Parametric Reinsurance for SCS, Wildfire and Wind: How Does It Work?

Watch the full webinare here: 

Why are secondary perils outgrowing traditional reinsurance structures?

• SCS is no longer a secondary consideration on the balance sheet. There were 1,250  SCS events in the US in 2025, with insured losses exceeding $50 billion — a claims cost increase of 156% since 2020.
• The reinsurance gap behind those numbers is structural, not just a function of a hard market:
• SCS losses are not comprehensively ceded. Cover is most commonly bought per severe event rather than against a high frequency of losses accumulating over a full year.
• Retentions have stayed high even as the market has softened. Exposure keeps growing, but cedents' aggregate retentions haven't come down to match.
• Retained SCS volatility eats into surplus. For US insurers, a shrinking surplus base is watched closely by rating agencies, regardless of how well a book is otherwise performing.
• Loss growth itself is driven by exposure as much as by hazard. Housing density in SCS-prone rural areas has grown sharply over the past two decades in many counties, meaning more structures now sit in the path of storms that would once have caused limited damage. Hazard intensity is also shifting geographically, so historical loss experience in a given region isn't always a reliable guide to what comes next. Add an aging housing stock — older roofs are more vulnerable to wind and hail — and sustained inflation in post-event construction and repair costs, and the result is a loss trend that outpaces the frequency of the underlying weather itself.

What is a modeled-loss parametric cover?

A modeled-loss parametric reinsurance cover pays based on an index calibrated to the cedent's own policy-level exposure, rather than on a claims-adjustment process or a generic industry loss index. The design work happens in three stages:

1. Hazard. Each peril is measured against an independent, third-party data source: hail against NOAA NEXRAD radar data, tornado against the National Weather Service's Enhanced Fujita Scale, and straight-line wind against physics-based atmospheric variables.

2. Vulnerability. Separate damage functions are applied per peril, tailored to the class of underlying asset — houses, commercial buildings, and auto are not assumed to respond to hail or wind in the same way.

3. Calibration. The index is tuned to reflect the cedent's actual underwriting terms: wind/hail deductibles, roof age schedules, cosmetic damage clauses, and other asset-class features that already shape how a real claim would pay out.

The output is a modeled gross loss designed to track the cedent's own ultimate net loss (UNL), location by location, rather than a broad regional proxy.

Illustrative example: for a hail structure, insured sites with different hazard intensities, deductibles and limits produce a modeled claim total by running each site's damage ratio against the mapped hazard footprint and applying that site's own deductible and limit — the same underwriting logic a claims adjuster would apply, computed directly from the index instead of after the fact.

How is the index calibrated against vendor catastrophe models?

Where a cedent already runs a vendor catastrophe model, that output becomes the benchmark for the parametric structure rather than a separate, disconnected exercise. The calibration sequence generally runs as follows:

1. Cedent exposure data is extracted from the vendor catastrophe model, or provided directly by the cedent.

2. Historical or modeled losses are extracted from the vendor model's output and matched event by event to the corresponding real historical storms.

3. A candidate parametric index and payout table are drafted — for example, the percentage loss assigned to each Saffir-Simpson category at different wind speed bands and distances.

4. The parametric proxy loss is calculated for each historical event using that candidate index.

5. Back-testing: the parametric proxy loss is compared against the vendor's modeled loss for the same set of events.

6. The index and payout table are refined, and steps 3–5 repeated, until the two loss series track closely across a run of historical storms.

7.  The calibrated structure is proposed for placement.

That back-tested correlation is what a cedent and their broker should expect to see and interrogate before binding: if the two loss series don't track well, the structure isn't ready to sell, regardless of how the trigger reads on paper.

How is a parametric reinsurance layer structured?

Once the index is calibrated, the layer itself is structured much like a traditional excess-of-loss treaty, with a few parameters unique to how modeled-loss covers respond:

The event cap matters more than it might first appear. Setting it at the same point where the cedent's traditional catastrophe tower attaches avoids double-indemnifying the same dollars of loss — the parametric layer is designed to sit alongside the existing program, not duplicate it.

How does Cat-in-a-Circle cover named windstorm exposure?

For tropical cyclone and named windstorm exposure, the equivalent structure is a 

Cat-in-a-Circle portfolio solution. The relationship runs:

Coverage is typically structured with two concentric zones around each location — an inner circle and a wider outer circle — with a higher payout percentage the closer the storm track passes and the higher its category. A Category 5 landfall inside the inner circle might trigger a full-limit payout, while the same category passing through the outer circle triggers a partial one, with the exact percentages set per exposure tier against the treaty's aggregate limit. Because the trigger is distance- and wind-speed-based rather than tied to a specific damage assessment, it responds the same way regardless of how quickly a traditional loss-adjustment process would otherwise conclude.

That speed is the main reason cedents use these structures to replace or supplement an industry-loss-warranty (ILW) position.

Illustrative case study — Florida coastal wind:
 A reinsurer replaced an ILW with a Cat-in-a-Circle-style cover mapped to three exposure tiers (low, medium, high) across its book. A Category 3 landfall with 120 mph one-minute sustained winds making landfall in a low-exposure county triggered a 50% payout factor against the annual aggregate limit — a recovery paid within weeks, without waiting on industry loss development or a third-party methodology to settle.

How does parametric wildfire reinsurance work?

Wildfire parametric structures use the same three-stage design — hazard, vulnerability, calibration — but with a different data source: burnt area rather than wind speed or hail size.

Two independent data sources are commonly used, depending on the solution:

• High-resolution satellite imagery from the European Space Agency's Copernicus Sentinel-2 mission. Descartes processes this imagery to a resolution of approximately 10 meters, with damage assessment refined through partnerships with remote-sensing specialists.
• Ground reports data, published by the National Interagency Fire Center through its Wildland Fire Interagency Geospatial Services (WFIGS), based on actual reports from fire management personnel on the ground — a source some local insurers already use internally, which can make the trigger more intuitive to explain to a board or a regulator.
• Each insured location gets a defined buffer — a circle, a building footprint, or in some cases county assessor parcel data for larger properties — and triggers independently when a fire scar breaches that boundary. The buffer size is a tunable parameter: a fire that reaches the edge of a property boundary frequently causes a full loss even where the roof itself looks intact, so buffers are calibrated with that in mind rather than requiring the burn scar to reach the structure itself.
• Three placement patterns illustrate how this scales — none is universally better; the right fit depends on what exposure data the cedent or ILS fund actually holds:

Structure	Data source	Granularity	Trigger	Buffer	Best suited for
Individual-location cover	Satellite (Sentinel-2) or ground reports (WFIGS)	Per insured location	Burn scar detected within buffer of the location	Simple circular buffer (e.g. 100m)	Portfolios with full location-level exposure data
Individual-location cover, tighter buffer	Ground reports (WFIGS)	Per insured location, building footprint	Burn scar breaches a footprint-based buffer	Tighter buffer (e.g. 50m), building-footprint-based	Carving wildfire out of specific high-risk regions to manage cost
Aggregated zone cover	Satellite (Sentinel-2)	Census block / zip code	% area burned per zone, applied to zone limit	Zone-level, not location-level	Cedents or ILS funds that only hold aggregated exposure data

 

What should reinsurance brokers and cedents ask before structuring a cover?

Does this replace the traditional program, or sit alongside it?
In most placements, it sits alongside. The structure is built around what the cedent already has — including setting the event cap at the attachment point of the existing cat tower — so the same dollars of loss aren't covered twice. Most placements are driven by balance sheet and earnings protection rather than by a need for statutory or regulatory capital relief.

How is exposure updated as the book changes mid-term?
Underwriting can be done on a projected basis rather than purely off the exposure snapshot at inception, so new locations, new underlying classes of business, or portfolio-level hardening (like roof replacements) can be reflected in how the cover is structured. Where those changes are material, the more common approach is to flex the structure itself — typically the attachment point — rather than simply adjust price mid-term.

If a cedent improves its primary underwriting, does that change the price?
It changes the structure more than it changes the price directly. Higher wind/hail deductibles, a move to actual cash value (ACV) claims settlement, or cosmetic damage provisions all change how a portfolio responds to an event. Rather than a straight discount on an unchanged structure, the more common adjustment is to the attachment point and the damage function, so the cover stays calibrated to the portfolio's improved resilience.

How is basis risk handled when satellite and ground evidence disagree?
Basis risk is the potential mismatch between the payout generated by a parametric trigger and the actual financial loss experienced by the cedent. The treatment is agreed upfront in the structure, not adjudicated case by case after an event. Wildfire structures are often designed with a slightly positive basis — paying somewhat more than the client may actually face — specifically so that any divergence between data sources doesn't leave the cedent short, and so a location-by-location dispute process (which would undercut the core speed advantage of a parametric trigger) isn't needed.

Is this only useful for large, sophisticated cedents?
No. National carriers use SCS-specific modeled-loss covers, but a substantial share of current placements are with regional and mutual carriers — often books with meaningful geographic diversification that previously relied on an aggregate reinsurance structure they've been unable to place in recent years, and are now looking for retention relief directly.

How do rating agencies and regulators treat these covers?
AM Best publishes a basis-risk scoring methodology for non-indemnity catastrophe bonds under which the maximum reinsurance credit is capped at 90%, with the actual credit depending on how the structure scores against factors such as the objectivity of the underlying index and how closely the trigger is expected to track the sponsor's actual losses. Descartes' underwriting team generally sees parametric reinsurance placements assessed on a comparable basis, though cedents should confirm the treatment of a specific structure directly with AM Best. NAIC risk-based capital rules give credit for ceded reinsurance more generally, but do not publish an equivalent peril-specific framework for non-indemnity structures — cedents should confirm applicable treatment with their own regulatory contacts.

Do you work with all vendor catastrophe models, or only certain ones?
Any vendor model. Descartes takes the cedent's exposure data and the model's output, reviews the events driving the cedent's modeled historical losses and PMLs, and calibrates the parametric structure against that view. Where a question arises about a particular model's methodology or assumptions, that gets worked through directly with the cedent.

Why is SCS covered through a modeled-loss structure rather than a simple index trigger like Cat-in-a-Circle?
Basis risk. SCS is a high-frequency peril with a wide geographic spread across the US, a different proposition from a single-track peril like a hurricane, where a Cat-in-a-Circle-style trigger works well. For SCS, the index instead measures the intensity of each underlying peril — hail, tornado, straight-line wind — at every individual insured location, applies a damage function to produce a ground-up loss, then applies the cedent's own deductibles and limits before aggregating, which is what lets the modeled gross loss track the cedent's actual UNL closely.

Can wildfire mitigation measures, such as defensible space or home hardening, be reflected in the structure?
Yes, provided the information is objective and verifiable. Ground-based fire reports can overestimate the loss for a well-mitigated property that sits inside a fire scar, so one adaptation is to adjust the buffer around a location, or to use high-resolution satellite imagery, which is more precise at determining whether defensible space was actually breached. Where a fire reaches the property boundary itself, a full loss is still assumed.

For more common questions about parametric reinsurance structuring, see Descartes' FAQ hub.

A practical starting point

The common thread across SCS, named windstorm and wildfire structures is the same: the trigger is only as good as how closely it's been calibrated to the cedent's own portfolio, and that calibration work happens before the treaty is bound, not after a loss. For a broker bringing a placement to market, the most useful thing to arrive with isn't a target price — it's exposure data, a view of historical losses (from a vendor model or otherwise), and a clear sense of where the existing program already attaches, so the parametric layer can be built to sit precisely alongside it rather than in competition with it.

Sources

• Descartes Underwriting. Severe Convective Storm (SCS) Reinsurance: Protecting Cedants From Uncovered Losses.
• National Oceanic and Atmospheric Administration, National Centers for Environmental Information. "Next Generation Weather Radar (NEXRAD)." https://www.ncei.noaa.gov/products/radar/next-generation-weather-radar
• NOAA Storm Prediction Center. "Enhanced Fujita Scale (EF Scale)." https://www.spc.noaa.gov/efscale/
•  European Space Agency. "Copernicus Sentinel-2." https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-2
•  National Interagency Fire Center. "WFIGS Interagency Fire Perimeters." https://data-nifc.opendata.arcgis.com/datasets/nifc::wfigs-interagency-fire-perimeters/about
• AM Best. "Gauging the Basis Risk of Catastrophe Bonds." Best's Methodology and Criteria, 19 December 2017. https://www3.ambest.com/ambv/ratingmethodology/openpdf.aspx?ubcr=1&ri=1501
•  National Association of Insurance Commissioners. "Risk-Based Capital." https://content.naic.org/insurance-topics/risk-based-capital