Ensemble distributional forecasting for insurance loss reserving (Q6632363)

The authors propose a systematic framework to objectively combine multiple stochastic loss reserving models such that the strengths offered by different models can be utilised effectively. The framework contains two main innovations compared to existing literature and practice. Firstly, the model combination considers the full distributional properties of the ensemble and not just the central estimate. Secondly, the framework is tailored for the features inherent to reserving data. These include, for instance, accident, development, calendar, and claim maturity effects. The framework is illustrated with a complex synthetic dataset. The framework can be implemented thanks to an R package, ADLP, which is available from CRAN.