California FAIR Plan's 35.8% Wildfire Rate Hike: Cat Models, Reinsurance Costs, and a Territory Dispersion Story

The percentage is the least informative number in the filing. It is a weighted average of territory changes running from a 78% cut in the most favored Central Valley groupings to over 300% in the most exposed wildland-urban interface ZIP codes, produced by replacing historical-loss ratemaking with catastrophe model output.

Key Takeaways

• The 35.8% statewide figure spans a 78% territory decrease and increases above 300%, and the FAIR Plan book is concentrated in the high-hazard territories, so the policy-weighted change runs well above the average.
• Cat model expected annual loss replacing the historical-loss burden contributes roughly 15 to 20 percentage points of the indication, the largest single component.
• The net cost of reinsurance load, newly permitted under the Sustainable Insurance Strategy, adds roughly 8 to 12 points on a reinsurance tower whose price has roughly doubled since 2019.
• The territory cuts are a defense against adverse selection: depopulation targets pull lower-hazard policyholders out first, and they leave fastest when smoothing overprices them.
• This is the first prior-approval filing to use both cat model output and a reinsurance load, so its CDI review sets the documentation template for every admitted carrier filing in 2026 and 2027.

Three Departures From Every Prior Indication

The FAIR Plan is the insurer of last resort, a statutorily mandated syndicate of every admitted property insurer in California, with premiums pooled and losses shared in proportion to voluntary market share. Its rates go to the Department of Insurance under the Prop 103 prior-approval regime in force since 1988.

The first departure is cat model output in the rate base. A stochastic wildfire model, drawing on vendor calibrations that completed Public Model Review in 2025, develops expected annual losses by territory in place of a historical-loss burden study. Historical ratemaking for California wildfire has never been stable: a single megafire year such as 2017, 2018 or the 2025 Los Angeles fires swamps any reasonable experience period, and smoothing across long horizons either ignores recent frequency or understates the pre-regime baseline.

The second is an explicit net cost of reinsurance load. California's prior-approval formula historically excluded reinsurance cost, treating it as a capital management choice rather than a cost of providing insurance. That was defensible while reinsurance was a stable share of gross premium. With wildfire rate-on-line multiplied several times over five years, the exclusion meant ceded carriers absorbed their catastrophe tower in equity instead of premium.

The third is territory differentiation at a scale California has not used. Cat model output is granular by construction, with hazard scores varying by orders of magnitude within a single county, and the filing translates that directly into relativities instead of smoothing it into one statewide number.

The Statewide Average Is Not the Rate Change Most Insureds Get

Consumer groups have pointed at the tail cases above 300% and trade coverage has led with 35.8%. The arithmetic between them is the story.

A 35.8% average built from a 78% decrease at one end and 300%-plus at the other is a different product from a 35.8% uniform increase. FAIR Plan policyholders are concentrated in high-hazard geographies, which is why they are at the Plan. Most sit in territories where the true change is well above the average, and a minority in lower-hazard transitional zones take the cuts that pull the average down.

Each component has a different trajectory, which matters more than the total. The cat model contribution should be stable across future filings once the methodology is accepted. The reinsurance load will move every year with the reinsurance market, so a soft market lowers next year's indication without any change in modeled hazard.

The territory cuts are the part that reads as a giveaway and is not one. Depopulation targets under the Sustainable Insurance Strategy give admitted carriers an incentive to take out lower-hazard FAIR Plan policyholders. Under smoothed rates those policyholders are the overpriced ones, so they are both the most attractive to take out and the most willing to go. Every one that leaves raises the Plan's average expected loss per remaining policy and forces the next increase onto a book that is even more concentrated in the wildland-urban interface. Charging them their indicated rate removes the cross-subsidy that made them targets.

That is also why the pre-filing rate structure was inadequate rather than merely low. A rate that undercharges high-hazard territories and overcharges low-hazard ones is insolvent in expectation for any book whose low-hazard segment can exit, and this book's low-hazard segment is being actively recruited away.

The Largest Component Rests on a Disclosed Assumption

The 15 to 20 points from cat model expected annual loss is the biggest single piece of the indication, and vendor models differ substantially in how they condition on recent climate trends. Some use a stationary historical calibration; others increase frequency in line with observed warming and vegetation-moisture trends. That conditioning choice moves the indicated rate more than almost any other model parameter.

Public Model Review requires vendors to disclose the choice and support it with validation data, which is a real advance in transparency over the Florida certification model. It also puts the assumption on the record. Prop 103 lets consumer advocacy groups intervene in filings and recover fees where their participation produces substantial modification, and Consumer Watchdog has a three-decade record of doing exactly that. The FAIR Plan filing drew multiple intervenor objections, and the admitted carrier filings that follow will carry higher profile and more contestable cross-subsidy arguments. Discovery on conditioning assumptions, reinsurance pricing and territory relativities is the predictable shape of that.

The reinsurance load faces a parallel test. The regulations permit it only with treaty terms, ceded premium, ceded loss projections and a demonstration that the structure is actuarially sound rather than arranged to inflate ceded cost, which puts captive and affiliated arrangements under specific scrutiny. And the load is not separable from the modeled loss it accompanies: a model-implied expected loss assumes a given retention and layer structure, so changing the treaty changes both the ceded cost and the risk left in the net indication. Filing and reinsurance program have to be built together or the indication is internally inconsistent.

Legislation is the constraint that could remove the mechanism outright. Hearings on FAIR Plan transformation ran through 2025, and further measures may address dispersion directly, either by capping territory-level rate changes, which would partially undo the territorialization the filing just introduced, or by expanding mitigation subsidies, which would soften the sticker shock without touching the method. A cap is the version that matters actuarially, because it would restore the cross-subsidy the 78% cuts were designed to end while leaving the depopulation incentives that punish it in place.

Further Reading

• California Court Upholds FAIR Plan Assessment Recoupment Surcharge – how the June 2026 ruling on the Plan's $1 billion assessment surcharge sits alongside this rate filing.
• NAIC's First National Homeowners Baseline Shows Non-Renewals Up 216% in the West – The Western Zone's countrywide non-renewal and premium figures this FAIR Plan filing's territory dispersion sits inside.
• Verisk Synergy Studio Rewrites the Cat Modeling Playbook – How cloud-native cat platforms are reshaping vendor workflows, Public Model Review submissions, and pricing actuary toolchains.
• Climate Risk and Catastrophe Modeling 2026 – Secondary peril integration, climate-adjusted conditioning, and the cross-peril model landscape relevant to California wildfire work.
• Allstate's $925M March Cat Bill Signals a Severe Convective Q1 – Q1 2026 cat loss context, ISO loss cost lag, and mid-year property-cat treaty dynamics that will spill into California renewals.
• Reinsurance Market 2026 – 1/1 renewal dynamics, rate-on-line trends, and the capacity backdrop that shapes the net cost of reinsurance load in California filings.
• Cat Bond Market Hits $63.9B as Pension Funds Scale Up – The ILS capital layer that absorbs a growing share of California wildfire risk above the traditional reinsurance tower.
• The P&C Market Cycle in 2026 – Broader market context on hard versus soft dynamics, rate adequacy, and combined ratio pressure.
• ZestyAI's AI Property Risk Models Expand Coverage in Wildfire Zones – How AI-native property-level risk scoring enables coverage for nearly one million previously uninsurable properties in catastrophe-exposed regions.
• NAIC Strengthen Homes Act: Standardizing Mitigation Premium Credits Across States – The NAIC model law framework for DOI-run mitigation grant programs and the actuarial evidence behind premium discounts for FORTIFIED and wildfire-resilient construction.
• How California's First Approved Cat Model Changes the Property Rate Indication Formula – Pricing mechanics of replacing Prop 103 backward-looking data with AAL, the reinsurance cost pass-through allocation, and the 85% writing mandate's territorial cross-subsidy.
• NAIC Homeowners Data Call Sets Nationwide Peril Baseline – The NAIC's first standardized homeowners data collection will produce zip-code-level wildfire loss, FAIR Plan enrollment, and non-renewal data across all participating states.
• Three Models, One Green Light: CDI Wildfire Cat Certifications and P&C Rate Filings – The multi-vendor landscape now that three PRID approvals are complete, what the conditions of use require beyond EAL substitution, and the adverse selection dynamics between SIS adopters and holdout carriers over the next development cycle.
• California's Public Wildfire Model Enters Its RFP Stage – SB 429's state-owned wildfire model moves into its RFP stage, and why the 29% approved increase against the 35.8% request previews the scrutiny a fully transparent model will bring to future FAIR Plan filings.

Sources

• California FAIR Plan: Key Statistics and Data
• California Department of Insurance: Press Releases and Sustainable Insurance Strategy Updates
• California Department of Insurance: Regulations and Rulemaking (Public Model Review, Net Cost of Reinsurance)
• California Department of Insurance: Studies and Reports on FAIR Plan Transformation
• California FAIR Plan: Rate and Form Filings
• NAIC: Property and Casualty Market Conduct and Rate Filings Guidance
• CAL FIRE: California Wildfire Statistics and Hazard Severity Zones
• Consumer Watchdog: Prop 103 Intervenor Filings and Commentary
• Casualty Actuarial Society: Research on Wildfire Ratemaking and Cat Model Use
• American Academy of Actuaries: ASOP No. 38 on Catastrophe Modeling and ASOP No. 43 on Reserving
• Verisk AIR Worldwide: Wildfire Catastrophe Model
• Moody's RMS: North America Wildfire HD Model
• Cotality (formerly CoreLogic): Wildfire Risk Score and Cat Model
• Insurance Information Institute: California Wildfire and Homeowners Market Data
• Aon Impact Forecasting: Wildfire Catastrophe Insight