Incurred But Not Reported (IBNR) Estimation Methods
A technical overview of the primary methods actuaries use to estimate IBNR reserves.
What Is IBNR?
IBNR (Incurred But Not Reported) represents claims that have occurred but have not yet been reported to the insurer. This includes both pure IBNR (truly unreported claims) and development on known claims (case reserve inadequacy). IBNR is often the largest and most uncertain component of an insurer's total reserves, particularly for long-tail lines where reporting delays can span years.
To understand why IBNR matters, consider a general liability insurer at the end of the calendar year. Some incidents that occurred during the year have been reported and have case reserves assigned by claims adjusters. But other incidents have occurred that the insurer does not yet know about. A slip-and-fall at a commercial property in November may not result in a claim until the injured person consults a lawyer months later. Even among reported claims, the case reserves set by adjusters may prove inadequate as more information becomes available. IBNR captures both of these sources of additional cost.
Pure IBNR vs. Case Reserve Development
It is useful to decompose total IBNR into two components. Pure IBNR is the expected cost of claims that have been incurred but not yet reported to the insurer. This component is driven by reporting delays, which vary significantly by line of business. In personal auto, most claims are reported within days. In medical malpractice, claims may not be reported for several years after the incident. The second component is case reserve development, which captures the expected increase (or decrease) in the value of claims that have already been reported. This arises because initial case reserves are often based on incomplete information and tend to be revised upward as claims mature.
For short-tail lines like property insurance, pure IBNR is typically the larger component because most reported claims are settled quickly and close to their initial case reserve. For long-tail lines like general liability, case reserve development often dominates because claims take years to settle and their values evolve substantially over time.
Common Estimation Methods
The chain ladder (or link ratio) method develops paid or incurred losses to ultimate using historical development patterns. The Bornhuetter-Ferguson method blends actual experience with an a priori expected loss ratio, providing stability when early development data is sparse. The Cape Cod method is a variant that uses earned premium to weight the expected loss ratio. Frequency-severity approaches estimate IBNR by separately projecting unreported claim counts and their expected average cost. Each method has strengths and weaknesses, and actuaries often use multiple methods to triangulate a best estimate.
Chain Ladder Method in Detail
The chain ladder method is the most widely used IBNR estimation technique. The actuary constructs a loss development triangle with accident years as rows and development periods as columns. Link ratios are calculated as the ratio of losses at one development period to losses at the prior period. For example, if accident year 2022 shows $10 million in incurred losses at 24 months and $12 million at 36 months, the 24-to-36-month link ratio is 1.20. Selected link ratios (often weighted averages of several years) are multiplied together to produce cumulative development factors from each maturity to ultimate.
To estimate ultimate losses, the actuary multiplies the current reported losses by the appropriate cumulative development factor. IBNR is then the difference between estimated ultimate losses and current reported losses. For example, if an accident year has $8 million in reported losses and a cumulative development factor of 1.50, the estimated ultimate is $12 million and the IBNR is $4 million.
Bornhuetter-Ferguson Method
The Bornhuetter-Ferguson (BF) method addresses a key limitation of the chain ladder: sensitivity to current reported losses. In the chain ladder, if current losses are unusually low (or high), the ultimate estimate will also be low (or high), even if the deviation is due to random fluctuation. The BF method stabilizes the estimate by combining the chain ladder with an a priori expected loss ratio.
The BF formula for IBNR is: IBNR equals (expected loss ratio times earned premium) times (1 minus 1 divided by the cumulative development factor). This formula assigns the unreported portion of losses based on the a priori expectation rather than the actual reported losses. The method is especially valuable for immature accident years where reported losses are sparse and volatile. Its weakness is the reliance on the a priori expected loss ratio, which introduces subjectivity.
Cape Cod Method
The Cape Cod method (also called the Stanard-Buhlmann method) is a refinement of the BF approach. Instead of using an external expected loss ratio, it derives the expected loss ratio from the data by dividing the sum of reported losses (weighted by the proportion reported) by the corresponding earned premium. This produces a data-driven expected loss ratio that is used in the same BF formula. The Cape Cod method balances the responsiveness of the chain ladder with the stability of the BF method.
Frequency-Severity Approaches
Frequency-severity methods estimate IBNR by separately projecting the number of unreported claims and their expected average cost. This approach requires a claim count triangle in addition to a dollar loss triangle. The actuary develops claim counts to ultimate to estimate the number of unreported claims, then multiplies by an expected average severity. This method provides additional insight into whether IBNR is being driven by claim frequency or severity, which can inform management decisions about claims handling and pricing.
Common Pitfalls
One common mistake is applying chain ladder development factors mechanically without examining whether the historical development patterns are still appropriate. Changes in claims handling practices, legal environment, or reinsurance structure can all alter development patterns. Another pitfall is using a single method and treating its output as the definitive answer. Best practice is to use multiple methods, understand why they produce different results, and select a best estimate based on professional judgment.
Actuaries should also be careful about the tail factor, which represents development beyond the oldest maturity in the triangle. For long-tail lines, losses may continue to develop for 20 years or more, and the tail factor can be a significant component of the total IBNR estimate. Understating the tail factor is one of the most common causes of reserve deficiency.
Study Tips
For exam preparation, practice constructing loss development triangles and calculating link ratios. Be able to apply the chain ladder, BF, and Cape Cod methods from start to finish. Understand the assumptions underlying each method and when each is most appropriate. Know how to decompose IBNR into its pure IBNR and case development components. Practice comparing results across methods and articulating the reasons for differences. Finally, review how tail factors are selected and the impact of tail factor selection on ultimate loss estimates.