Year-end reserving: rapid growth, a new scheme, a TPA backlog and a disputed claim
An Indian health insurer grew 40% in policy year 2024, which ended on 31 December; earned premium was ₹410 crore. At the valuation date, paid claims for 2024 were ₹240 crore and case reserves ₹28 crore, so reported claims were ₹268 crore. Development factors built from the 2021-2023 years are 1.26 from 12 to 24 months, 1.05 from 24 to 36 months and 1.02 from 36 months to ultimate. One newly onboarded group scheme, written from July 2024 at a thin margin, now supplies 30% of exposure and is expected to run at a 95% loss ratio; the rest of the book is expected at 88%. One third-party administrator's average reporting lag worsened by 12 days in November and December. Sixty per cent of 2024 paid claims are from October-December. A ₹6 crore claim has been denied by the administrator but the insurer's medical team believes it is payable. Fifty per cent quota-share reinsurance applies.
1. Determine an appropriate IBNR for policy year 2024, comparing chain-ladder and Bornhuetter-Ferguson and adjusting for the known data problems.
Read the worked answer
Chain-ladder. 12-to-ultimate factor = 1.26 x 1.05 x 1.02 = 1.350. Ultimate = 268 x 1.350 = ₹362 crore; IBNR = ₹94 crore; loss ratio 88%.
Why chain-ladder is weak here. The factors come from a smaller, older, differently mixed book. Thirty per cent of 2024 exposure is a scheme that did not exist in the base years, and 40% growth concentrated in the second half makes the 2024 year more immature than the average historical year at the same age. Chain-ladder also assumes the reporting pattern is stable when the evidence says it is not.
Bornhuetter-Ferguson. Expected loss ratio = 0.7 x 88% + 0.3 x 95% = 90.1%. Reported proportion = 1/1.35 = 74%. Ultimate = 268 + 0.26 x (0.901 x 410 = 369) = 268 + 96 = ₹364 crore; IBNR = ₹96 crore.
Adjustments.
- TPA backlog: 12 extra days on a ₹240 crore annual paid run-rate is about 3.3%, or ₹8 crore unreported, which becomes roughly ₹11 crore after development.
- Seasonality: with 60% of paid claims in Q4, the December cohort is more immature than the average year; add ₹5-8 crore.
- Disputed claim: ₹6 crore, expected value around ₹4 crore.
- Prudence margin for the new scheme: 3-5% of its IBNR.
Selection. Ultimate of ₹385-390 crore, gross IBNR of ₹117-122 crore; select ₹120 crore. Use BF as the base and present chain-ladder as a cross-check, not the other way round.
2. Explain how the result feeds into the financial statements and what further reserves or disclosures you would consider.
Read the worked answer
Net position. Fifty per cent quota share reduces the gross IBNR to ₹60 crore, but check the treaty's experience-refund or loss-ratio cap: a refund mechanism can reduce recoveries precisely when claims are worst, and the reinsurer's credit standing must be assessed before the recoverable is recognised.
Premium deficiency. Run the test product by product, not in aggregate. The new scheme is priced at a thin margin and, if unexpired premium plus expected investment income is insufficient to cover expected claims, expenses and a reasonable margin for the unexpired period, a deficiency reserve is required against the UPR. This is the most likely source of an additional charge and it is invisible in the IBNR.
UPR adequacy. With 40% growth, UPR is large relative to the claim reserves. Test the unexpired risk on the assumption that the 95% loss ratio persists and that medical trend runs at 10-11% into 2025.
Sensitivity and disclosure. A +/-₹15 crore swing in gross IBNR moves net profit by about ₹7.5 crore. Disclose the method, the switch in emphasis from chain-ladder to BF, the TPA data weakness, the disputed claim and the seasonality adjustment in the valuation report. The judgement is the substance here: a mechanically applied chain-ladder would have understated the reserve by roughly ₹26 crore, and the actuarial sign-off has to show why the adjustment was made rather than simply presenting the number.