Q1 classified 316 properties at four fidelity levels — 91% changed tier from L0 to L3. Now we ask: does that reclassification change the portfolio answer? If EAL is the same regardless of fidelity, the reclassification is academic. If it changes materially, fidelity has real dollar consequences.
Expected Annual Loss by Fidelity Level
Each bar represents the portfolio-wide Expected Annual Loss computed at that fidelity level. L0 uses fire hazard zone membership alone. L1 adds slope, aspect, and vegetation proximity. L2 incorporates structure materials and defensible space. L3 runs full Rothermel fire spread simulation with Monte Carlo weather sampling.
Model: 249 WUI properties, California FAIR Plan base rates, fidelity-adjusted loss factors. EAL = probability of ignition × conditional loss × insured value.
Rate-Based vs. Simulation-Based EAL
This study computes EAL two ways, and they disagree by an order of magnitude:
Rate-based EAL ($127M at L3): Applies industry-standard annual loss rates to each risk class (LOW: 0.1%, MEDIUM: 0.5%, HIGH: 2%, EXTREME: 5%). These rates represent the annual probability of fire AND the fraction of value destroyed — a small number because most years, this neighborhood doesn’t burn at all.
Simulation-based EAL ($1,597M at L3): Uses the Monte Carlo burn probability directly as the loss rate. This answers a different question: “If a fire starts nearby, how much does this neighborhood lose?” The burn probability is high (>50% for most EXTREME cells) because the simulation is conditional on fire occurring.
The rate-based number is what an insurer uses for annual pricing. The simulation-based number is what a reinsurer uses for catastrophe pricing. Both are correct — they answer different questions. The 133% divergence between L0 and L3 holds under both methods.
| Method | L0 EAL | L3 EAL | Divergence | Use Case |
|---|---|---|---|---|
| Rate-based | $54M | $127M | 133% | Annual premium pricing |
| Simulation | — | $1,597M | — | Catastrophe reinsurance |
Rate-based uses industry annual loss rates per risk class. Simulation-based uses Monte Carlo burn probability as the conditional loss rate.
The rate-based EAL incorporates the annual probability of fire occurrence implicitly through industry loss rates. The simulation-based EAL is conditional on fire occurring. To convert: simulation EAL × P(fire per year) ≈ rate-based EAL. At $1,597M × ~8% annual fire probability ≈ $128M, consistent with the rate-based $127M.
Loss Ratio by Fidelity Level
The loss ratio — expected loss as a fraction of insured value — tells the same story. Zone-based pricing (L0) produces an artificially low 1.93% loss ratio that looks profitable on paper. Once terrain and fire behavior enter the model, the true loss ratio is more than double. L1 and L3 converge near 4.2–4.5%, suggesting that basic terrain features capture the portfolio-level answer even without full simulation.
Loss ratio = EAL / Total Insured Value. Industry combined ratio benchmark: 3–5% for catastrophe-exposed lines.
Who’s Paying for Whom?
Zone-based pricing assigns every property in the same fire zone the same rate. That means a property on a steep, chaparral-covered ridge pays the same premium as a property on flat irrigated lawn — as long as they share a zip code.
Under zone-based pricing, 272 properties underpay their actual risk while 17 overpay — a $2.2M annual cross-subsidy from low-risk to high-risk properties. The 17 overpayers are subsidizing the 272 who carry the real exposure. When those 17 leave the pool, the math collapses.
This is the adverse selection spiral that collapsed the California voluntary market. Low-risk homeowners who know they overpay leave first. The pool shrinks to the highest-risk properties, premiums spike, and the insurer is left holding concentrated exposure it never priced correctly.
Probable Maximum Loss and Capital
PML 99 from L3 Monte Carlo: $2.3B. Capital requirement at 1.5× PML99: $3.4B — exceeding the $2.8B total insured value. This neighborhood-level concentration risk is exactly what reinsurers need to see.
The 1-in-100-year loss for this portfolio exceeds the total insured value when capital loading factors are applied. This means a single catastrophic fire season could exhaust the insurer’s capital reserves allocated to this book of business. The concentration risk is not visible in the EAL — it only appears in the tail.