status: draft

Climate risk and emerging perils

Climate change is reshaping ILS risk assessment. Historical loss data may not predict future losses, and modeling firms are scrambling to incorporate climate trends into their hazard simulations.

The core problem

Cat bond pricing relies on models calibrated to historical event patterns. If climate change increases hurricane intensity, shifts storm tracks, or makes wildfire seasons longer, models trained on 1970-2020 data will underestimate future losses.

You’re betting on stationarity in a non-stationary world.

What climate science tells us

Trend	Confidence	ILS Impact
Sea surface temperatures rising	High	Increased hurricane intensification potential
Sea level rise	High	Higher storm surge damage
Precipitation intensity increasing	Moderate-High	Flood losses rising
Wildfire seasons lengthening	High	Extended exposure period
Tornado frequency shifting	Moderate	Geographic pattern changes
Hurricane frequency	Low-Moderate	Unclear direction

How models are adapting

Long-term average views

Traditional approach: use historical event catalog weighted by long-term climate averages. This assumes the next 3-5 years will resemble the last 50-100 years on average.

Limitations:

• Ignores recent warming trend
• May systematically underestimate future risk
• Appropriate if you believe mean reversion

Near-term climate conditioning

Adjusts historical catalog for recent climate state:

• Current sea surface temperatures (SST)
• El Nino/La Nina (ENSO) phase
• Atlantic Multidecadal Oscillation (AMO)
• Recent precipitation patterns

Typical impact on US hurricane EL: +10-25% vs. long-term average

Rationale: The climate system today is warmer than the historical average. Conditions that produced the historical catalog are not the conditions we have now.

Long-term climate scenarios

Projects event distributions under future warming scenarios:

• RCP 4.5 (moderate emissions)
• RCP 8.5 (high emissions)
• 2030, 2050, 2100 time horizons

Limitations:

• High uncertainty in projections
• Less relevant for 3-year bonds
• Scenario selection is subjective

Hybrid blending

Combines historical with climate-adjusted projections. Introduces subjective weighting between views.

Approach	Weighting
Equal blend	50% long-term, 50% near-term
Conservative	100% near-term conditioned
Weighted by confidence	Varies by peril

Ask which approach was used in any modeling presentation.

Emerging perils

Some perils were historically unmodeled or under-modeled and are now getting attention.

Wildfire

California wildfire was a $10B+ insured loss year in 2017, 2018, and 2020. Traditional cat bonds rarely covered wildfire explicitly. Now it’s mainstream.

Key challenges:

Challenge	Implication
Interface zones (WUI)	Losses concentrate where wildland meets urban
Fire behavior modeling	More complex than wind or shake
Defensible space	Property-level mitigation varies wildly
Utility ignition	Human-caused fires add uncertainty
Model maturity	Less validated than hurricane models

Investor considerations:

• Some investors remain skeptical of wildfire models
• Demand wider spreads for wildfire exposure
• Concentrate on structure/trigger rather than just modeled EL
• Consider parametric triggers (burn perimeter) for faster settlement

Severe convective storm (SCS)

Tornadoes, hail, and straight-line winds. Historically considered attritional (frequent, small losses), SCS is now producing $50B+ annual insured losses.

Key issues:

Issue	Description
Highly localized	Events are small-scale, hard to model at property level
Climate trends	Increasing frequency and severity signals
Aggregate accumulation	Most SCS losses aggregate rather than single mega-events
Trigger challenge	Per-occurrence triggers may miss SCS; aggregate triggers better

Cat bond implications:

• Few pure SCS cat bonds exist
• Often included as secondary peril in multi-peril deals
• Aggregate triggers capture SCS better than per-occurrence
• Attachment points need to account for frequency

Flood

NFIP (National Flood Insurance Program) has issued cat bonds. Private flood insurance is growing.

Challenges:

Challenge	Description
Model maturity	Flood models less mature than hurricane or earthquake
Exposure data quality	Basement coverage, first floor elevation often unknown
Climate intensification	Increasing precipitation intensity
Correlation with hurricane	Storm surge often accompanies wind

Flood cat bond considerations:

• Parametric triggers (rainfall depth, river gauge) common
• Basis risk higher than for wind
• Secondary market even less liquid
• Growing interest from World Bank and sovereigns

Cyber

A new frontier for ILS. Cyber cat bonds have been issued but remain nascent.

Unique challenges:

• No physical footprint to model
• Correlation risk (single vulnerability hits many insureds)
• Accumulation scenarios unclear
• Silent cyber exposure in traditional policies

Most ILS investors treat cyber cautiously, demanding very wide spreads for the model uncertainty.

Investor responses to climate risk

Climate-skeptic investor approach

Decision	Approach
Model view	Long-term average
Acceptable multiple	2.0x spread/EL acceptable
Preferred perils	US earthquake, European wind (less climate-affected)
Wildfire	Avoid or require very wide spreads
Duration	Comfortable with 5-year
Rationale	Climate impacts uncertain, models adequate

Climate-adjusted investor approach

Decision	Approach
Model view	Near-term conditioned (+15-25% EL)
Acceptable multiple	Requires 2.5x+ on climate perils
Preferred perils	Diversified, shorter tenors
Wildfire	Selective exposure with strong modeling
Duration	Prefers 3-year for reassessment
Rationale	Models underestimate, trends accelerating

Pricing dislocations

Divergence between climate views creates pricing dislocations:

• Climate-adjusted investors demand wider spreads
• Climate-skeptic investors accept tighter spreads
• Same bond, different required return

These dislocations may resolve as loss experience accumulates. Or they may persist if climate trends accelerate.

Practical due diligence questions

When reviewing a cat bond with climate-exposed perils:

Model view questions

1. What climate view does this model use?
2. What’s the expected loss under near-term vs. long-term view?
3. How has the model vendor’s view evolved over recent versions?
4. Are sensitivity runs available for different climate assumptions?

Trend questions

1. How would expected loss change under +1.5C vs. +2.0C scenarios?
2. Is the attachment point sufficient if loss trends continue upward?
3. What’s the historical trend in losses for this peril/region?
4. How has the sponsor’s loss experience compared to modeled?

Portfolio questions

1. What percentage of portfolio is climate-sensitive perils?
2. How correlated are climate-exposed positions?
3. Is there geographic diversification within climate perils?
4. What’s the duration profile of climate-exposed holdings?

Loading for climate uncertainty

Some investors apply explicit climate loads to modeled expected loss:

Peril	Climate Load
US hurricane (Florida)	+15-25%
US hurricane (Gulf)	+10-20%
US wildfire	+20-30%
European windstorm	+5-15%
US earthquake	None (not climate-affected)
Japan typhoon	+10-15%

Illustrative loads. Investor views vary significantly.

After applying loads, recalculate the spread multiple. If the loaded multiple is below your threshold, pass or demand wider spread.

Duration strategy

Climate uncertainty favors shorter durations:

Duration	Climate Strategy
1 year	Annual reassessment of view
3 year	Standard, manageable climate drift
5 year	Significant climate risk over period
7+ year	Climate projections increasingly relevant

Some investors cap duration at 3 years for climate-sensitive perils, accepting higher execution costs for the option to reprice.

Secondary peril accumulation

Climate change is making secondary perils (historically attritional) more relevant:

Primary Peril	Secondary Peril	Climate Trend
Hurricane	Storm surge	Increasing (sea level + intensity)
Hurricane	Inland flood	Increasing (precipitation intensity)
Earthquake	Fire following	Stable
Wildfire	Smoke damage	Increasing (fire intensity)
Winter storm	Freeze damage	Variable

Cat bonds may not explicitly cover secondary perils or may sub-limit them. Read trigger definitions carefully.

What sponsors should consider

If issuing in climate-exposed perils:

1. Offer both views: Present long-term and near-term model output. Transparency builds investor confidence.

2. Acknowledge uncertainty: Sophisticated investors know models are imperfect. Acknowledge this rather than overselling precision.

3. Consider trigger choice: Parametric and index triggers avoid arguments about sponsor exposure data accuracy as climate shifts.

4. Duration and reset: Consider shorter tenors or annual resets to give investors repricing options.

5. Premium for climate: Expect to pay wider spreads for climate-sensitive perils, especially wildfire and Florida wind.

---

status: draft

For catastrophe modeling fundamentals, see Catastrophe modeling. For ESG considerations including climate disclosure, see ESG and green ILS.