Discussion on “Experience rating in the classic Markov chain life insurance setting: an empirical Bayes and multivariate frailty approach” (Furrer)

Excerpt

A challenge in life insurance is the shrinkage estimation for well-defined groups of insured persons, especially taking into account transitions from one group to another. A classic example of this is the Disability Insurance, which assumes an “active” death order to estimate a probability of “disabled” from “active”, “dead” from “active” and “dead” from “disabled”. The procedure is based on the assumption that there is a Markov chain with jump processes in order to estimate the parameters of an a priori distribution from empirical data and to optimize them with a maximum likelihood approach (empirical Bayes). In practice, the requirement has arisen for a multivariate extension of the concept of multivariate frailty, that is, different groups of “active” arise through further diversification, for example, by “occupation” or causes and extent of “disabled” (frailty approach, ADL). The variables for the parameterization of the shrinkage estimators are usually not directly observable (latent) and they are no longer independent. …