Deductible Change and a One-Event Catastrophe Load in Homeowners
A homeowners writer is preparing its 2026 filing. The experience period is accident years 2022-2024, during which every policy carried a $500 deductible. Effective January 1, 2025, all new and renewing business moved to a $1,000 deductible, and the company has not yet given any premium credit for it. Selected industry loss elimination ratios for this book are 0.04 at $500 and 0.07 at $1,000. Incurred loss and DCCE for 2022-2024 totals $72.0M, of which $18.0M came from a single June 2023 hailstorm; there were no other catastrophe events in the period. On-level earned premium at current rates is $90.0M for the three years, on 45,000 earned policy years. Fixed expenses are $55 per policy; variable expenses are 22% of premium; the profit and contingency provision is 5%. The product team proposes to load for catastrophe using the three-year average cat loss ($6.0M per year) plus a 5% annual climate trend, and to ignore the deductible change because it will wash out in the next filing.
1. Restate the experience losses to the prospective $1,000 deductible level and explain why the adjustment runs in the direction it does.
Read the worked answer
The experience is at a $500 deductible; the prospective period is at $1,000. A deductible eliminates losses below its level, so the loss elimination ratio measures the share removed. The ratio of remaining (retained) losses is:
(1 - LER at $1,000) / (1 - LER at $500) = (1 - 0.07) / (1 - 0.04) = 0.93 / 0.96 = 0.96875
Applied to the $72.0M total, indicated losses at the new deductible are 72.0 x 0.96875 = $69.75M, a reduction of $2.25M, or about 3.1% of losses.
The direction is intuitive once stated: the company is selling less coverage in 2026 than it sold in 2022-2024, so the same rate level produces fewer dollars of loss. The indication must therefore come down, not up.
Two cautions. First, apply each LER to the losses it was derived from. If these are non-catastrophe LERs, apply 0.96875 to the $54.0M non-cat component only. Second, the product team's argument that the change will wash out is wrong for 2026: the wash-out only happens if there is a subsequent filing in which the deductible is already embedded in the base. For the filing at hand, the 2026 rate must reflect the 2026 coverage or the company writes a full year of $1,000-deductible business on a $500-deductible rate.
2. Critique the proposed catastrophe load and describe a better approach.
Read the worked answer
Four problems:
A three-year average containing exactly one event is not an average annual loss. It embeds the frequency of a three-year sample and the severity of one storm. The sampling error is enormous, and the statistic cannot be distinguished from noise.
Double counting. If the $18.0M storm stays in the $72.0M base and a $6.0M annual load is added on top, the event is counted once in the experience and again in the load. The storm must be removed from the experience base before any cat load is added.
The 5% climate trend is unsupported. It is not a frequency or severity trend estimated from the company's data; it is an assumption applied to an already unstable base. If the source used to derive the load already reflects current climate conditions, the trend double counts.
It mixes a loss-based and a premium-based view. Adding a dollar load to a dollar loss base and then dividing by premium requires the load to be on the same coverage basis, including the new deductible.
Better approach. Derive the load from a catastrophe model's average annual loss for the company's actual territory, construction and limit mix, or from a much longer historical record using a frequency-severity cat method. Hold it as a separately supported provision with a stated return period and a range. Set the non-catastrophe experience on non-cat years only, and state explicitly whether the AAL is quoted at a $500 or $1,000 deductible.
3. Compute the indicated change using the corrected deductible treatment and a supported $12.0M average annual cat load, and compare it with the product team's proposal.
Read the worked answer
Corrected basis.
- Non-cat losses: 72.0 - 18.0 = $54.0M; x 0.96875 = $52.31M
- Cat load: $12.0M (if quoted at a $500 deductible, also x 0.96875 = $11.63M)
- Total loss and DCCE = 52.31 + 12.00 = $64.31M
- Fixed expense = 45,000 x $55 = $2.475M
- Indicated premium = (64.31 + 2.475) / (1 - 0.22 - 0.05) = 66.785 / 0.73 = $91.49M
- Indicated change = 91.49 / 90.0 - 1 = +1.7%
Product team basis. 52.31 + (6.0 x 1.05) = $58.61M; (58.61 + 2.475) / 0.73 = $83.68M, or -7.0%.
The cat load choice alone swings the filing by nearly nine points, which is the real issue in this filing; the deductible adjustment is worth about three points of loss and is straightforward. The $12.0M load should be documented with its source, return period and model vintage, and a range presented.
One further point for the memorandum: a 5% profit and contingency provision applied to a cat-exposed homeowners book is thin. If the AAL is a mean, roughly half of all years will exceed it, and the surplus strain of a bad year is not reflected in a 5% margin. A risk load or a separate cat margin should be considered and disclosed.