Policy Provision Changes

Study guide: Exam 5

Policy Provision Changes are changes in the terms of the contract itself — deductibles, limits, sub-limits, exclusions, coverage triggers, benefit levels — that alter how loss is shared between insurer and insured for the same underlying risk.

$\text{Adjusted Loss} = \sum_i \left[\min(X_i, L_{\text{new}}) - \min(X_i, d_{\text{new}})\right]$

Unlike underwriting changes, which change which risks are written, provision changes change what is covered on the risks already there. That makes them far easier to quantify: the effective date is known, the terms are documented, and the historical claims can be re-valued under the new provisions.
The adjustment is a pro-forma restatement: take each historical claim and recompute what the insurer would have paid under the new terms, then develop the restated history. This is exactly the loss elimination ratio calculation applied retrospectively.
Provision changes affect the development pattern, not only the level. A higher deductible eliminates small, fast-closing claims, so the surviving claims are larger and slower — the restated triangle has a longer tail and a different shape, and simply scaling the old factors is wrong.
Changes attach to the policy, so they phase in over the policy term. On a Policy Year basis the change is clean; on an Accident Year basis it phases in over roughly two years, which is one of the few situations where policy year is the better reserving basis.
Benefit-level changes in statutory lines (workers compensation reform, auto no-fault changes) are the same problem at industry scale, and are handled by a law-level adjustment factor applied to historical losses before trend — deliberately kept separate from trend so the two are not confused.
On the pricing side the same restatement is required, or the indication prices coverage the insurer no longer sells.

Restating for a Deductible Increase A commercial auto programme raises its per-occurrence deductible from $\$1{,}000$ to $\$5{,}000$ effective $1/1/2024$ . A sample of $1{,}000$ historical claims:

Claim size Count Total losses

$\$1{,}000$ – $\$5{,}000$ $600$ $\$1{,}800{,}000$

over $\$5{,}000$ $400$ $\$6{,}200{,}000$

(All figures are already net of the old $\$1{,}000$ deductible.) Restate the historical losses to the new deductible.
Answer Under the new $\$5{,}000$ deductible, the insurer pays nothing on the first group and $\$4{,}000$ less on each claim in the second:
 $\begin{align*} \text{Eliminated (small claims)} &= \$1{,}800{,}000 \\ \text{Eliminated (layer on large)} &= 400 \times \$4{,}000 = \$1{,}600{,}000 \\[4pt] \text{Total eliminated} &= \$3{,}400{,}000 \\[4pt] \text{Restated losses} &= \$8{,}000{,}000 - \$3{,}400{,}000 \\ &= \$4{,}600{,}000 \end{align*}$ 
The loss elimination ratio is $\$3{,}400{,}000/\$8{,}000{,}000 = 42.5\%$ .

But note what else changed: claim counts fell from $1{,}000$ to $400$ — $60\%$ of claims disappear entirely. Average severity on the restated basis is $\$4{,}600{,}000/400 = \$11{,}500$ against $\$8{,}000$ before. The restated book has fewer, larger, slower claims, so its development pattern is longer-tailed. Applying the old factors to the restated triangle would understate ultimate.

Claim size	Count	Total losses
$\$1{,}000$ – $\$5{,}000$	$600$	$\$1{,}800{,}000$
over $\$5{,}000$	$400$	$\$6{,}200{,}000$

A Provision Change Across Accident and Policy Years The same deductible change takes effect for policies written on or after $1/1/2024$ . All policies are annual and written uniformly.

How does the change appear on each aggregation basis?

Answer Policy year — PY 2023 is entirely old-deductible business and PY 2024 entirely new. Clean, immediate, and fully comparable after restatement.

Accident year — AY 2024 losses arise from both $2023$ -written policies (old deductible, running off) and $2024$ -written policies (new). With uniform writing, roughly half of AY 2024's exposure is on each basis, and AY 2025 is the first fully-new year.

So on an accident year basis the change phases in over two years, at approximately $50\%$ and $100\%$ . Restating AY 2024 requires applying the loss elimination ratio to only the affected half of the exposure — which requires knowing, claim by claim, which policy it arose under.

This is one of the specific situations where Policy Year earns its longer development period: when the thing that changed attaches to the policy, only the policy-year cohort isolates it. The general rule from Ratemaking Data Organization applies — use the most responsive basis that can be corrected to current conditions, and here the correction on an accident year basis is materially harder.