Approximations of the ruin probability in a discrete time risk model (Q2218139)

The main result of the paper consists of an approximation technique, by which the ultimate ruin probability in the classical Gerber-Dickson risk model can be derived, when claims follow a negative binomial mixture distribution. In particular, based on this methodology, the ruin probability is expressed as the expected value of a deterministic sequence \(\{Ck\}\), with \(k\) the value of a negative binomial distribution. This sequence is obtained recursively. Then, the result is framed within a mixed Poisson scenario. Numerical examples show the approximation methods when the mixing distribution is Erlang, Pareto and lognormal.