## Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies

## Source details

**Canonical URL:** [Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022189-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022189-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022189-print-pdf.pdf.json)

---

### Introduction: observed and projected climate trends
- Global temperatures have increased at unprecedented levels over the past 40 years and are expected to increase by 1.5ºC by 2030 and by 2.2 to 3.5ºC by 2100 (IPCC, 2022).
- These trends will increase the frequency and severity of extreme weather events, such as heat waves, droughts, and floods (IMF, 2020a).
- Recent analyses show the risk of infectious diseases increases around five-fold in the year following a heatwave event such as El Niño (WHO, 2020).

### Executive summary and scenario design
- Paper stress tests the global economy to severe country-specific climate change shocks that outstrip an economy’s capacity to cope.
- Dataset: 63 economies representing around 80 percent of global GDP.
- Framework components:
  - Identify climate-vulnerable, large, interconnected economies.
  - Quantify conditional effects of climate shocks on full distribution of external financing needs using a value-at-risk based forecasting methodology.
  - Quantify global spillovers with a multilayered network model incorporating trade and financial linkages and domestic and multilateral policy responses.
- Scenario: tail risk where weakened access to international financial markets following climate shocks impairs debt service, triggering default, contagion, and spillovers; model embeds domestic monetary and fiscal policy responses and multilateral financing support through the global financial safety net (GFSN).

### Key empirical findings and magnitudes
- Probability and EFN outcomes:
  - There is a 50 percent probability that external financing needs in two large emerging economies could exceed $19 billion (5 percent of GDP) and $342 billion (12 percent of 2019 GDP) following a climate change-related shock.
- Global reserve loss estimates:
  - Global losses could reach up to $1.8 trillion in the absence of any macroeconomic policy action.
  - Exchange rate and fiscal consolidation in line with historical IMF program patterns could reduce losses to about $1.2 trillion.
  - Multilateral financing support through the GFSN, including IMF financing, could further reduce losses to about $0.8 trillion.
- Literature evidence:
  - A persistent increase in average global temperature by 0.04°C per year would reduce, absent mitigation, world real per capita GDP by more than 7 percent by 2100 (Kahn et al., 2019).
  - Productivity can decline by about 1.7 percent for each 1°C increase in daily average temperature above 15°C (Deryugina and Hsiang, 2014).
  - Historical losses: about 12,000 extreme weather events between 1999 and 2018 led to about 495,000 deaths and an economic loss of around $3.54 trillion in purchasing power parity.

### Mechanisms of impact
- Physical risks:
  - Supply-side shocks reducing productive capacity, price volatility from commodity shortages, damage to capital stock and infrastructure.
  - Demand-side shocks reducing household wealth and consumption.
  - Ambiguous effects on investment due to reconstruction needs versus uncertainty and insurance gaps.
- Transition risks:
  - Policy responses (carbon pricing, environmental regulation, trade policies) can raise energy and commodity prices and impose adjustment costs.
  - Example policy magnitudes cited:
    - A carbon price of 75 dollars per ton of CO2 by 2030 suggested to help limit temperature increase to at most 1.5°C.
    - A carbon tax increase to $75 a ton of CO2 would boost average electricity prices by 45 percent cumulatively and gasoline prices by 15 percent for ten years; coal prices would rise by more than 200 percent above baseline levels in 2030 (IMF, 2019c).
- Financial amplification:
  - Weakened insurance coverage: 26 percent of losses from world’s largest natural disasters between 1980 and 2015 were covered by insurance.
  - Falling real estate collateral values, potential insurance company stress, and repricing of long-term climate risks in equity and sovereign markets.

### Interconnectedness, spillovers, and contagion
- Localized climate events can propagate globally through trade (export restrictions, commodity price shocks) and finance (contagion via sovereign risk premiums and capital markets).
  - Exogenous international food commodity price shocks can explain on average 25 to 30 percent of inflation volatility in the euro area (Peersman, 2018).
- Network modeling captures feedbacks across portfolios, institutions, sectors, and countries; the paper applies a novel multilayered network at the global macro level.

### Measures of climate change vulnerability and sample identification
- Vulnerability indices described:
  - ND-GAIN (University of Notre Dame’s Global Adaptation Index): captures vulnerability and readiness; ND GAIN score = (Readiness -Vulnerability + 1) *50.
  - Global Climate Risk Index (CRI) by Germanwatch: analyses fatalities and economic losses; example CRI Score calculation given as 114.17.
- Sample and identification:
  - 63 economies representing about 80 percent of 2018 global GDP.
  - Six large economies identified as highly vulnerable (over 75th percentile of ND-GAIN and above 90th percentile of economic size); three of these (A, B, C) display substantial financial and trade interconnectedness.
  - The three focus economies together represent about 5 percent of global GDP and 4 percent of world total imports; over 30 percent of their trade is directly linked to global value chains (GVCs).

### External Financing Needs-at-Risk (F@R): methodology and econometric results
- Definition: EFNs = short-term debt + amortization of medium- and long-term debt − current account balance.
- Forecasting approach:
  - Quantile regressions estimate conditional density of EFNs as a function of climate indicators (ND-GAIN and CRI), domestic variables, and global conditions.
  - Controls include real GDP growth, inflation, fiscal deficit-to-GDP ratio, public debt-to-GDP ratio, external debt-to-GDP ratio, credit-to-GDP gap, IMF GFSR financial stability index, VIX, credit-to-GDP ratio, real effective exchange rate, world real GDP growth, U.S. real GDP growth, China’s real GDP growth, oil prices.
  - Sample period: 1999q1-2019q4; forecast horizon: four quarters-ahead.
  - Conditional pdf derived using the parametric skewed t-distribution.
- Baseline four-quarter ahead F@R density (absent COVID-19):
  - Economy A: 50 percent probability EFNs > $ 2 billion (0.5 percent of GDP); 5 percent tail EFNs > $ 16 billion (4.2 percent of GDP).
  - Economy B: 50 percent probability EFNs > $ 25 billion (0.9 percent of GDP); 5 percent tail EFNs > $ 161 billion (5.6 percent of GDP).
- Two-standard-deviation climate shock to ND-GAIN:
  - After shock:
    - 50 percent probability EFNs in country A exceed $19 billion (5.0 percent of GDP).
    - 50 percent probability EFNs in country B exceed $342 billion (11.9 percent of GDP).
  - 5.0 percent tail under shock:
    - EFNs exceed $ 40 billion in economy A (10.6 percent of GDP).
    - EFNs exceed US$ 494 billion in economy B (17.2 percent of GDP).
- Contextual facts from source:
  - Floods cost economy B about $7 billion every year.
  - A 2013 weather event in economy A killed more than 6,000 people, damaged 1.1 million homes, and caused $802 million in agricultural and infrastructure damages.
  - IPCC (2021) warns economy B will suffer more frequent and intense heat/humid waves, extreme rainfall, erratic monsoons, and cyclonic activities.

### Multilayered network model: structure, simulations, and amplification channels
- Model focus: captures impact of climate shock–induced defaults in countries A and B, simulates balance of payments dynamics via trade and cross-border asset/liability positions, and incorporates GFSN layers (bilateral swap lines, RFAs, IMF).
- Reserve dynamics and default trigger:
  - Changes in foreign reserves, ∆RR_i,t, reflect trade revenues, net interest payments on foreign assets and foreign liabilities, exchange rate changes and net external financing from the GFSN.
  - Notation specifics preserved for asset holdings a_ij, interest rates r_j,t, exchange rates e_j,t, and EEF_i,t.
  - An economy stops servicing external debt when reserves reach 80 percent of the IMF’s assessed reserve adequacy (ARA) metric.
- Amplification channels:
  - Risk premium reacts nonlinearly to distance-to-default; fitted regression example: 푅푅_{푖,푡} = 150 퐷퐷_{푡푡퐷퐷}^{1.8} + 60.
  - Sovereign spreads and climate vulnerability relationship estimated as: 푅푅_{푖,푡} = 2151exp (−0.05퐺퐺퐺퐺퐺퐺푁푁).
  - Asset-price co-movement contagion calibrated to maximum changes in sovereign risk premia during the 2013 Taper Tantrum.
- Simulation results — tail risk scenario (defaults in A and B):
  - 'No policy' outcome:
    - Number of economies defaulting: 26 economies.
    - Global reserve losses: $ 1.8 trillion.
  - Domestic policy responses (currencies allowed depreciation and fiscal adjustment) reduce global reserve losses by about US$ 0.6 trillion relative to 'no policy'.
  - GFSN financing effects:
    - Financing by GFSN excluding the IMF reduces global reserve losses by over US$ 0.3 trillion.
    - IMF financing further significantly reduces contagion and global reserve losses.
  - Residual systemic impact after all financing avenues:
    - Even after all financing, an extreme shock in A and B could push seven other economies into default and trigger global reserve losses of US$ 800 billion.
  - Aggregate policy effect:
    - Full deployment of domestic and multilateral policies can reduce global reserve losses by more than one-half, from $ 1.8 trillion to about $ 800 billion.
- Interpretation:
  - Direct exposures alone generate relatively small global reserve losses; amplification channels (distance-to-default and asset price comovements) substantially increase systemic impact.
  - Interconnectedness and contagion materially increase global costs when large, interconnected economies default.

### Policy implications and recommendations
- Principal recommendations:
  - Pursue adequate adaptation and mitigation policies to reduce vulnerability and the likelihood of events that could trigger economic and financial spillovers.
  - Maintain sustainable macroeconomic policies to preserve policy space to respond to climate shocks and attendant spillovers.
  - Enhance global financial safety nets (GFSN) — bilateral swap lines, RFAs, and IMF facilities — to materially limit global losses from climate-induced sovereign distress.
  - Support vulnerable countries to build capacity and readiness; global collective efforts are warranted because mitigation and adaptation are global public goods.
  - Establish common global principles and disclosure standards to strengthen climate risk assessment and help control cross-country spillovers.
- Additional notes:
  - Domestic macroeconomic policies (monetary and fiscal responses, exchange rate and fiscal consolidation) are critical in containing systemic effects.
  - The global benefits from individual countries’ mitigation and adaptation efforts are likely much larger than commonly estimated because such policies reduce external vulnerabilities and global contagion risk.
  - Policy design should consider distributional impacts of mitigation (e.g., carbon pricing) and include compensation for adversely affected groups.

### Data, robustness checks, and annex material
- Data and coverage:
  - Network model uses 2019 data for 63 economies (36 advanced and 27 emerging market) representing about 85 percent of global GDP.
  - Data inputs: DOTS, BIS Locational International Banking Statistics, IMF CPIS, IMF CFDIS, nominal bilateral exchange rates, sovereign spreads, interest rates, foreign exchange reserves, and access to bilateral swap lines, RFAs, and Fund financing.
- Robustness checks:
  - Alternative vulnerability indicator: frequency of climate-related natural disasters from IMF Climate Change Dashboard confirms robustness of ND-GAIN-based identification; three large economies A, B, and C are over the 95th percentile of disaster frequency.
- Annex summaries:
  - Quantile regressions for F@R presented for Country A and Country B (Annex III).
  - Multilayered network model formal setup, amplification mechanisms, policy/GFSN features, and data sources elaborated (Annex IV).
  - Annex Table A.1 lists vulnerable economies identified by ND-GAIN, CRI, and IMF studies.

*Source: IMF Working Paper — Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies, Working Paper No. WP/2022/189*

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

### Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies

### Introduction: observed and projected climate trends
- Global temperatures have increased at unprecedented levels over the past 40 years and are expected to increase by 1.5ºC by 2030 and by 2.2 to 3.5ºC by 2100 (IPCC, 2022).
- These trends will lead to a significant increase in the frequency and severity of extreme weather events, such as heat waves, droughts, and floods (IMF, 2020a).
- Extreme weather events that were once considered tail-risk events are becoming more frequent and need careful analysis.

### Economic and human impacts documented
- Climate change-driven events can:
  - Adversely impact natural capital.
  - Result in the destruction of infrastructure.
  - Disrupt livelihoods and cause mass migration.
  - Produce devastating humanitarian consequences (IMF, 2019a).
- Climate change is changing the pattern of diseases and mortality of the global population.
  - Recent analyses show that the risk of infectious diseases increases around five-fold in the year following a heatwave event such as El Niño (WHO, 2020).
  - Going forward, climate change may also increase the likelihood of new costly and deadly diseases and pandemics (Stanford Woods Institute for the Environment, 2019).
- Extreme weather-related events have been shown to trigger severe and persistent output losses:
  - Damage associated with natural disasters in the small states of the Caribbean and the Pacific is estimated at an average of 2-3 percent of GDP on an annual basis (IMF, 2019d).
  - In the longer run, the per capita GDP of a representative low-income country could be 9 percent lower in 2100 than what it would have been in the absence of temperature increases (IMF, 2017).
  - Other studies estimate that a 1°C rise in temperature in a given year can reduce that year’s annual economic growth by about [text truncated in source].

### Financial and sovereign risk linkages
- Climate risks pose large risks to the global economy and are ranked among the top risks in likelihood and impact over the next ten years by the World Economic Forum (WEF, Global Risks Report 2022).
- Sovereign risk and climate change risks are interwoven:
  - Extreme weather has often played a prominent role in sovereign default episodes (Mallucci, 2020).
  - Extreme climate-related events reduce governments’ ability to issue debt and restrict access to financial markets.
  - Financial markets have started to price-in the risk of climate change on sovereign debt (Zheng and Tovar, 2022).

### Gaps and objectives of the analysis
- Despite growing evidence on adverse economic and financial effects, the magnitude and channels through which climate change operates are not clearly understood.
- The paper sets out to stress test the global economy to climate change-related shocks in large and interconnected economies, analyzing:
  - Economic size, interconnectedness, and climate change vulnerability (Figure 1).
  - Global Value Chains (GVCs) and climate vulnerability (Figure 2).
  - External Financing Needs-at-Risk (F@R) density estimates for baseline and climate change-related shock scenarios (Figure 3).
  - CDS spreads and climate change vulnerability (Figure 4).
  - CDS spreads and reserve adequacy levels (Figure 5).
  - An extreme scenario of global systemic impact of a combined climate change-related shock in Country A and Country B (Figure 6).
- Assessment framework components include:
  - Assessing country vulnerabilities to climate change risks (Table 1).
  - Data sources and robustness checks (Annex I and Annex II).
  - Quantile regressions for External Financing Needs-at-Risk (Annex III).
  - A multilayered network model (Annex IV).

*IMF Working Paper — Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies*

### 1.3 percentage points, with both level and growth effects that persist in the medium term (Dell et al., 2012), or that a

### Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies

### Executive summary and scenario design
- Paper stress tests the global economy to severe country-specific climate change shocks (e.g., floods, heat waves, droughts) that manifest as economic shocks and outstrip an economy’s capacity to cope.
- Focus on economies that are large, interconnected, and vulnerable to climate change; dataset covers 63 economies, representing around 80 percent of global GDP.
- Framework: identify climate-vulnerable, large, interconnected economies; quantify conditional effects of climate shocks on the full distribution of external financing needs using a value-at-risk based forecasting methodology; quantify global spillovers with a multilayered network model that incorporates trade and financial linkages and domestic and multilateral policy responses.
- Scenario is a tail risk: weakened capacity to access international financial markets following climate shocks impairs debt service, triggering default, contagion, and spillovers.
- Model embeds domestic monetary and fiscal policy responses and multilateral financing support through the global financial safety net (bilateral swap lines, regional financing arrangements, IMF).

### Key empirical findings and magnitudes
- There is a 50 percent probability that external financing needs in two large emerging economies could exceed $19 billion (5 percent of GDP) and $342 billion (12 percent of 2019 GDP) following a climate change-related shock (e.g., record breaking floods or droughts).
- Global losses, captured by aggregate global losses of international reserves, could reach up to $1.8 trillion in the absence of any macroeconomic policy action.
- Exchange rate and fiscal consolidation in line with historical patterns seen in IMF programs could reduce these losses to about $1.2 trillion.
- Multilateral financing support through the global financial safety net, including IMF financing, could further reduce losses to about $0.8 trillion.
- Literature evidence cited: a persistent increase in average global temperature by 0.04°C per year would reduce, in the absence of mitigation policies, world real per capita GDP by more than 7 percent by 2100 (Kahn et al., 2019).
- Other literature: productivity can decline by about 1.7 percent for each 1°C increase in daily average temperature above 15°C (Deryugina and Hsiang, 2014).
- Historical losses: global losses from about 12,000 extreme weather events between 1999 and 2018 led to about 495,000 deaths and an economic loss of around $3.54 trillion in purchasing power parity.

### Mechanisms of impact
- Physical risks: supply-side shocks reducing productive capacity (price volatility from shortages of commodities; damage to capital stock and infrastructure); demand-side shocks reducing household wealth and consumption; potential ambiguous effects on investment (reconstruction vs. uncertainty and insurance gaps).
- Transition risks: policy responses (carbon pricing, environmental regulation, trade policies) can raise energy and commodity prices and impose adjustment costs on energy-dependent sectors and communities.
  - Example: a carbon price of 75 dollars per ton of CO2 by 2030 is suggested to help limit temperature increase to at most 1.5°C (IMF Fiscal Monitor).
  - A carbon tax increase to $75 a ton of CO2 would boost average electricity prices by 45 percent cumulatively and gasoline prices by 15 percent for ten years (IMF, 2019c); coal prices would rise by more than 200 percent above baseline levels in 2030 (IMF, 2019c).
- Financial amplification: weakened insurance coverage (26 percent of losses from world’s largest natural disasters between 1980 and 2015 were covered by insurance), falling real estate collateral values, potential insurance company stress, and repricing of long-term climate risks in equity and sovereign markets.

### Interconnectedness, spillovers, and contagion
- Localized climate events can have global effects through trade (export restrictions, commodity price shocks) and finance (contagion via sovereign risk premiums and capital markets).
  - Exogenous international food commodity price shocks can explain on average 25 to 30 percent of inflation volatility in the euro area (Peersman, 2018).
- Network modeling is necessary to capture feedbacks across portfolios, institutions, sectors, and countries; few prior studies apply network models at the aggregate global macro level—this paper presents a novel multilayered network application.

### Policy implications and recommendations
- Current economic and financial estimates appear to be underestimating overall effects of climate change-related shocks; localized tail events can produce large systemic effects.
- Domestic macroeconomic policies are critical in containing systemic effects (monetary and fiscal responses, exchange rate and fiscal consolidation).
- Multilateral cooperation and the global financial safety net (including IMF financing) can play a key role in containing spillovers and contagion and materially reduce global reserve losses.
- The global benefits from individual countries’ mitigation and adaptation efforts are likely much larger than commonly estimated because such policies reduce external vulnerabilities and global contagion risk.
- Policy design should consider distributional impacts of mitigation (e.g., carbon pricing) and include compensation for adversely affected groups.

### Research contributions and structure
- First attempt, to authors’ knowledge, to quantify and stress test the global economy to a country-specific climate change-related shock using risk management tools and a global multilayered network model.
- Paper structure: literature review (Section 2); measures of climate change vulnerability and identification of vulnerable large, interconnected countries (Section 3); quantification of external vulnerability (Section 4); global stress testing with multilayered network model (Section 5); conclusions and policy implications.

*Source: IMF Working Paper — Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies*

### 2.5 percent of 2019's global GDP (Germanwatch, 2020). This relatively modest economic loss may be the

### wpiea2022189-print-pdf - 2.5 percent of 2019's global GDP (Germanwatch, 2020). This relatively modest economic loss may be the

### Observed economic losses from natural disasters and trends
- Global economic losses of natural disasters at $280 billion in 2021 (of which $120 billion was insured).
- 2020 global losses: $210 billion.
- 2019 global losses: $166 billion.
- United States losses in 2021: $145 billion (of which $85 billion were insured).
- Floods in Germany in 2021: $54 billion (of which $13 billion were insured); described as the costliest natural disaster in Germany’s history (Munich RE, 2022).
- Earlier estimate: 2.5 percent of 2019's global GDP (Germanwatch, 2020) cited as a relatively modest economic loss.

### Measures of climate change vulnerability (composite indices)
- Two alternative composite indices described:
  - ND-GAIN (University of Notre Dame’s Global Adaptation Index)
    - Captures both vulnerability and readiness.
    - Vulnerability assessed with six life-supporting sectors—food, water, health, ecosystem services, human habitat, and infrastructure.
    - Readiness encompasses investment climate (economic readiness), institutional arrangements (governance readiness), and social conditions (social readiness).
    - GDP related measures and data on recent climate change-related disasters are explicitly excluded.
    - Index scores computed as: ND GAIN score = (Readiness -Vulnerability + 1) *50
  - Global Climate Risk Index (CRI) by Germanwatch
    - Analyses quantified impacts of extreme weather events based on fatalities and economic losses: fatalities, fatalities per 100,000 inhabitants, loss in millions of $ (PPP), and losses per unit of GDP in percent.
    - Creates an average ranking of economies in four indicating categories with greater weight on relative indicators.
    - Example calculation: CRI Score = 137 x 1/6 + 130 x 1/3 + 114 x 1/6 + 87 x 1/3 = 114.17

- Comparative note:
  - ND-GAIN is emphasized as best suited for estimating the global impact of a climate shock because it includes readiness and vulnerability, allowing assessment of an economy’s capacity to leverage adaptation investments.
  - Readiness indicators include relatively long-term indicators such as political governance, doing business, social inequality, and education.

### Identifying large, interconnected, and vulnerable economies
- Sample includes 63 economies representing about 80 percent of 2018 global GDP.
- Identification criteria (illustrated in Figure 1):
  - Six large economies identified as highly vulnerable to climate change: over the 75th percentile of the ND-GAIN distribution and above the 90th percentile of economic size in the sample.
  - Of these six, three economies (anonymized as A, B, and C) display substantial financial and trade interconnectedness.
- Characteristics of the three focus economies:
  - Together represent about 5 percent of global GDP.
  - Represent 4 percent of world total imports.
  - Over 30 percent of their trade is directly linked to global value chains (GVCs), much of which corresponds to intra-regional GVC activities (WTO et al., 2019).
  - Economy C’s strong external position resulted in a positive external financing position and is not found to be vulnerable to a climate change-related shock as to trigger significant spillovers.

### Global Value Chains (GVCs) and amplification risk
- Economies highly integrated in GVCs are vulnerable to climate change shocks in trade-linked sectors, potentially reverberating and amplifying effects across the world.
- Panel B of Figure 2 displays economies considered highly integrated in GVCs alongside ND-GAIN vulnerability.

### External Financing Needs-at-Risk (F@R): methodology
- Builds on value-at-risk approaches (e.g., IMF’s growth-at-risk).
- External financing needs defined as: short-term debt plus the amortization of medium- and long-term debt minus the current account balance.
- Forecasting method:
  - Uses quantile regressions to estimate conditional density of EFNs as a function of climate change indicators (ND-GAIN and CRI), domestic variables, and global conditions.
  - Domestic controls: real GDP growth, inflation, fiscal deficit-to-GDP ratio, public debt-to-GDP ratio, external debt-to-GDP ratio, credit-to-GDP gap.
  - Financial controls: IMF GFSR financial stability index, CBOE volatility index (VIX), credit-to-GDP ratio, real effective exchange rate.
  - Global controls: world real GDP growth, U.S. real GDP growth, China’s real GDP growth, oil prices.
  - Data sources: IMF’s International Financial Statistics, IMF’s World Economic Outlook database, IMF’s GFSR, Haver Analytics.
- Density estimation:
  - Conditional quantiles used to obtain conditional cumulative distribution function (cdf) of EFNs.
  - Probability density function (pdf) derived using the parametric skewed t-distribution.
  - Sample period: 1999q1-2019q4.
  - Density forecast horizon: four quarters-ahead.

### Econometric results and scenario findings for economies A and B
- Baseline four-quarter ahead F@R density (absent COVID-19 effects):
  - Under baseline estimate, EFNs in economy A have a 50 percent probability of exceeding $ 2 billion, equivalent to 0.5 percent of GDP.
  - Under baseline estimate, EFNs in economy B have a 50 percent probability of exceeding $ 25 billion, equivalent to 0.9 percent of GDP.
  - Baseline 5 percent tail: 5 percent probability that EFNs will exceed $ 16 billion in economy A (4.2 percent of GDP) and $ 161 billion in economy B (5.6 percent of GDP).
- Two-standard-deviation climate change-related shock applied to ND-GAIN (increases vulnerability):
  - The shock induces a substantial shift in the EFN distribution for both economies.
  - After the shock:
    - 50 percent probability that external financing needs in country A would exceed $19 billion, equivalent to 5.0 percent of GDP.
    - 50 percent probability that external financing needs in country B would exceed $342 billion, equivalent to 11.9 percent of GDP.
  - Tail (5.0 percent probability) under the shock:
    - EFNs will exceed $ 40 billion in economy A, i.e., 10.6 percent of GDP.
    - EFNs will exceed US$ 494 billion in economy B, i.e., 17.2 percent of GDP.
- Additional contextual notes from the source:
  - Floods are the single source of annual losses in economy B, costing about $7 billion every year.
  - A single weather-related event in economy A in 2013 killed more than 6,000 people, devastated nine regions, damaged 1.1 million homes, and triggered agricultural and infrastructure damages of $802 million.
  - The IPCC (2021) warns that economy B will suffer more frequent and intense heat and humid waves, extreme rainfall events and erratic monsoons, as well as cyclonic activities in the coming decades.
  - Climate change vulnerability for economies A and B have improved over the past decades according to the ND–GAIN index.

### Key implications highlighted
- Even economies representing about 5 percent of global GDP can, if highly interconnected and vulnerable, amplify climate change-related shocks and trigger global spillovers via trade and financial channels.
- The ND-GAIN index’s inclusion of readiness implies that economies with higher readiness could mitigate cross-border economic and financial effects by leveraging adaptation investments.
- The F@R framework can distinguish between normal and extreme circumstances (tail risks), enabling assessment of whether climate change is a macro-critical factor that might push an economy toward default-level external financing stress.

*Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022189-print-pdf.pdf*

### Annex 1    for more details.

### Annex 1

### V. Stress Testing the Global Economy to Climate Change-Related Shocks — Model and Mechanisms

- Model focus:
  - Explicitly captures impact of a climate change-related shock–induced default in country A and country B and quantifies systemic impact through global losses of international reserves.
  - Simulates balance of payments dynamics via trade, net foreign asset, and liability positions.
  - Incorporates external financing provided by the global financial safety net (GFSN): bilateral swap lines, regional financial arrangements (RFAs), and the IMF.
- Formal reserve dynamics (as presented):
  - Changes in foreign reserves, ∆RR_i,t, reflect trade revenues, net interest payments on foreign assets and foreign liabilities, exchange rate changes and net external financing from the GFSN. (Equation presented in source.)
  - Notation: a_ij denotes country i’s asset holding against country j; r_j,t is country j’s specific interest rate on its liabilities; e_j,t denotes the nominal effective exchange rate; EEF_i,t is the external financing provided by the GFSN.
- Default trigger and distance-to-default:
  - An economy is assumed to stop servicing its external debt when reserves reach 80 percent of the IMF’s assessed reserve adequacy (ARA) metric.
  - The total loss in international reserves for a creditor economy equals its exposure to the debt service amount not serviced: ∆RR_i = − a_i h r_h,t e_h,t (as presented).
- Amplification channels beyond direct exposures:
  - Risk premium reacts to the distance-to-default (how far reserves are from ARA); nonlinear relationship estimated between sovereign spreads and reserve adequacy is shown in Figure 5. The estimated fitted regression line is presented as:
    - 푅푅_{푖,푡} = 150 퐷퐷_{푡푡퐷퐷}^{1.8} + 60. 
  - Asset-price co-movement contagion: risk premium can react stepwise to changes in asset prices in economies falling into default; calibrated according to maximum changes in sovereign risk premia during the 2013 Taper Tantrum episode.
  - Sovereign spreads and climate vulnerability relationship (Figure 4) is estimated as:
    - 푅푅_{푖,푡} = 2151exp (−0.05퐺퐺퐺퐺퐺퐺푁푁).
- Domestic policy responses included in simulations:
  - Currencies allowed to depreciate by up to 10 percent per year.
  - Fiscal adjustment program equal to 0.7 percent of GDP over two years.
  - Multilateral policy responses captured by GFSN financing, calibrated economy-by-economy; IMF financing treated as provided by the General Resources Account (GRA) and instruments such as Stand-by Arrangements (SBA), Extended Fund Facility (EFF), Rapid Financing Instrument (RFI), Flexible Credit Line (FCL), and Precautionary Liquidity Line (PLL).

### A. Simulations — Tail Risk Scenario and Systemic Outcomes

- Scenario setup:
  - Tail risk scenario: an extreme climate change-related shock triggers defaults on external debt in country A and country B.
  - Results presented sequentially: contagion in number of economies defaulting, global reserve losses (in trillions of U.S. Dollars), and mitigation contributions from policy responses.
- Key simulation outcomes (as reported):
  - 'No policy' outcome (direct exposures + amplification channels, but no policy responses):
    - Number of economies defaulting: 26 economies may default on their debt service.
    - Global reserve losses: $ 1.8 trillion.
  - Domestic policy responses (currencies allowed depreciation and fiscal adjustment) reduce contagion and reserve losses:
    - Global reserve losses decline by about US$ 0.6 trillion compared to the ‘no policy’ scenario.
  - GFSN financing effects:
    - Financing provided by the GFSN excluding the IMF reduces global reserve losses by over US$ 0.3 trillion.
    - IMF financing further significantly reduces contagion and global reserve losses (substantial decline relative to ‘no policy’ scenario shown in figures).
  - Residual systemic impact after all financing avenues used:
    - An extreme climate change-related shock in economies A and B could push seven other economies into default and trigger a global reserves loss of US$ 800 billion even after all other financing avenues (domestic and external) have been utilized.
  - Aggregate policy effect:
    - When fully deployed, domestic and multilateral policies together can reduce global reserve losses due to the climate change-related shock by more than a half, to about $ 800 billion (from $ 1.8 trillion).
- Interpretation and implications from simulations:
  - Direct exposures alone generate relatively small global reserve losses; amplification channels (distance-to-default and asset price co-movements) substantially increase systemic impact.
  - Interconnectedness and contagion materially increase global costs of climate change-related shocks when large, interconnected economies default.
  - Adequate adaptation and mitigation policies, together with sustainable macroeconomic policies that maintain sufficient policy space, are necessary to manage spillovers.

### VI. Conclusion and Policy Implications — Key Findings and Recommendations

- Principal findings:
  - Extreme climate change-related shocks can have systemic economic and financial impacts on the global economy through spillovers and contagion, even when physical impacts are confined to individual economies.
  - In the absence of appropriate domestic and multilateral macroeconomic policy responses, global aggregate international reserve losses could reach $ 1.8 trillion in the simulated extreme scenario.
  - Full deployment of domestic and multilateral policies can reduce those losses by more than one-half, to about $ 800 billion.
  - Current macroeconomic estimates likely understate potential losses from climate change because they often omit indirect effects via spillovers and contagion.
- Policy implications and recommendations:
  - Economies should pursue adequate adaptation and mitigation policies to reduce vulnerability and the likelihood of events that could trigger economic and financial spillovers.
  - Maintaining sustainable macroeconomic policies to preserve policy space is important to respond to climate shocks and attendant spillovers.
  - Enhancing global financial safety nets (GFSN) — including bilateral swap lines, RFAs, and IMF facilities — can materially limit global losses from climate-induced sovereign distress.
  - Global collective efforts are warranted because climate mitigation and adaptation are global public goods; support to vulnerable countries to build capacity and readiness is essential.
  - Establishing common global principles and disclosure standards would strengthen assessment of climate risks and help control cross-country spillovers.

### Annex I. Data Sources

- Cross-country imports and exports: Direction of Trade Statistics (DOTS).
- Balance sheet positions against the rest of the world:
  - Interbank asset and liabilities positions: BIS Locational International Banking Statistics.
  - Portfolio investment positions: IMF Coordinated Portfolio Investment Survey (CPIS).
  - Foreign direct investment positions: IMF Coordinated Foreign Direct Investment Survey (CFDIS).

*Source: IMF Working Paper — Annex 1 (as provided).*

### Annex Table A.1. Vulnerable Economies Identified by Various Studies (Concluded)

### Annex Table A.1. Vulnerable Economies Identified by Various Studies (Concluded)

### Vulnerable economies identified (by study)
- ND GAIN (89 Economies):
  - Afghanistan, Bangladesh, Bhutan, Cambodia, Lao People’s Democratic Republic, Myanmar, Nepal, Papua New Guinea, Solomon Islands, Timor-Leste, Vietnam, Haiti, Honduras, Nicaragua, Djibouti, Somalia, Sudan, Tajikistan, Uzbekistan, Yemen, Benin, Burkina Faso, Burundi, Cameroon, Chad, Comoros, Democratic Republic of Congo, Republic of Congo, Cote D’Ivoire, Eritrea, Ethiopia, Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Madagascar, Mozambique, Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Sierra Leone, Tanzania, Togo, Uganda, Zambia, Zimbabwe.

- Global Climate Risk Index (98 Economies):
  - Afghanistan, Bangladesh, Burundi, Cambodia, Djibouti, Ethiopia, Haiti, Honduras, Kenya, Lao People’s Democratic Republic, Madagascar, Malawi, Mauritania, Moldova, Mozambique, Myanmar, Nepal, Niger, Nicaragua, Yemen, Sierra Leone, Solomon Islands, Tajikistan, Gambia, Uganda, Vietnam, Zimbabwe.

- IMF (64 Economies):
  - Afghanistan, Bangladesh, Cambodia, Kiribati, Myanmar, Nepal, Papua New Guinea, Solomon Islands, Timor-Leste, Vietnam, Haiti, Honduras, Nicaragua, Djibouti, Sudan, Tajikistan, Benin, Burkina Faso, Cameroon, Chad, Comoros, Gambia, Guinea, Guinea-Bissau, Kenya, Liberia, Madagascar, Mali, Mozambique, Niger, Sao Tome and Principe, Senegal, Sierra Leone, Zimbabwe.

### ND-GAIN over time (Annex Figure 1)
- Sources: University of Notre Dame Global Adaptation Index (ND-GAIN) and IMF staff calculations.
- Figure panels:
  - Panel A: ND-GAIN of Economies A and B Over Time (ND-GAIN INDEX expressed as deviation from sample average, years: 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 shown with scale -10 to -5 and 0 to 80 in separate displays).
  - Panel B: ND-GAIN of All Economies Over Time (ND-GAIN INDEX, years 2000–2017 shown).
- Numeric fidelity preserved as presented in figures.

### Robustness check using frequency of climate-related natural disasters (Annex II)
- Alternative vulnerability indicator: frequency of climate related natural disasters from the IMF Climate Change Dashboard (annual frequency economy-level data on drought, extreme temperature, flood, landslide, and storm).
- Method:
  - Investigate all types of natural disasters and compute the annual average number of natural disasters for each country.
  - Replot Figure 1 using number of climate related natural disasters instead of ND-GAIN.
- Key finding:
  - As Figure A.1 shows, three large economies A, B, and C are highly vulnerable to climate change.
  - These economies are large in economic size measured by GDP and are over the 95th percentile of the distribution of the number of natural disasters.
  - Panel B shows that of these three economies, also display a substantial degree interconnectedness, confirming the robustness of the results using ND-GAIN.
- Annex Figure A.2:
  - Panel A: Economic Size and Climate Change Vulnerability (economic size captured by GDP from the World Bank data).
  - Panel B: Financial Interconnectedness and Climate Change Vulnerability (interconnectedness indices correspond to the geometric average of the degree and strength of the network (see Porter et al, 2021)). All figures expressed as deviation from sample average.

### External financing needs-at-risk: Quantile regressions (Annex III)
- Reports quantile regressions for external financing needs-at-risk (F@R) for Country A and Country B.
- Annex Figure 2 and Annex Figure 3 present Country A and Country B quantile regression coefficient estimates for F@R (notes: see text, Section IV, for definitions and details).

### Multilayered network model: structure and mechanisms (Annex IV)
- Purpose:
  - Examine how economic and financial contagion spread through the global economy when economies deemed highly vulnerable suffer exogenous shocks to the return on their external liabilities.
  - Capture spillovers from direct trade and financial exposures, amplification via risk premia and asset price comovements, cascading effects, domestic policy responses, and tapping of GFSN layers.
- Formal setup (key model elements and notation preserved):
  - Each economy i is endowed with foreign exchange reserves at t = 0 in amount of RR_{i0} dollars.
  - Change in foreign exchange reserves of economy i at time t, ∆RR_{i,t}, is given by the expression combining trade terms, cross-border asset/liability positions, interest rates r_{j,t}, nominal effective exchange rates e_{i,t}, asset holdings a_{ij}, and external financing from GFSN layers EEEF_{it} = EEEF_{BIGA;i,t} + EEEF_{RGA;i,t} + EEEF_{IIFG;i,t}.
  - Economy i’s asset holdings and liabilities a_{ij} include FDI, portfolio investment, and cross-border interbank positions.
  - In a liquidity crisis triggered by a shock to economy h’s external liabilities (unable to fulfill interest payments a_{ih,t} r_{h,t}), international reserves decline for all economies exposed to h.
  - Initial financing need for a non-defaulting economy i is ∆RR_{i,t} = − a_{ih,t} r_{h,t} e_{h,t}.
  - If reserves are depleted or fall below a threshold, the economy cannot fulfill some external payments and external financing needs exceed available resources; it can resort to other GFSN layers and the residual is a potential call for Fund resources.
- Amplification mechanisms:
  - Sovereign risk premia: sensitive to distance-to-default measured by stock of international reserves minus reserve adequacy (ARA); as reserves approach the ARA threshold, sovereign risk premium increases.
  - Asset price comovements: calibrated using interest rate correlations and CDS shifts during periods of financial stress (e.g., Taper Tantrum).
- Policy and GFSN features:
  - Monetary policy: currencies can adjust by up to 15 percent cumulatively during the first two years in which an economy sees its counterparts not servicing debt obligations.
  - Fiscal policy: allows a once-and-for-all fiscal adjustment program equal to 1 percent of GDP; the fiscal adjustment improves the trade balance by 0.5 percent of GDP through reduced imports.
  - GFSN layers drawn first: bilateral swap lines (EEEF_{BIGA;i,t}) and RFAs (EEEF_{RGA;i,t}) when available; drawdowns support reserve accumulation up to limits determined by maximum access in each GFSN layer.
  - RFAs included: European Stability Mechanism, European Financial Stability Facility, European Financial Stabilization Mechanism, EU Balance of Payments Assistance Facility, Chiang Mai Initiative Multilateralization, BRICS Contingent Reserve Arrangement, Latin American Reserve Fund, Arab Monetary Fund, Eurasian Fund for Stabilization and Development.
  - Demand for Fund resources is the residual after accounting for these sources.

### Data and coverage of the network model
- Data year and coverage:
  - Uses 2019 data for 63 economies (36 advanced and 27 emerging market) representing about 85 percent of global GDP.
- Data inputs:
  - (i) cross-economy imports and exports from Direction of Trade Statistics (DOTS);
  - (ii) interbank asset and liability positions from BIS Locational International Banking Statistics;
  - (iii) portfolio investment positions from IMF Coordinated Portfolio Investment Survey (CPIS);
  - (iv) foreign direct investment positions from IMF Coordinated Foreign Direct Investment Survey.
  - Additional inputs: nominal bilateral exchange rates, sovereign spreads, interest rates, foreign exchange reserves, and access to bilateral swap lines, RFAs, and Fund financing.
- Data sources include Haver, Bloomberg, IMF International Financial Statistics, the U.S. Federal Reserve System, and RFAs websites.

*Stress Testing the Global Economy to Climate Change-Related Shocks in Large and Interconnected Economies, Working Paper No. WP/2022/189*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022189-print-pdf.pdf_