What it means
Insurance claims often arrive and settle long after the policies' coverage period, so an insurer estimates ultimate losses before all claims are known. The gap from recorded losses informs reserves, and reported and paid data differ.
Start with expected ultimate losses for a group of policies, often based on earned premium times an expected loss ratio. Then estimate what fraction of ultimate claims is already reported at the valuation date; the remainder is the unreported share for which the method retains the prior expected loss.
In a simple reported-loss form, estimated ultimate losses equal losses reported to date plus expected ultimate losses multiplied by the unreported percentage. If $4 million has been reported, expected ultimate losses were $10 million, and 40% is expected to remain unreported, the estimate is $4 million + $4 million = $8 million.
A pure development approach gives $4 million / 60% = about $6.67 million, whereas Bornhuetter-Ferguson uses the $10 million prior for the unreported share, reaching $8 million. The assumptions differ; neither answer is automatically correct.
Development patterns estimate how claims emerge over time, often using historical accident-year data. Claims may be measured on an incurred or paid basis, but mixing those measures in one calculation creates confusion, and a paid-loss approach needs its own pattern and a clear definition of what remains unpaid or unreported.
The Casualty Actuarial Society's discussion of parameter estimation describes selecting both a development pattern and an initial ultimate loss ratio for each accident year. It also notes that a conventional chain-ladder-derived pattern can conflict with assumptions of the Bornhuetter-Ferguson method, so actuaries should evaluate rather than mechanically import a pattern.
At early development ages, a small number of claims may not reflect the eventual total, and the prior expected loss provides stability for the portion that has not yet emerged. As a year matures and a higher share is reported, the method places more weight on observed claims in the ultimate estimate.
For a non-actuary reviewing results, ask for expected loss ratio, earned premium, reported-to-date amount, selected emergence percentage, and the basis of each. Compare the result with prior valuations and explain the effect of changed assumptions separately from newly reported losses.
The method does not replace claim-level case reserves or professional actuarial judgment; it is one way to estimate aggregate ultimate losses and related reserve needs. Management should document uncertainty and test sensitivity, especially for a new line with limited credible experience.
In practice
Real-world examples.
Example
An insurer expects $10 million ultimate losses and estimates that 60% should already be reported. Claims reported to date total $4 million. The unreported percentage is 40%, so Bornhuetter-Ferguson estimates $4 million + 40% x $10 million = $8 million ultimate losses.
Example
A new liability policy has few claims in its first year, though claims often arrive late. Dividing the handful reported by a small emergence percentage could swing sharply after one large claim. The actuary uses a documented prior expectation for the unreported share and revisits it as evidence arrives.
Example
A claims-processing change makes reporting faster. Last year's development pattern may overstate the unreported share, so the team reviews emerging data and tests sensitivity before presenting its estimate.
Formula
Calculation
Bornhuetter-Ferguson ultimate losses = reported losses to date + expected ultimate losses x (1 - estimated proportion reported). With $4 million reported, $10 million expected, and 60% reported: $4 million + $10 million x 0.40 = $8 million. Simplified IBNR over reported losses = $8 million - $4 million = $4 million.Case study
Seen in the real world.
Fictional example: Vale Mutual launched a specialty policy and saw just $2 million in claims reported after one year. Its finance director proposed using $2 million as the final cost because claims appeared quiet. Actuary Imran explained that similar claims often arrived later and selected a 25% reported proportion for the age of the policies. Vale's initial expected ultimate loss was $8 million.
The unreported proportion was 75%, making the additional estimate $6 million and ultimate losses $8 million when added to $2 million reported. Imran showed a sensitivity table using alternative emergence assumptions and checked whether the prior loss ratio still fit the new policy. The board recorded the assumptions and asked for quarterly comparisons of actual claim development with the estimate.
Watch out
Common mistakes.
- Treating reported losses to date as ultimate losses when material claims may still emerge.
- Using a paid-loss development pattern with reported-loss amounts without reconciling the definitions.
- Presenting an estimate as certain despite an outdated expected loss ratio or unstable emergence pattern.
Questions
People also ask.
What does IBNR mean here?
It refers to losses incurred but not yet reported; simplified calculations subtract reported losses from estimated ultimate losses.
Why use an expected loss ratio?
It supplies a prior estimate for the portion of claims that has not emerged, especially useful early on.
Is it better than chain ladder?
Neither is universally better. Data maturity, claim volatility, assumptions, and validation determine which estimate is useful.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%Related