Economics & ClaimsAdvanced

Experience vs. Exposure Rating

Two ways to price a layer

Two competing techniques for pricing excess-of-loss reinsurance: one trends the account’s own loss history, the other builds a price from first principles using a severity curve.

Open the interactive version → definition, quiz, structure diagram and progress tracking

Definition

Experience rating prices a layer from the cedent’s own historical losses — trended to current cost levels, developed to ultimate, and run back through the specific retention and limit being priced — to produce a burning cost. Exposure rating instead largely sets the account’s own history aside and builds a price from the size of the exposure and a market severity curve, allocating a share of the expected loss to the layer based on where it sits in the tower. Actuaries typically blend the two, weighting each by credibility: experience rating dominates where loss history is large and stable, exposure rating dominates for high or volatile layers where an individual account simply has not produced enough losses to be statistically meaningful.

Experience rating: trend historical losses to today’s cost level, develop them to ultimate to capture IBNR, restate each year through the layer’s retention and limit, then average into a burning cost.
Exposure rating: apply a severity curve (often built from increased-limit factors) to the subject premium to estimate what share of expected loss falls inside the layer, independent of the account’s own claims record.
Credibility weighting blends the two: Z × experience rate + (1 − Z) × exposure rate, with Z rising toward 1 as the account’s own history grows larger and more stable.
High layers lean heavily on exposure rating — a handful of loss-free years in a $20M xs $10M layer does not mean the true expected loss is zero, only that the account has not produced enough data to say otherwise.

Blended Rate = Z × Experience Rate + (1 − Z) × Exposure Rate

Z is the credibility assigned to the account’s own experience; it rises as that experience becomes larger and more stable.

Worked example

A $5M xs $5M layer on a mid-sized account with ten clean years of data might be priced 70% from trended experience and 30% from an exposure curve. A $40M xs $10M catastrophe layer on the same account, where the retained history shows zero losses, is priced almost entirely from exposure.

Scenario · figures in USD

Blending two views of the same layer

A $10M xs $5M layer is being priced on an account with ten years of trended, developed loss history and one loss ever recorded in the layer.

Experience-rated burning cost	$420,000
Exposure-rated loss cost	$650,000
Credibility assigned to experience (Z)	35%
Blended technical loss cost	0.35 × $420,000 + 0.65 × $650,000 = $569,500
Expense & profit loading (15%)	$569,500 × 1.15 = $654,925
Resulting rate on line	$654,925 ÷ $10.0M = 6.5%

So whatPure experience rating would have priced the layer at $420,000 — cheap, because the account got lucky and produced only one loss in ten years. Pure exposure rating would have priced it at $650,000. Blending at 35% credibility lands at $569,500, much closer to the exposure view: ten years and a single loss is simply not enough data to trust on its own.

Check your understanding

An account has thirty years of stable, credible loss history on a working layer. Which rating method should dominate the price?

Experience rating, because the account’s own history is credible enough to trust. High credibility means the account’s own claims record is statistically reliable, so experience rating carries most of the weight in the blend.

Why is a high catastrophe layer almost always priced primarily by exposure rating rather than experience?

A handful of loss-free years in that layer does not mean the true expected loss is zero — the account lacks enough history at that height to be credible. High layers are hit rarely by nature. Zero recorded losses reflects thin data, not zero expected loss — exactly the situation exposure rating is built for.

Word problem

A $15M xs $10M layer is being priced for a regional account. Ten years of trended, developed losses produce an experience burning cost of $310,000. An industry exposure curve applied to the same subject premium produces an exposure loss cost of $540,000. The actuary assigns 25% credibility to the experience result, then adds a 20% loading for expenses and profit. Calculate the blended loss cost, the technical premium, and the rate on line.

Show a hint

Blend the two loss costs by credibility first, then apply the loading, then divide by the limit to get the rate on line.

Reveal the worked answer

Blended loss cost = 25% × $310,000 + 75% × $540,000 = $77,500 + $405,000 = $482,500
Technical premium after a 20% loading = $482,500 × 1.20 = $579,000
Rate on line = $579,000 ÷ $15,000,000 = 3.86%

At 25% credibility the exposure view dominates the blend — appropriately, since ten years is thin history for a layer this high. The technical premium of $579,000 works out to a 3.86% rate on line, the number a broker would actually take to market.