Two-sided Lundberg inequalities in a Markovian environment (Q1324885)

An insurance risk process \(X(t)\) is considered, such that the risk process has a premium rate, a Poisson arrival intensity and a claim size distribution depending on the state \(Y(t)\) of a background Markov process on a general state space. Let \(\psi_ y (u)\) be the ruin probability with initial reserve \(u\) and initial state \(Y(0) = y\). A Lundberg exponent \(R\) is defined and inequalities of the form \(C_ - h(y)e^{- Ru} \leq \psi (u) \leq C_ + h(y) e^{-Ru}\) are proved for suitable constants \(C_ -\), \(C_ +\) and a suitable function \(h(y)\). This extends results of the reviewer and \textit{T. Rolski} [Math. Oper. Res. 19, No. 2, 410-433 (1994; Zbl 0801.60091)], who consider a finite state space for \(Y(t)\).