Joint density of the number of claims until ruin and the time to ruin in the delayed renewal risk model with Erlang(\(n\)) claims (Q1936140)

Let \(U(t) = u + c t - \sum_{k=1}^{N_t} X_k\) be the surplus of an insurance portfolio, where \(u\) is the initial capital, \(c\) is the premium, \(N\) is a (delayed) renewal process and \(\{X_k\}\) is an i.i.d. sequence, independent of \(N\). In particular, the distribution of \(X_k\) is an Erlang distribution. Let \(\tau = \inf\{ t \geq 0: U(t) < 0\}\) denote the time to ruin. The quantity of interest is \(P[N_\tau = k+1, 0< \tau \leq t]\), or the density \(p(k+1,t) = {\text{d} \over \text{d} t} P[N_\tau = k+1, 0< \tau \leq t]\). For this purpose, one first conditions on \(T_0\) and considers \[ p(k+1,t \mid v) = {\text{d} \over \text{d} t} P[N_\tau = k+1, v\leq \tau \leq t \mid T_0 = v]. \] Following an idea of \textit{K. A. Borovkov} and \textit{D. C. M. Dickson} [Insur. Math. Econ. 42, No. 3, 1104--1108 (2008; Zbl 1141.91486)], the problem can be transferred to a dual problem where the rôle of the claim size distribution and of the interarrival time is interchanged. Following an approach by \textit{D. Landriault, T. Shi} and \textit{G. E. Willmot} [Insur. Math. Econ. 49, No. 3, 371--379 (2011; Zbl 1229.91161)], \(p(k+1,t \mid v)\) is found. Integration over \(v\) gives \(p(k+1,t)\). It is used that \(p(1,t)\) (ruin at the first occurrence of a claim) is easy to find, and that for \(p(k,t)\) with \(k > 1\) ruin is not allowed to occur at the first occurrence time. The main idea of the proof is that for Erlang interarrival times, the joint Laplace transform \(E[r^{N_\tau} e^{-\delta \tau}; \tau < \infty]\) can be found explicitly in terms of the Laplace transform \(E[e^{s X_k}]\) of the claim times by techniques used for phase-type distributions. In particular, the approach would also work for a phase-type claim size distribution. The inversion of the transform is done via Lagrange's expansion theorem. Several examples are discussed.