Following Clark (CAS) and the Swiss Re technical literature, disciplined layer pricing reconciles two independent estimates.
| Experience rating | Exposure rating |
|---|
| Data source | Cedant own historical losses | Industry exposure curves / ILFs |
| Strength | Reflects the actual book | Stable for high / thin layers |
| Weakness | Unreliable for rare layers | May not fit the specific book |
| Best when | Credible, stable data | Sparse data or high attachment |
Experience rating, the steps
- Collect historical losses; cap and cut to the layer \\(\ell\text{ xs }d\\).
- Trend losses for inflation to the prospective period.
- Develop to ultimate for late reporting / development.
- On-level the exposure base; divide to get the loss cost.
Exposure rating, the steps
- Estimate the ground-up expected loss by band of sum insured.
- Apply the exposure curve \\(G\\) (or ILFs) to allocate loss to the layer: share \\(=G(b)-G(a)\\).
- Aggregate across bands to the layer loss cost.
Reconciliation by credibility
The two estimates are blended by credibility \\(Z\\), with \\(Z\\) reflecting the volume and stability of the experience. The result is a single, defensible technical loss cost, the foundation of Power Re's pricing discipline.
