Proxies

This paper illustrates a Bayesian method to estimate the predictive distribution for outstanding loss liabilities that can be applied when there is either insufficient data or little actuarial expertise. The assumptions made are that the unpaid losses can be described by the collective risk model and that the scenarios that make up the prior distribution contain the possible loss ratio and loss development factors. As one should expect, the more data one puts into the method, the tighter the predictive distribution will be.

Keywords: Reserving Methods, Reserve Variability, Uncertainty and Ranges, Collective Risk Model, Fourier Methods, Bayesian Estimation