Methods for estimating the reserve risk of incurred but not reported provisions are usually based on the assumption that these provisions were calculated using the chain ladder method. Thus, in practice, these methods relate to various ways of quantifying the prediction error of the chain-ladder estimates, often the bootstrap procedure of England and Verall (2011) for Mack’s chain-ladder method.
On the other hand, incurred but not reported estimates are often quantified using exposure-based methods such as the Bornhuetter-Ferguson or Cape-Cod methods. We define bootstrap procedures for the Cape-Cod method, show how these can be applied and how the results compare to more established methods used in practice. We consider which of the methods are more appropriate for use in reserve risk estimation under Solvency II and accounting estimates in the context of IFRS 17, with a focus on stability and realism of the results. Finally, we provide R code to reproduce these methods on representative data.
Practical outcomes
Ronald Richman – Old Mutual Insure