par Lefèvre, Claude ;Coulibaly, I.
Référence Insurance. Mathematics & economics, 42, page (935-942)
Publication Publié, 2008
Article révisé par les pairs
Résumé : This note discusses a simple quasi-Monte Carlo method to evaluate numerically the ultimate ruin probability in the classical compound Poisson risk model. The key point is the Pollaczek-Khintchine representation of the non-ruin probability as a series of convolutions. Our suggestion is to truncate the series at some appropriate level and to evaluate the remaining convolution integrals by quasi-Monte Carlo techniques. For illustration, this approximation procedure is applied when claim sizes have an exponential or generalized Pareto distribution. © 2007 Elsevier B.V. All rights reserved.