Paperlane

no. 16 · Tuesday 17 November 2026

Asymptotics for a continuous-time risk model with multivariate strongly regularly varying claims and stochastic investment returns

Zhangting Chen and Charalampos D. Passalidis

math.PR · via q-fin.RM

Abstract

In this paper we study the finite-time asymptotic behavior of the entrance probability of the discounted aggregate claims into an extreme region in a multidimensional risk model. Our non-standard risk model is not restricted to the Levy-Renewal framework, that represents a usual assumption for the stochastic returns and the counting process of the claim vectors, respectively. For the modeling of the distribution of claim vectors, we introduce the notion of set-based multivariate (strongly) regularly varying distributions. For the new set-based classes we provide certain closure properties and some examples to show the difference from the usual definitions of multivariate regularly varying distributions. This set-based framework provides a broad and flexible framework to the actuarial applications, and facilitates the modeling of dependence among the claim vectors that arrive at different time epochs. The two main results of the paper establish a unified asymptotic expression, that demonstrates the presence of multivariate linear single big jump principle. Despite of the assumptions the asymptotic expression remains explicit. In the first main result we suppose that the common distribution of the claim vectors belongs to the class of set-based multivariate regularly varying distributions, and the stochastic returns satisfy a general moment condition. In the second main result, we restrict the claim vectors in the class of setbased multivariate strongly regularly varying distributions, relaxing the moment condition on the stochastic returns. Finally, we conduct a numerical study to illustrate the accuracy of the proposed asymptotic approximation and investigate its sensitivity to the main model parameters.

fields
Probability
pdf ↗ page arXiv:2611.18681 ↗