What About an Alternative Chain-Ladder Predictor for the Ultimate Claims Amount?

We highlight that the classical chain-ladder ultimate claims predictor for the total over all accident years just results to be conditionally unbiased given the first triangle column and propose a correction term making it conditionally unbiased given the latest triangle diagonal. This might be an enhancement since averaging over triangles with the same latest triangle diagonal is preferable then the classical averaging over triangles with the same first triangle column. However, unfortunately, the required correction term cannot be calculated within the famous Mack’s chain-ladder model but only estimated according to a backwards simulation procedure inspired by the related time series model. This leads to an alternative ultimate claims predictor which may be worth to be considered as an alternative to the classical one, as also illustrated in a numerical example