Loss Distributions

Loss distributions are currently used by loss reserve analysis in special situations to estimate loss reserves, or to estimate the variability of loss reserve estimates. An example of such a special situation is excess or catastrophe losses, where data is frequently sparse and traditional techniques are difficult to apply. This session will introduce those unfamiliar with the use of loss distributions in insurance to models that have been used to analyze reserving data. The panelists will discuss the uses of one or two parameter distributions. Specifically, the single parameter and shifted pareto and the lognormal distribution will be used to illustrate the applications of loss distributions. The shortcomings of the distributions, such as goodness of fit of the tail will be discussed. Application of the distribution to practical reserving situations including the evaluation of per occurrence and aggregate excess loss reserves and the estimation of reserve variability will be illustrated. This session assumes no prior knowledge of loss distribution except for what is covered on the Part 2 exam.