Hitting probabilities in a Markov additive process with linear movements and upward jumps: applications to risk and queueing processes. (Q1879901)

A risk process in which premium rates, claim arrivals and claim sizes depend on a background state, which is governed by a continuous time Markov chain, is considered. The author considers a risk process with positive and negative premium rates. Asymptotic behavior of the ruin probability when the initial reserve goes to infinity is studied. The author models the risk process as a Markov additive process \(Y(t)\) with upward jumps and \(Y(0) =0.\) For \(x \geq 0\), \(x - Y(t)\) is a risk process with initial reserve \(x\) and ruin occurs when \(Y(t)\) hits level \(x.\) The background continuous-time Markov chain has a finite state space. The additive component linearly increases or decreases while the background state is unchanged. The author introduces the additive process with upward jumps under the stationary regime and computes the ascending ladder height distribution in terms of the dual process and he specializes this result to the Markov additive processes. The author derives a Markov renewal equation to the hitting probabilities. He shows that these probabilities have asymptotically exponential decay. The results apply to risk processes and fluid queues.