Critical illness: cancer incidence double the pricing assumption and a closed quota-share treaty
A critical illness product covers 85,000 lives and has been in force six years; benefits are paid on diagnosis. Pricing assumed a cancer incidence of 1.4 per 1,000 and a 5% annual improvement in morbidity, with a 90-day waiting period and standard underwriting up to ₹25 lakh sum insured. Latest experience shows cancer incidence of 2.1 per 1,000, while cardiovascular incidence is in line with pricing; average severity is ₹8.2 lakh against ₹7.0 lakh assumed. Cancer is 60% of all claims, and 30% of cancer claims arise in the first two policy years against 20% expected. The owning channel is agency, with a growing online affiliate contribution. Quota-share reinsurance at 70% applied to the first three years only, and the treaty is now closed to new cessions on this product. A competitor has launched a longer waiting period and higher rates.
1. Assess whether the experience is credible and quantify a repricing indication, allowing for the morbidity assumption error.
Read the worked answer
Credibility. 85,000 lives over six years at around 2 per 1,000 produces roughly 1,000 cancer claims. That is credible for incidence; the cardiovascular result confirms the data are not systematically distorted. Severity is less credible, but the count is adequate for an indication.
Build-up.
- Cancer incidence gap: 2.1 / 1.4 = +50%
- Severity gap: 8.2 / 7.0 = +17%
- Experience-only indication, cancer being 60% of claims: 0.6 x 1.5 x 1.17 + 0.4 x 1.0 = 1.45, i.e. +45%
- Morbidity assumption error: pricing assumed -5% a year, actual is a modest deterioration, say +1%. A 6-point gap compounded over an average duration in force of about three years is roughly 1.06^3 = 1.20. Applied to the cancer block: 0.6 x 1.5 x 1.17 x 1.20 + 0.4 = 1.66, i.e. +66% on an unadjusted basis.
- Early-duration concentration (30% of cancer claims in the first 24 months against 20% expected) suggests anti-selection or weak underwriting. This is not a random fluctuation and will persist unless the underwriting is fixed.
Indication. Present a range of +45% to +65% on the cancer element, not a single figure. The width reflects uncertainty over whether the early-duration claims are a cohort effect or a permanent feature, and over the true trend sign. Do not average away the assumption error: a 6-point error in the trend assumption compounds silently and is the most important single finding.
2. Recommend actions, distinguishing what can be done to new business, to the in-force, and to the portfolio's risk transfer, and explain the constraints.
Read the worked answer
New business. File a replacement product: cancer rates +45-55%, a 180-day cancer waiting period, a two-year pre-existing exclusion, and mandatory medicals plus a family-history questionnaire above ₹25 lakh. A competitor has already moved, so the market will accept it. This is the only action that is fully within the insurer's control.
In-force. Check the policy wording: CI rates are usually reviewable at renewal, but if the contract is rate-guaranteed, a repricing cannot be applied and a premium deficiency test is required instead. If reviewable, apply +15-20% with design tightening, which is a smaller shock than +55% and recognises that the anti-selection finding is partly an underwriting failure the insurer owns. If guaranteed, hold a deficiency reserve for the expected shortfall and let the product run off.
Risk transfer. The 70% quota share is closed, so the whole deterioration is now retained. Options: buy a cancer-specific stop-loss, or place a new quota share on the replacement product. Expect the reinsurer to require the waiting-period change as a condition, and to price the anti-selection explicitly.
Constraints and sequencing. Brief agents before filing: a sudden +50% on a product that is 60% cancer claims will cause a sales collapse if it lands without warning. The online affiliate channel is where the early-duration claims concentrate, so tighten its underwriting rules first. Accept a one to two year profit hit on the in-force block; the alternative is subsidising it from the rest of the book indefinitely.