Tools › Finance and Business › Insurance › Loss Reserving › Example
Mack 1993: chain-ladder reserves and standard errors of the Taylor–Ashe triangle
| Tool | Loss Reserving |
| Source | Mack, T. (1993). "Distribution-free Calculation of the Standard Error of Chain Ladder Reserve Estimates." ASTIN Bulletin, Vol. 23, No. 2, pp. 213–225. Tables 1–3. |
| Level | Professional |
Problem
A 10 × 10 cumulative paid-claims run-off triangle (Taylor and Ashe 1983, accident years 1 to 10) is given. Estimate the age-to-age development factors, the variance parameters σ_k², the reserve of every accident year and the total, and the standard error of each reserve by Mack's distribution-free formulas.
Inputs to Enter
- Cumulative triangle, 10 accident years × 10 development ages (loaded by default; first row 357,848 … 3,901,463, last row 344,014)
- Volume-weighted averages over all years, no tail factor
Answers
- f_k = 3.49, 1.75, 1.46, 1.174, 1.104, 1.086, 1.054, 1.077, 1.018
- σ_k²/1000 = 160, 37.7, 42.0, 15.2, 13.7, 8.19, 0.447, 1.15, 0.447 (the paper prints 0.477 for k = 9, a misprint; Mack's own rule gives 0.447)
- Reserves (thousands) by accident year 2–10: 95, 470, 710, 985, 1,419, 2,178, 3,920, 4,279, 4,626; total 18,681 (tool 18,680,856)
- Standard error as % of the reserve: 80, 26, 19, 27, 29, 26, 22, 23, 29; total 13 % (tool 2,447,095, equal to R's MackChainLadder(GenIns))
- Total ultimate 53,038,946; the ODP GLM on the same triangle gives the same reserves with dispersion φ = 52.6 thousand (ChainLadder glmReserve)
Notes
The tool's default inputs are this triangle; press Submit to reproduce it. The Mack standard errors are given for the volume-weighted ladder without a tail factor.
More Examples for This Tool
- ChainLadder vignette: RAA triangle by Mack, the LDF exhibit with a 1.05 tail and the ODP bootstrap
- Friedland 2010: U.S. Industry Auto reported claims by development, expected claims, Bornhuetter–Ferguson and Cape Cod
- England and Verrall 2001 Table 6.1: chain-ladder reserve and ODP prediction error of a triangle with a negative increment