## wpiea2020003-print-pdf

## Source details

**Canonical URL:** [wpiea2020003-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020003-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020003-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020003-print-pdf.pdf.json)

---

### Introduction: role of disaster insurance
- Natural disasters are emerging as a macro-critical risk, especially in small, low-income and other disaster-vulnerable states.
- Ex-ante disaster risk management focuses on:
  - (i) reducing exposure through more resilient infrastructure;
  - (ii) transferring risk through insurance;
  - (iii) managing retained risk ex-ante through self-insurance and contingency financing.
- Insurance spreads expected and potentially large one-off costs through smaller annual payment of insurance premia, reducing volatility of financial losses and mitigating macroeconomic implications for households and businesses.
- Literature note: uninsured losses account for strong negative effects of disasters on economic activity, while well insured disasters can be inconsequential for growth (e.g. BIS, 2012).

### Purpose, methodology, and simulation design
- Focus: sovereign natural disaster insurance as a tool to manage financial costs and assess adequacy relative to government preferences over debt and growth.
- Approach:
  - Draw disaster shocks from model-based probability distributions and industry estimates of economic and fiscal costs.
  - Run stochastic simulations comparing a no-insurance scenario (disaster costs financed ex-post through borrowing) with alternative insurance packages using the same disaster shocks.
  - Compare debt and output dynamics to identify trade-offs between debt sustainability and growth.
- Simulation algorithm (exact parameters preserved):
  - Natural disaster simulations: two sampling methods; reported results use Loss Exceedance (LE) curve mapping uniform[0,1].
  - Insurance packages: combinations of
    - five attachment points, ranging from return periods of 5 to 25;
    - eleven exhaustion points, ranging from return periods 50 to 300;
    - ten ceding percentage options, ranging from 10 to 100 percent.
  - Monte Carlo design: four steps repeated 1000 times.
  - Growth dynamics coefficients:
    - a = -0.3 (negative contemporaneous effects)
    - b = 0.1 (positive lagged effects)
    - potential growth adjustment includes coefficient 0.1 on uncovered losses
  - Government post-disaster financing limit: bG...t,i = 5% of GDP for both smaller and larger country.
  - Debt dynamic equation preserved as presented:
    - d_t,i ≡ [(1+g_t,i)^(-1)] [ (1+r) d_{t-1,i} − p_t,i ]
  - Primary balance:
    - p_t,i ≡ p_MMB − premium_{t,i} − min[ disaster-related expenditures within access limits, bG..._{t,i} ]
  - Interest rate response to large debt increases written as:
    - b_t,i = b_{t-1,i} + 0.0000*(d_{t-1,i} − d_{t-2,i})
  - Baseline potential growth referenced to IMF’s WEO projection as of 2023.
- Government utility and selection:
  - Utility: U_i = (1−z) * [function of growth outcomes with coefficient ρ] + z * Δd_i
  - ρ = 1.2 (baseline coefficient of constant risk aversion)
  - z = 0.5 (baseline weight on debt sustainability)
  - Reporting: analysis of top 50 packages by government utility rather than a single optimum.

### Current landscape and nature of exposure (key statistics)
- Disasters can cause large and long-lasting macroeconomic effects: near-term and longer-term growth declines, capital destruction, and higher public debt.
- Example: Acevedo (2016) — average hurricane in the Caribbean causes cumulative decline in GDP per capita of 4.4 percent over seven years.
- Relative annual damages (1980–2017 averages, as cited):
  - Small Caribbean countries: 3.1 percent of GDP a year on average.
  - Pacific small states: around 2 percent of GDP a year on average.
  - Larger states: 0.3 percent of GDP a year on average.
- Top single-disaster intensity examples (1980–2017): Dominica 2017 Storm 374 (% of GDP), Grenada 2004 Storm 184, Maldives 2004 Earthquake 179 (reported as sample entries).
- Current insurance penetration examples:
  - Belize and Grenada: traditional indemnity insurance covered only 4.5 percent of total damage in recent large disasters.
  - Munich Re estimates for 2010-16:
    - Cuba and Haiti: less than 2 percent of natural disaster losses insured.
    - El Salvador: less than 6 percent insured.
    - Jamaica: around 10 percent insured.
- Caribbean countries (2017-18) insured on average some 35 percent of estimated losses to government assets:
  - 50 percent for earthquakes
  - 38 percent for hurricanes
  - 21 percent for excess rainfall
- CCRIF current policy limit: US$100 million per hazard per year.
- ARC example: US$30 million per country per season for drought events that occur with a frequency of 1 in 5 years or less.

### Taxonomy of instruments and regional pooling
- Pre-arranged instruments with immediate disbursement: self-insurance via fiscal buffers, risk-transfer to insurance/capital markets (catastrophe bonds), pre-arranged contingent loans from IFIs/central banks.
- Ex-post options: market or official borrowing, grants, humanitarian assistance.
- World Bank multi-layer risk approach: self-insurance for smaller/frequent disasters → contingent credit lines → insurance → catbonds for most infrequent/severe disasters (example figure uses US$100 million).
- Regional pools described (avg. premium income / avg. coverage preserved):
  - CCRIF (2007): Avg. premium income/Avg. coverage: US$21.5m / US$650m; hazards: Earthquake, Tropical cyclone, Excess rainfall, Drought; insured members (21); other eligible (14).
  - PCRAFI (2013): Avg. premium income/Avg. coverage: US$2m / US$45m; hazards: Tropical cyclone, Earthquake/tsunami, Excess rainfall; insured members (5); other eligible (10).
  - ARC (2013): Avg. premium income/Avg. coverage: US$22m / US$50m; hazards: Drought, Extreme weather; insured members (6); other eligible (6).
  - SEADRIF: agreement signed December 2018; premium/coverage: TBD.
- Advantages: lower cost parametric coverage via larger portfolios, country-specific risk models, reduced administrative costs; quick payouts (CCRIF payouts in 14 days or less); policy holders as owners.
- Challenges: pools cannot reliably rely on fresh capital inflows; small size (about US$750 million overall average annual coverage) raises reinsurance premia; parametric basis risk.

### Catastrophe bonds and state-owned insurers
- Catbonds: mechanism similar to parametric insurance that taps capital markets; sovereign uptake increasing but limited by sophistication and setup costs.
- Issuers to date include Mexico and Turkey; World Bank issued first joint sovereign catbond for Pacific Alliance (Chile, Colombia, Mexico, Peru).
- State-owned insurers used where local private markets are inefficient or prohibited; examples: Sri Lanka’s Insurance Corporation (SLIC), Costa Rica’s Instituto Nacional de Seguros, the Philippines Government Service Insurance System (GSIS).

### Case example — Philippines Catastrophe Risk Insurance Program
- Introduced 2017 as part of Disaster Reduction Financing and Insurance Strategy with World Bank and UK DFID support.
- GSIS provides protection to national government and 25 participating local governments.
- Coverage: up to US$206 million (in local currency) against typhoons and earthquakes with parametric triggers.
- GSIS transfers risk to private international reinsurers via competitive bidding; the World Bank Group acts as intermediary.

### Drivers of low coverage and parametric design trade-offs
- Drivers of low sovereign insurance coverage:
  - Low perceived vulnerability due to rarity of catastrophic events.
  - Cost: parametric insurance and catbonds cost estimated in the range of 1.2-  3.2 times the expected annual payout.
  - Basis risk: parametric triggers may not align with actual loss locations.
  - Political economy: trade-offs between paying premia and financing development needs; expectation of government assistance ("charity hazard").
- Parametric parameter choices:
  - Attachment point: events expected once in 5 to 30 years.
  - Exhaustion point: coverage capped at levels estimated for once in 50-to-140-year events.
  - Ceding percentage: share of risk ceded between attachment and exhaustion points.
- Observation: low exhaustion points and CCRIF US$100 million cap imply payouts cover a small proportion of losses under larger disasters; parametric insurance intended primarily for emergency liquidity.

### Core analytical results on trade-offs (preserve numeric examples)
- Result 1 (systemic exposure and tail risks):
  - Example: government ex-ante expected losses of 0.1-0.4 percent of GDP on average; an illustrative expected loss of 0.4 percent of GDP a year can correspond to realized government losses of 16-46 percent of GDP in scenarios.
- Result 2 (debt-growth tradeoff):
  - More coverage → faster rebuilding and stronger growth but higher premia and fiscal costs.
  - Less coverage → smaller premium-driven debt increases but limited protection in realizations.
- Result 3 (size and exposure differences):
  - Small, more exposed countries: larger shares of GDP at risk; higher premia and payouts as shares of GDP; tend to favor less costly packages.
  - Larger countries: often choose maximal risk transfer where payout limits do not bind.
- Result 4 (insurance multiple and value for money):
  - Insurance multiple = premium / expected payout.
  - Packages with lowest insurance multiple provide best value-for-money and lie on the efficient frontier of debt-growth tradeoffs.
- Result 5 (insuring smaller, more frequent disasters):
  - Within a given price range, packages with lower deductibles (insuring smaller disasters) tend to have the lowest insurance multiples and improve debt-growth tradeoffs.
- Result 6 (risk aversion and prohibitive optimal transfer):
  - Higher risk aversion to negative growth yields more growth-biased (and potentially higher-cost) optimal transfers.
  - Simulation sensitivity: baseline ρ = 1.2; alternative example ρ = 3.5 discussed.
- Result 7 (preferences over growth vs debt):
  - Higher weight on fiscal sustainability (example z = 0.9) → choose less costly packages.
  - Balanced preferences (z = 0.5) → intermediate choices.
  - Growth-prioritizing (z = 0.1) → choose more expensive, higher-coverage packages.
- Result 8 (price discounts and donor support):
  - Halving the insurance multiple (proxy for donor subsidy) leads to:
    - (i) smaller increases in debt from insurance;
    - (ii) improved tradeoffs (same debt increase yields larger growth gains; smaller “sacrifice ratio”);
    - (iii) shift toward more growth-biased packages with higher ceding and wider insured windows.
- Result 9 (borrowing constraints and insurance value):
  - Tighter post-disaster borrowing limits (example scenarios: 5% of GDP vs 10% of GDP) increase the value of insurance and steepen the tradeoff slope, making insurance more attractive.

### Policy implications and options
- Evaluate insurance on tail risks and potential catastrophic fiscal impacts, not only expected losses.
- Optimal insurance choice depends on:
  - country size, exposure, fiscal space, payout limits, borrowing capacity, and preferences over debt versus growth.
- Donor-subsidized premiums or discounts can materially improve growth-debt tradeoffs and enable higher-coverage packages.
- Where payout limits bind (e.g., CCRIF US$100 million cap), consider:
  - raising effective payout ceilings via recapitalization or risk-sharing among pools,
  - expanding pool capital to lower reinsurance reliance and premia.
- Expand private property insurance penetration to reduce sovereign contingent liabilities via:
  - developing local insurance markets and regulation,
  - regional pooling of private risks,
  - government-sponsored national pools with international reinsurance.
- Improve trigger measurement and design to reduce basis risk:
  - more granular geographic triggers,
  - complement parametric triggers with indemnity-type or industry-loss triggers.
- Consider issuance of catbonds at regional scale and introduction of hurricane clauses in sovereign bonds (debt repayment stand-still) as alternative contingent financing.

### Operational recommendations for implementing insurance strategies
- Integrate risk transfer decisions within a Disaster Resilience Strategy (DRS) to balance risk retention, risk transfer, mitigation, and macro-fiscal objectives.
- Scale and finance regional pools by:
  - diversifying membership and perils;
  - recapitalizing pools (donor contributions, trust funds) to reduce reinsurance costs;
  - sharing risks among pools while avoiding cross-subsidization (portfolio segregation).
- Use donor support conditionally to avoid dependency:
  - conditionality examples: IMF program participation, sound macroeconomic policies, matching country purchases, commitments to resilient investment.
- Prioritize packages with low insurance multiples to maximize expected payouts for given premia.

### Areas for further analytical work
- Optimal allocation of fiscal space between financial resilience (insurance) and structural resilience (resilient infrastructure).
- Detailed trade-offs between purchasing insurance and self-insuring (market borrowing).
- Expanding the analysis across a broader range of countries to map how optimal risk transfer depends on country-specific characteristics.

*Source: wpiea2020003-print-pdf — References and sections 17, 25, 35.*

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

### wpiea2020003-print-pdf - References .............................................................................................................

### Introduction: role of disaster insurance
- Natural disasters are emerging as a macro-critical risk, especially in small, low-income and other disaster-vulnerable states.
- Ex-ante efforts to manage disaster risk focus on: (i) reducing exposure through more resilient infrastructure; (ii) transferring risk through insurance; (iii) managing retained risk ex-ante through self-insurance and contingency financing.
- Insurance does not reduce physical losses but spreads expected and potentially large one-off costs through smaller annual payment of insurance premia, reducing volatility of financial losses and mitigating macroeconomic implications for households and businesses.
- Literature suggests uninsured losses account for strong negative effects of disasters on economic activity, while well insured disasters can be inconsequential for growth (e.g. BIS, 2012).

### Purpose and methodology of the paper
- Focus: sovereign natural disaster insurance as a tool to manage financial costs of disasters and assess adequacy of insurance relative to government preferences over debt and growth.
- Approach:
  - Draw disaster shocks from model-based probability distributions of disasters of various strengths and industry estimates of economic and fiscal costs.
  - Run stochastic simulations comparing:
    - A scenario without insurance (disaster costs financed ex-post through borrowing).
    - Alternative insurance packages using the same disaster shocks.
  - Compare debt and output dynamics to illustrate benefits and costs and identify trade-offs between debt sustainability and growth.

### Key findings on trade-offs and country heterogeneity
- Choosing higher risk transfer:
  - Provides higher protection to growth by enabling faster recovery.
  - Entails higher fiscal costs (higher premia and payouts).
- Policy trade-offs:
  - Prioritizing fiscal sustainability → choose less costly insurance packages with lower payouts and less beneficial growth outcomes.
  - Prioritizing growth outcomes → choose more expensive insurance packages with higher payouts.
- Country-specific considerations:
  - In smaller and more vulnerable countries, natural disasters carry a higher social cost, inducing higher government risk aversion and greater benefits from insurance.
  - Cost of significant insurance coverage is higher in these countries, often producing suboptimal insurance choices.
  - International support can loosen fiscal constraints and help scale up risk transfer to improve growth outcomes.

### The current landscape — nature of exposure
- Disasters can cause large and long-lasting macroeconomic effects: near-term and longer-term growth declines, capital destruction, and higher public debt, perpetuating high debt-low growth cycles.
- Example estimate: Acevedo (2016) — the average hurricane in the Caribbean causes a cumulative decline in GDP per capita of 4.4 percent over seven years.
- Relative annual damages (1980–2017 averages, as cited):
  - Small Caribbean countries: 3.1 percent of GDP a year on average.
  - Pacific small states: around 2 percent of GDP a year on average.
  - Larger states: 0.3 percent of GDP a year on average.
- Single-disaster intensity: for some countries, a single natural disaster caused economic losses of multiples of GDP (Top Ten Natural Disasters: 1980-2017 list included country-year-event-damage entries with values such as Dominica 2017 Storm 374 (% of GDP), Grenada 2004 Storm 184, Maldives 2004 Earthquake 179, etc.).

### Implications of systemic damages in smaller states
- Larger contingent liabilities for sovereigns: direct costs (public property and infrastructure) and implicit contingent liabilities from pressures to assist uninsured households.
- Higher social costs of damages imply higher sovereign risk aversion and higher social benefits of insurance, especially when private sector damages are highly correlated.
- Missing local risk markets: highly exposed countries with shallow/weak financial and insurance systems may face missing risk markets where risks are not underwritten at any price, motivating government intervention or regional solutions.

### Implicit contingent liabilities (Box 1 highlights)
- Low- and lower middle-income households and small businesses are often uninsured or underinsured due to high premia, lack of suitable products, unfit construction, lack of insurance tradition, or expectation of government assistance ("charity hazard").
- Empirical coverage examples:
  - Belize and Grenada: traditional indemnity insurance of physical assets reportedly covered only 4.5 percent of the total damage in recent large disasters.
  - Munich Re estimates for 2010-16:
    - Cuba and Haiti: less than 2 percent of natural disaster losses insured.
    - El Salvador: less than 6 percent insured.
    - Jamaica: around 10 percent insured.
- Inadequate private-sector insurance delays recovery and reduces tax revenues, especially where tax incidence concentrates in specific industries (e.g., tourism in the Caribbean).

### Taxonomy of risk-transfer instruments used by sovereigns
- Pre-arranged instruments that disburse immediately after disasters include:
  - Self-insurance via fiscal buffers.
  - Risk-transfer to insurance or capital markets (e.g., catastrophe bonds).
  - Pre-arranged contingent loans from international financial institutions or central banks.
- Ex-post options: market or official borrowing, grants, and humanitarian assistance.

### World Bank multi-layer risk approach
- Objective: determine most cost-efficient combination of instruments to achieve a pre-determined level of coverage (example figure uses US$100 million).
- Prioritization by cost and timeliness: self-insurance for smaller/frequent disasters → contingent credit lines → insurance → catbonds for most infrequent/severe disasters.
- Sovereigns often combine instruments into multi-instrument strategies.

### Regional risk pooling facilities (overview and examples)
- Three well-established regional pools:
  - CCRIF (2007): Caribbean Catastrophe Risk Insurance Facility Segregated Portfolio Company (extended to Central America in 2015).
    - Hazards insured: Earthquake, Tropical cyclone (hurricanes), Excess rainfall, Drought.
    - Insured members (21) and other eligible members (14).
    - Avg. premium income/Avg. coverage: US$21.5m / US$650m.
  - PCRAFI (2013): Pacific Catastrophe Risk Assessment and Financing Initiative insurance program (established 2013).
    - Hazards insured: Tropical cyclone, Earthquake/tsunami, Excess rainfall.
    - Insured members (5): The Cook Islands, the Marshall Islands, Samoa, Tonga, Vanuatu.
    - Other eligible members (10).
    - Avg. premium income/Avg. coverage: US$2m / US$45m.
  - ARC (2013): African Risk Capacity LTD, specialized agency of the African Union.
    - Hazards insured: Drought, Extreme weather (drought, excess rainfall, heatwaves and tropical cyclones).
    - Insured members (6): Burkina Faso, Mali, Mauritania, Niger, Senegal, The Gambia.
    - Other eligible members (6).
    - Avg. premium income/Avg. coverage: US$22m / US$50m.
- SEADRIF: agreement signed in December 2018 to establish the Southeast Asia Disaster Risk Insurance Facility (mainly flood risk); signatories include Cambodia, Indonesia, Japan, Lao, Myanmar, Singapore; premium/coverage: TBD.
- Note: regional pools have benefited from donor support for start-up, capitalization, and sometimes partial premium financing.

### Advantages of regional pools
- Lower cost parametric coverage relative to individual purchases due to larger aggregate portfolios, country-specific risk models, and reduced administrative costs.
- Quick payouts that help maintain essential government functions (CCRIF payouts in 14 days or less).
- Policy holders are owners of the facility (CCRIF, PCRAFI, ARC), allowing benefits via dividends or lower premia and avoiding private-profit conflicts.

### Challenges facing regional risk pooling
- As non-private entities, pools cannot reliably rely on fresh capital inflows; options include transferring risk to reinsurance markets at cost, donor contributions, or limiting payout sizes.
- Small size of existing pools (about US$750 million overall in terms of average annual coverage) increases reinsurance and insurance premia.
- Parametric insurance exposes countries to basis risk and may need to be combined with indemnity-based instruments.

### Catastrophe bonds (catbonds)
- Catbonds are an emerging sovereign risk-transfer instrument, functionally similar to parametric insurance but tapping capital markets.
- Sovereign uptake is increasing but remains limited due to sophistication and high setup costs.
- Issuers to date include Mexico and Turkey for individual catbonds; the World Bank issued the first joint sovereign catbond for Pacific Alliance members (Chile, Colombia, Mexico and Peru), achieving cost savings and record-low premium rates due to investor demand for diversification.
- Affordability has improved as coupons have come down and World Bank Treasury has offered lower setup costs.

*Source: wpiea2020003-print-pdf - References .............................................................................................................*

### 17.      Many sovereigns rely (fully or partially) on state-owned insurance companies for

### 17.      Many sovereigns rely (fully or partially) on state-owned insurance companies for

### State-owned insurers and examples
- State-owned insurance companies are used because insuring public assets through local private markets is often inefficient due to small scale and lack of competition, and/or explicit prohibition of hiring private insurance brokers for the public sector.
- Examples of state-owned insurance companies named:
  - Sri Lanka’s state-owned Insurance Corporation (SLIC)
  - Costa Rica’s Instituto Nacional de Seguros
  - The Philippine’s Government Service Insurance System (GSIS) covering national government assets and local governments

### Box 2 — Philippines: Catastrophe Risk Insurance Program (summary)
- Introduced in 2017 as part of the Disaster Reduction Financing and Insurance Strategy, with support from the World Bank Group and UK’s DFID.
- GSIS (government-owned) provides protection to the national government and 25 participating local governments.
- Coverage: up to US$206 million (in local currency) against typhoons and earthquakes, with parametric triggers for payouts.
- GSIS transfers risk to private international reinsurers via competitive bidding; the World Bank Group acts as intermediary.
- The program complements natural disaster-related reserves and contingency credit lines and was developed alongside the Department of Finance’s Disaster Risk Finance Strategy and the country’s first catastrophe risk model.

### Current levels of risk transfer (observations)
- Many sovereigns have chosen relatively low levels of insurance.
- Example: In 2017-18, Caribbean countries insured on average some 35 percent of estimated losses to government assets from different hazards:
  - 50 percent for earthquakes
  - 38 percent for hurricanes
  - 21 percent for excess rainfall
- Parametric insurance parameter choices described:
  - Attachment point: events expected once in 5 to 30 years (trigger/attachment point)
  - Exhaustion point: coverage capped at levels estimated for once in 50-to-140-year events
- CCRIF current maximum payouts noted:
  - Maximum payout limits: US$100 million per hazard (CCRIF)6
  - ARC example: US$30 million per country per season for drought events that occur with a frequency of 1 in 5 years or less6
- Observation: Low exhaustion points imply that payouts cover a small proportion of losses under larger disasters; parametric insurance is intended for emergency liquidity rather than full loss coverage.

### Box 3 — Decisions Faced by Sovereigns in Choosing Parametric Insurance (based on CCRIF)
- Countries choose three parameters:
  - Attachment point: severity at which coverage is triggered (akin to a deductible)
  - Exhaustion point: disaster threshold at or above which the insurance pays only the maximum payout
  - Ceding percentage: share of risk between attachment and exhaustion points ceded to insurer
- Payouts are model-based (e.g., based on wind speed triggers) and facilitate rapid disbursement but can mismatch actual losses (basis risk).
- CCRIF policy limit: US$100 million per hazard per year, which can become binding before exhaustion point—likely for larger countries.

### Drivers of relatively low insurance coverage
- Low perceived vulnerability: catastrophic events are rare, leading to low perceived vulnerability until disasters occur.
- Cost considerations:
  - Parametric insurance and catbonds cost is estimated to be in the range of 1.2-  3.2 times the expected annual payout.
  - Ex-post, insurance payout often far exceeds the annual premium in most disasters, especially larger ones.
  - Given higher frequency of smaller disasters, many sovereigns experience smaller or no payouts depending on chosen parameters.
- Basis risk: parametric triggers may not align with locations of main impact; actual losses can exceed modelled losses.
- Political economy factors: trade-offs between paying premiums and financing development needs; risk that damaging events may not trigger payouts.

### Question of adequacy and criterion for adequacy
- Adequacy depends on the criterion used; BIS (2012) judges coverage adequate if it produces no output loss.
- This paper uses a criterion: expected payouts (minus paid premiums) should help countries avoid significant deterioration in debt dynamics while offering some protection for growth.
- The assessment accounts for country-specific exposure, government risk preferences, and actual insurance options.

### Methodology for identifying debt and output implications (overview)
- Approach: stochastic simulations of disaster incidence and ensuing debt and output dynamics over a 10-year horizon.
- Data and shocks:
  - Current insurance choices, costs, and payouts based on CCRIF information.
  - Shocks simulated using catastrophe risk models by CCRIF.
- Range of insurance packages:
  - About 550 insurance packages considered, covering wide parametric combinations.
- Country coverage in simulations:
  - Two representative countries:
    - a smaller country heavily exposed to natural disasters (“smaller country”)
    - a somewhat larger country less exposed (“larger country”)

### Simulation algorithm (steps and exact parameter ranges)
- Step 1: natural disaster simulations
  - Two sampling methods: (i) draw from Loss Exceedance (LE) curves via uniform[0,1] mapping; (ii) non-parametric kernel density estimation from LE data. Results qualitatively similar; reported results use the first method.
- Step 2: insurance cost and payouts
  - Insurance packages defined by combinations of:
    - five attachment points, ranging from return periods of 5 to 25;
    - eleven exhaustion points, ranging from return periods 50 to 300;
    - ten ceding percentage options, ranging from 10 to 100 percent.
  - Each package has ex-ante known payouts and premiums computed from probability-weighted payouts.
- Step 3: output path simulation
  - Growth dynamics specified (as in source) with coefficients:
    - a = -0.3 (negative contemporaneous effects)
    - b = 0.1 (positive lagged effects)
  - Potential growth adjustment includes coefficient 0.1 on uncovered losses; ex-post borrowing by government is limited and enters potential growth.
  - Baseline potential growth projection referenced to IMF’s WEO projection as of 2023.
  - Government post-disaster financing limit set at 5% of GDP for both smaller and larger country:
    - bG...t,i = 5% of GDP (explicitly stated)
- Step 4: debt path simulations
  - Conventional debt dynamic equation used:
    - d_t,i ≡ [(1+g_t,i)^(-1)] [ (1+r) d_{t-1,i} − p_t,i ] (equation as presented in source)
  - Primary balance p_t,i construct:
    - p_t,i ≡ p_MMB − premium_{t,i} − min[ disaster-related expenditures within access limits, bG..._{t,i} ]
  - Real interest rate assumed affected by large increases in debt via:
    - b_t,i = b_{t-1,i} + 0.0000*(d_{t-1,i} − d_{t-2,i}) (as stated)
  - Assumptions: natural disaster losses affect only current primary balance; no differentiation between domestic and foreign currency debt.
- Monte Carlo design:
  - The four steps repeated 1000 times to obtain distributions of disasters, debt, and growth under varied states of the world.

### Government utility and selection criterion
- Government assumed risk averse regarding growth, risk neutral regarding financial outcomes.
- Utility function (additive constant risk aversion) specified:
  - U_i = (1−z) * [function of growth outcomes with coefficient ρ] + z * Δd_i
  - ρ (coefficient of constant risk aversion) assumed equal to 1.2 (sensitivity discussed in source)
  - z (weight on debt sustainability) assumed at 0.5 in baseline (sensitivity discussed in source)
- Growth outcome metric:
  - ΔG_i = difference (averaged over 1000 simulations) between average growth over the 10-year strings under package i and average growth under no-insurance for the same disaster strings.
- Debt outcome metric:
  - Δd_i = difference (averaged over 1000 simulations) between debt ratio in year 10 under package i and debt ratio in year 10 under no-insurance.
- Reporting approach:
  - Rather than a single package, the top 50 packages (by government utility) are analyzed to parse their characteristic features.

*Source: Excerpt from wpiea2020003-print-pdf — section 17 and accompanying boxes and methodology.*

### 25.      In choosing how much natural disaster risk to transfer to insurance, countries face a

### wpiea2020003-print-pdf - 25.      In choosing how much natural disaster risk to transfer to insurance, countries face a

### Summary of core trade-off
- Countries face a trade-off between the implications of insurance for debt and for growth; the optimal risk transfer varies with country size, exposure to disaster, fiscal space, and risk aversion to growth losses.
- Decisions for larger (less exposed) countries differ materially from those for smaller and more exposed countries.

### Result 1 — Systemic exposure and tail risks
- Where a large share of output and capital are at risk from low frequency events (natural disaster effects could be systemic), governments should guard against tail risks and not just expected average loss.
- Example statistics (Caribbean states): government ex-ante expected losses of 0.1-0.4 percent of GDP on average; a stated illustrative expected loss of 0.4 percent of GDP a year ex-ante can correspond to actual government losses of 16-46 percent of GDP in realization scenarios.
- Implication: insurance decisions based solely on expected value of loss can lead to suboptimal financial protection and catastrophic ex-post outcomes that derail development.

### Result 2 — Debt-growth tradeoff of insurance coverage
- More insurance coverage supports faster rebuilding and thus supports growth but requires higher premia, increasing borrowing and fiscal costs.
- Less coverage reduces premium-driven debt increases but provides limited protection for growth and debt in disaster realizations.
- The tradeoff depends on country preferences over fiscal costs versus growth and on limits to insurance payouts.

### Result 3 — Size and exposure differences
- Insurance is relatively more impactful in smaller, more exposed countries because disaster losses and premia are larger shares of GDP, increasing debt, while payouts and growth-mitigation benefits are also larger.
- Optimal risk transfer can differ:
  - Small, more exposed countries tend to favor less expensive/less coverage packages (bias toward lower cost).
  - Larger countries tend to choose packages with larger coverage (more expensive but more protective of growth).
- Payout limits (example: US$100 million CCRIF limit) can bind for larger countries, limiting premiums and shifting optimal choices toward maximum growth protection.
- Within optimal packages:
  - Smaller countries: can choose either (i) narrower insured windows but cede a larger share of risk to insurer, or (ii) wider insured windows but limit ceded risk.
  - Larger countries: optimal insurance often implies maximum risk transfer with high insured ranges and high degree of risk transfer.

### Result 4 — Insurance multiple and value for money
- The best growth-debt tradeoffs are offered by packages with the lowest insurance multiple (cost relative to expected payout).
- Every package has a fiscal cost (premium) and fiscal benefit (expected payout = predetermined payout × probability of state).
- If chosen on cost alone, countries pick lowest premium packages (higher deductible, lower maximum payout, or lower ceding percentage).
- Optimal (utility-maximizing) packages are not necessarily the cheapest; for larger countries they may be among the most expensive.
- Packages with the lowest insurance multiple yield the highest expected payouts for a given premium and lie on the efficient frontier of debt-growth tradeoffs.

### Box 4 — The Insurance Multiple
- Definition: insurance multiple = premium / expected payout (how much one pays for every dollar of expected payouts).
- For a given premium there exist packages that provide higher expected payout; these have lower multiples (best value: green, followed by yellow in figures).

### Result 5 — Insuring smaller (more frequent) disasters
- Under existing options, insuring smaller disasters (lower deductible) can improve debt-growth tradeoffs because the packages with lowest insurance multiples tend to trigger at more frequent disasters.
- While conventional advice emphasizes insurance for medium-to-large disasters as cost-efficient, within a given price range the most value-for-money packages are those with the lowest deductible.
- For a chosen insured window, it is more cost-efficient to start coverage at lower-frequency disasters.

### Result 6 — Risk aversion and prohibitive optimal transfer
- Higher risk aversion to negative growth implies a more growth-biased (but potentially prohibitive-cost) optimal risk transfer.
- Smaller, highly exposed countries with higher risk aversion will seek higher coverage and more growth protection (packages that cede large shares of losses and have wide insured windows), but these can carry prohibitive costs and large debt implications.
- Empirical simulation note: risk aversion examples shown with different coefficients (baseline 1.2 and an alternative 3.5).

### Result 7 — Preferences over growth versus debt
- Countries prioritizing fiscal sustainability choose less costly packages with lower payouts and limited coverage (low ceding percentages, especially if insured window is wide).
- Countries prioritizing growth (often more systemically exposed) choose more expensive packages with higher payouts, higher ceding percentages, and generally wide insured windows.
- Simulation weights illustrate outcomes for preferences (example weights z: debt 0.9, growth 0.1; z: debt 0.5, growth 0.5; z: debt 0.1, growth 0.9).

### Result 8 — Price discounts and donor support
- A discount on the insurance premium (e.g., donor support) allows countries to choose more expensive packages with better coverage and growth protection.
- Proxied in analysis by halving the insurance multiple for each package, effects include:
  - (i) smaller increases in debt from insurance;
  - (ii) improved tradeoffs (same debt increase associated with larger growth gains; smaller “sacrifice ratio”);
  - (iii) shift in optimal choices toward more growth-biased packages that cede higher risk and have higher insured windows.
- Conclusion: lower price enables sovereigns to select packages that provide better coverage and growth protection.

### Result 9 — Borrowing constraints and insurance value
- Countries facing larger borrowing constraints derive greater benefit from disaster insurance.
- Tighter post-disaster borrowing limits (example scenarios: “borrowing” =5% of GDP vs “borrowing” =10% of GDP) improve the tradeoffs between growth and debt that insurance offers (steeper tradeoff slope).
- Interpretation: when assistance or post-disaster borrowing is anticipated (non-binding constraint), countries may opt for lower insurance coverage (charity hazard).

### Policy implications and options highlighted
- Governments should evaluate insurance not only on expected losses but on tail risks and potential catastrophic fiscal impacts.
- Optimal insurance choice depends on country-specific factors: size, exposure, fiscal space, payout limits, borrowing capacity, and preferences over debt versus growth.
- Donor-subsidized premiums or discounts (reducing the insurance multiple) can materially improve the attainable growth-debt tradeoff and make higher-coverage packages feasible.
- Where payout limits bind, consider instruments or arrangements that raise effective payout ceilings for larger exposures or coordinate regional pools with adequate caps.
- For smaller, highly exposed countries, consider balancing higher risk aversion and growth protection needs against prohibitive premium-driven debt costs; targeted subsidy or concessional financing can help.

*Source: Fund staff estimates.*

### 35.      This section discusses options for expanding insurance coverage for sovereigns,

### Expanding insurance coverage for sovereigns (section summary)

### Integrating insurance decisions within Disaster Resilience Strategy (DRS)
- Risk transfer and risk retention decisions should be made in the context of a comprehensive DRS to: 
  - ensure an optimal allocation of resources between risk transfer and risk retention;
  - balance risk mitigation and risk reduction through resilient investment;
  - maintain policy consistency with government’s longer-term macro-fiscal objectives.
- Many countries are developing disaster risk management or disaster risk financing strategies with World Bank assistance, though some strategies may not take a comprehensive and quantified approach to resilience building.

### Scaling and financing regional insurance pools
- Diversification of membership and perils lowers operational costs and reinsurance costs, which can reduce premia for member states.
- Regional pools can share risks among themselves to reduce costs; in extreme cases pools could merge, but risks can also be shared while maintaining portfolio segregation to avoid cross-subsidization (example: CCRIF segregation of Caribbean and Central American cells).
- Increasing capital of regional pools (recapitalization) can:
  - lower reinsurance costs because pools can retain more risk;
  - increase availability of coverage, addressing limits such as the current CCRIF coverage cap of up to US$100 million per hazard per year.
- Sources of capital currently include donor contributions to trust funds and country insurance premia; the Global Risk Financing Facility (GRiF) is cited as an instrument to strengthen financial resilience of vulnerable countries.

### Donor support and conditionality to make insurance affordable
- Direct donor support for insurance premia can increase country coverage and encourage participation in regional pools, further reducing premia.
- To avoid dependency on donor funding, donor-financed premium support could be temporary and conditional on:
  - countries entering or having an IMF program, or already pursuing sound macroeconomic policies (allowing higher access to post-disaster payouts to protect reforms or provide incentive for structural adjustment);
  - countries purchasing their own insurance while donors match the premium for a higher payout (maintains incentive to purchase higher insurance levels and increases payout amounts);
  - countries committing to invest in climate-resilient infrastructure and risk reduction, operationalized through donor-supported programs within regional pools with ex-ante agreements and compliance monitoring (noting challenges: enforcement and potential weak resilience gains if enforcement is weak).
- Conditional donor support could improve growth-debt tradeoffs and be mutually beneficial when tied to demonstrated fiscal efforts or strong macroeconomic policies.

### Indirect subsidization and recapitalization options
- Indirect premium subsidization via recapitalization of regional pools can:
  - lower costs by enabling pools to retain more risk and reduce reinsurance reliance;
  - increase coverage limits available to countries.
- Reinsurance costs are the largest contributor to pool costs and thus to countries’ insurance premia; capital increases would allow pools to hold more risk and lower reinsurance needs.

### Increasing penetration of private property insurance to reduce government contingent liabilities
- Options to incentivize broader private insurance uptake by households include:
  - developing local insurance markets and strengthening regulation and supervision;
  - regional pooling of private insurance risk to overcome local market failures;
  - government-sponsored pools for natural disasters (example: Turkish Catastrophe Insurance Pool, TCIP);
  - incentives to private providers of risk transfer instruments where market failures exist.
- For smaller states, a combined approach could include mandatory property insurance with targeted subsidies for poorest households, pooling these risks nationally (potential government support), reinsuring internationally, and using existing regional pools to pool national pools’ risks to achieve further diversification and cost gains.
- Considerations: appetite and capacity of local insurers to take on higher risk, need for enhanced supervision, and additional financial costs to support capital of national or regional pools.

### Trigger measurement, design, and expanding instrument range
- Better trigger measurement and design can reduce underinsurance by:
  - refining triggers to be less complex/uncertain;
  - ensuring more granular geographic measurement to avoid non-triggering events outside trigger areas.
- Expanding insurance instrument types:
  - Issuing catbonds: increasingly affordable and viable, especially at regional scale, though still costly.
  - Introducing hurricane clauses in sovereign bonds: a debt repayment stand-still following disasters (akin to Grenada’s clause) can provide immediate post-disaster financing; such clauses can be structured to be NPV-neutral for investors, though other costs (e.g., reprofiling risks) may arise.
  - Complementing parametric triggers with indemnity-type or industry-loss triggers: indemnity-type reimbursements benchmarked to easier proxies for actual losses (for example, aggregate losses reported by insurance companies) can mitigate basis risk and increase overall coverage, albeit with potential delays relative to parametric payouts.

### Key conclusions and policy implications
- Pre-arranged financing, including insurance, is compelling because disasters can have large economy-wide implications and immediate payouts can alleviate binding financial constraints post-disaster.
- Insurance typically adds to debt ex-ante because expected payouts are kept below yearly premia due to low-probability catastrophic events; however, ex-post disasters can be far costlier than ex-ante expectations, and risk aversion to catastrophic macroeconomic effects can make insurance desirable despite its costs.
- Optimal insurance choices require a benchmark that internalizes macroeconomic trade-offs between insurance costs (impact on debt) and benefits (impact on growth); the paper uses a government utility function framework to capture these trade-offs and determine insurance choices that maximize utility.
- Countries with more systemic disaster exposure and tighter borrowing constraints (often smaller or low-income states) would benefit more from higher risk transfer but are also more likely to be underinsured due to lack of fiscal space and high public debt.
- International donor support (direct or indirect) can help vulnerable states scale up financial resilience without creating notable debt sustainability risks, improve growth-debt tradeoffs, and reduce ex-post donor support costs if conditioned on fiscal efforts or sound macroeconomic policies.
- Practical decisions about risk transfer should be anchored on three elements:
  - understanding and quantifying exposure to natural disasters;
  - clarity over government preferences regarding growth protection versus fiscal sustainability (finding that keeping insurance multiples low and insuring smaller disasters can improve the debt-growth tradeoff for a given cost range);
  - taking risk transfer decisions within broader disaster resilience strategies that weigh financial resilience versus structural resilience and risk transfer versus risk retention.

### Areas for further analytical work
- Optimal allocation of fiscal space between financial resilience (purchasing insurance) and structural resilience (investing in resilient infrastructure).
- Detailed trade-offs between purchasing insurance and self-insuring against disasters (treated here as market borrowing).
- Expanding analysis across a broader range of countries to identify how optimal risk transfer choices depend on country-specific characteristics such as size and risk exposure.

*Source: wpiea2020003-print-pdf - 35.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020003-print-pdf.pdf_