Home / Financial Tools / P&C Loss Reserving: Chain-Ladder & Bornhuetter-Ferguson Development Engine

Graduate-Level Modeling · Property & Casualty Actuarial Reserving

P&C Loss Reserving: Chain-Ladder & Bornhuetter-Ferguson Development Engine

The newest accident year in every triangle is the one Chain-Ladder trusts least and gets wrong most. Bornhuetter-Ferguson exists for exactly that reason.

How To Use This Model

Reading This Tool

Generate a realistic loss development triangle from premium, loss ratio, development speed and volatility assumptions, then run both the Chain-Ladder and Bornhuetter-Ferguson reserving methods against it side by side.

Set your own Bornhuetter-Ferguson prior independently of what actually generated the triangle, and watch exactly what the textbooks describe: BF leans hard on your prior for the least mature accident years, and converges to pure Chain-Ladder as an accident year matures.

Triangle Generation

Number Of Accident Years 8

Accident Year 1 Earned Premium $50M

Premium Growth Rate 6.0%

Underlying (True) Loss Ratio 65%

Development Speed 0.60

Tail Factor (Beyond Observed Triangle) 1.020

Accident-Year Volatility (Noise) 15%

Bornhuetter-Ferguson Prior

BF A Priori Expected Loss Ratio 65%

The BF prior is set independently of the underlying loss ratio that actually generated the triangle, mismatch them deliberately to see how BF behaves when the actuary's a priori assumption is wrong.

Triangle Summary

Total Earned Premium

Total Paid To Date (Latest Diagonal)

Chain-Ladder Total Ultimate

Implied Ultimate Loss Ratio

Completed Loss Triangle (Known + Chain-Ladder Projected)

Property & Casualty Actuarial Reserving

The Core Formulas

Age-To-Age Factor: f_j = Σ_i C_i,j+1 / Σ_i C_i,j (volume-weighted, upper triangle only)
Cumulative Development Factor: CDF_j = f_j·f_j+1·…·f_n−1·Tail
Chain-Ladder Ultimate: U_i = C_i,latest·CDF_latest
Bornhuetter-Ferguson Ultimate: U_i = C_i,latest + (Premium_i·ELR)·(1 − 1/CDF_latest)

When To Actually Use Each Method

Chain-Ladder: mature, stable lines of business with a long, credible development history and no major volume swings.
Bornhuetter-Ferguson: newer or fast-growing books, catastrophe-exposed lines, or any accident year too immature for Chain-Ladder's leverage on a tiny reported amount to be trustworthy.
Teaching property and casualty actuarial reserving methodology in an actuarial science or general insurance course.

Key Assumptions & Limitations

This tool generates a synthetic triangle from a smooth parametric development pattern with random noise, real triangles are built from actual claims data and often show irregular patterns Chain-Ladder alone cannot smooth away.
Uses paid-loss development only, real reserving practice typically also runs an incurred (paid plus case reserves) triangle and reconciles the two.
A full actuarial opinion would also test multiple development factor selections (simple average, weighted average, excluding outliers) rather than a single volume-weighted average, and would apply Mack's method or bootstrapping to quantify reserve variability.

Foundational References

Friedland, J. (2010). Estimating Unpaid Claims Using Basic Techniques. Casualty Actuarial Society.

Bornhuetter, R. L., & Ferguson, R. E. (1972). The Actuary and IBNR. Proceedings of the Casualty Actuarial Society, 59, 181-195.

Want the life insurance side of actuarial reserving too?

Open The Life Contingencies Engine →