 7) and 0.00105** (Column 3, Table 7) — positive association holding normalized damage fixed.
- Implication: stronger bank capitalization and higher factoring activity mitigate negative CAR responses in the financial sector to foreign climatic disasters.

### Equity pricing of foreign climate change physical risk (P/E analysis)
- Upstream and downstream foreign climate risk exposures (U_i and D_i) are constructed as trade-weighted sums of partner-country climate risks R_j, excluding the home country.
- Cross-section (year fixed at 2018) pooled results (Table 8, Columns 1–2):
  - foreign_exp (upstream pooled) coefficient: -43.04*** (standard error 15.11).
  - foreign_exp (downstream pooled) coefficient: -43.94** (standard error 20.06).
  - One standard deviation increase in upstream or downstream exposures corresponds to about 0.05 standard deviation decline in the P/E ratio; inter-quartile increase associated with P/E reductions of about 3.0 (upstream) and 3.7 (downstream).
- Sector heterogeneity:
  - Industrial producers (tradable) show strong negative correlations (Columns 3 and 5).
  - Real estate (non-tradable) shows no significant correlation (Columns 4 and 6).
- Interaction with tradability:
  - Adding TDIM_s * U_i removes level significance of foreign_exp — tradable sectors drive the negative association (Columns 7–8).
  - Example magnitudes for upstream U_i (Table 8, Column 7):
    - Sector at 50th percentile importing tradability (food and beverages): one standard deviation increase in U_i → 0.0488 standard deviation decline in P/E.
    - Sector at 25th percentile importing tradability (travel and leisure): 0.0286 standard deviation decline.
    - Sector at 75th percentile importing tradability (industrial producers): 0.0742 standard deviation decline.
  - For downstream D_i (Table 8, Column 8):
    - 25th percentile exporting tradability (insurance): 0.0075 standard deviation decline.
    - 50th percentile (media): 0.0169 standard deviation decline.
    - 75th percentile (industrial producers): 0.1066 standard deviation decline.
- Placebo tests:
  - Placebo upstream/downstream foreign exposures (Ũ_i, D̃_i) are not significantly correlated with home P/E ratios (Columns 9–10).
  - Interactions of placebo exposures with tradability are also not significant (Columns 11–12).
  - Interpretation: openness to trade per se does not explain the negative P/E association; trading with high-risk partners does.

### Key conclusions and policy implications
- Foreign climatic disasters generate negative asset-price spillovers to aggregate markets and especially to tradable sectors in partner countries via trade and supply-chain linkages.
- Tradable-sector exposure is the main channel; non-tradable sectors show little or no direct asset-price sensitivity to foreign climate risk in these results.
- Bank capitalization and factoring markets can cushion the financial-sector valuation impact from foreign climatic disasters.
- Policy implications highlighted:
  - Enhancing resilience through adaptation is a common international responsibility because trade transmits climate risk across borders.
  - Collective international policy action complements domestic adaptation and cross-border economic incentives.
  - The analytical framework can be applied to transition risk, other crises (for example, COVID-19), and other globalization channels (multinational production, remittances, tourism).
- Caveats and further research:
  - Financial stability implications depend on country-specific factors (size of tradable sectors, bank exposures).
  - Quantifying country-specific stability effects requires further calibrated modeling.
  - More granular data could test additional global supply chain characteristics (for example, input-specificity) and firm- or region-level spillovers.

*Source: wpiea2021013-print-pdf - 3.3 percent declines in market cumulative abnormal return on trading day 40.*

### REFERENCES

### REFERENCES

### Climate and Temperature Effects
- Addoum, J., D. Ng, and A. Ortiz-Bobea, 2019. Temperature Shocks and Industry Earnings News. Available at SSRN 3480695.
- Burke, M., S.M. Hsiang, and E. Miguel, 2015. “Global Non-Linear Effect of Temperature on Economic Production,” Nature, 235-39.
- Dell, M., B.F. Jones, and B.A. Olken, 2012. “Temperature Shocks and Economic Growth: Evidence from the Last Half Century,” American Economic Journal: Macroeconomics, 4(3), 66-95.
- Dingel, J.I., K.C. Meng, and S.M. Hsiang, 2019. “Spatial Correlation, Trade, and Inequality: Evidence from the Global Climate,” National Bureau of Economic Research, No. w25447.
- Hsiang, S.M., 2010. “Temperatures and Cyclones Strongly Associated with Economic Production in the Caribbean and Central America,” proceedings of the National Academy of Sciences of the United States of America 107:15367–72.
- Hsiang, S.M., and A.S. Jina, 2014. “The Causal Effect of Environmental Catastrophe on Long-Run Economic Growth: Evidence from 6,700 Cyclones,” National Bureau of Economic Research, No. w20352.
- Kahn, M.E., K. Mohaddes, R.N. Ng, and others, 2019. “Long-Term Macroeconomic Effects of Climate Change: A Cross-Country Analysis,” . National Bureau of Economic Research, No. w26167.
- McKinsey, 2020. “Climate Risk and Response: Physical Hazards and Socioeconomic Impacts.” January.
- BlackRock, 2019. “Getting Physical: Scenario Analysis for Assessing Climate-Related Risks,” BlackRock Investment Institute, https://www.blackrock.com/ch/individual/en/insights/physical-climate-risks.
- Hong, H., F. Li, and J. Xu, 2019. “Climate Risks and Market Efficiency,” Journal of Econometrics, 208(1), 265-281.

### Supply Chains, Trade, and Production Networks
- Antras, P., T.C. Fort, and F. Tintelnot, 2017. “The Margins of Global Sourcing: Theory and Evidence from US Firms, American Economic Review, 107(9), 2514-64.
- Auboin, M., H. Smythe, and R. Teh, 2016. Supply Chain Finance and SMEs: Evidence from International Factoring Data.
- Baldwin, R., and R. Freeman, 2020. “Supply Chain Contagion Waves: Thinking Ahead on Manufacturing ‘Contagion and Reinfection’ from the COVID Concussion,” VoxEU. org.
- Barrot, J.N., and J. Sauvagnat, 2016. “Input Specificity and the Propagation of Idiosyncratic Shocks in Production Networks,” The Quarterly Journal of Economics, 131(3), 1543-92.
- Carvalho, V.M., M. Nirei, Y. Saito, and others, 2016. “Supply Chain Disruptions: Evidence from the Great East Japan Earthquake,” Columbia Business School Research Paper, (17-5).
- Cadestin, C., K. De Backer, I. Desnoyers-James, and others, 2018. “Multinational Enterprises and Global Value Chains,” OECD AAMNE Database.
- Chaney, T., 2018. “The Gravity Equation in International Trade: An Explanation,” Journal of Political Economy, 126(1), 150-77.
- Eaton, J., S. Kortum, B. Neiman, and others, 2016. “Trade and the Global Recession,” American Economic Review, 106(11), 3401-38.
- Johnson, R.C., and G. Noguera, 2017. “A Portrait of Trade in Value-Added Over Four Decades,” Review of Economics and Statistics, 99(5), 896-911.
- Korniyenko, M.Y., M. Pinat, and B. Dew, 2017. “Assessing the Fragility of Global Trade: The Impact of Localized Supply Shocks Using Network Analysis,” IMF Working Paper 17/30.
- Kouvelis, P., and F. Xu, 2019. A Supply Chain Theory of Factoring and Reverse Factoring. Available at SSRN 3241484.
- Tingergen, J., 1962. Shaping the World Economy, An Analysis of World Trade Flows. New York: Twentieth Century Fund.

### Disasters, Firm-Level Impacts, and Labor
- Anttila-Hughes, J.K., and S.M. Hsiang, 2013. Destruction, Disinvestment, and Death: Economic and Human Losses Following Environmental Disaster. Available at SSRN 2220501.
- Leiter, A.M., H. Oberhofer, and P.A. Raschky, 2009. “Creative Disasters? Flooding Effects on Capital, Labour and Productivity Within European Firms,” Environmental and Resource Economics 43:333–50.
- Strobl, E., 2011. “The Economic Growth Impact of Hurricanes: Evidence from US Coastal Counties,” Review of Economics and Statistics 93:575–89.
- Li, H., and G. Souza, 2020. The Aggregate Labor Market Consequences of Anti-Dumping Tariffs. University of Chicago mimeo.

### Methods, Data, and Nowcasting
- Cerdeiro, D.A., A. Komaromi, Y. Liu, and others, 2020. “World Seaborne Trade in Real Time: A Proof of Concept for Building AIS-Based Nowcasts from Scratch,” IMF Working Paper No. 20/57.
- Caliendo, L., F. Parro, E. Rossi-Hansberg, and others, 2018. “The Impact of Regional and Sectoral Productivity Changes on the US Economy,” The Review of Economic Studies, 85(4), 2042-96.
- Di Giovanni, J., A.A. Levchenko, and I. Mejean, 2018. “The Micro Origins of International Business-Cycle Comovement,” American Economic Review, 108(1), 82-108.
- Timmer, M.P., E. Dietzenbacher, B. Los, and others, 2015. “An Illustrated User Guide to the World Input–Output Database: The Case of Global Automotive Production,” Review of International Economics, 23(3), 575-605.

### IMF and Institutional Reports
- International Monetary Fund, 2020. Global Financial Stability Report: Markets in the Time of COVID-19. Washington: International Monetary Fund.
- International Monetary Fund, 2019. Fiscal Monitor: How to Mitigate Climate Change. Washington: International Monetary Fund.

*Source: wpiea2021013-print-pdf - REFERENCES.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021013-print-pdf.pdf_