On the analysis of a class of loss models incorporating time dependence (Q362057)

A unified model for quantities related to the claims on a portfolio of insurance business incorporating the incidence and severity of claims is defined and analyzed. Assume that the number of claims which result from a claim-causing event occurring at time \(x\) has Poisson probability generating function \(B_x ( z )\) and for each claim incurred at time \(x\) there is a random reporting lag with distribution function \(W_x(\cdot)=1-\overline{W}_x(\cdot)\) until the claim is reported, independent of all others. Suppose that the value of a claim incurred at time \(x \in (0, t]\) depends on \(x\) as well as \(t\), has distribution function \(F_{t,x}(y)\) and Laplace-Stieljes transform \(\tilde{f}_{t,x}(s) = \int_0^\infty e^{-sy} dF_{t,x} (y)\). The inflation and incurred-but-not-reported claims models are unified and generalized to a model with inflation as \[ \tilde{f}_{t,x}(s) = B_x\{W_x(t-x) + \overline{W}_x(t-x)\tilde f( s/\sigma x ) \}. \] As particular cases this model includes the usual aggregate claims models with or without inflation, as well as a model for the incurred-but-not-reported claims, also with or without inflation. Connections between the inflation and incurred-but-not-reported claims models are established, and the notions of self-decomposability and discrete self-decomposability are seen to be relevant. A numerical example is considered to demonstrate how Panjer-type recursions may be employed to evaluate distributions of interest.