Bayesian optimal investment and reinsurance with dependent financial and insurance risks (Q2135611)

A Cramér-Lundberg risk model is considered where proportional reinsurance may be taken and the surplus is invested in a Black-Scholes market. The claim size distribution is not completely known. One parameter, taking a value in a finite set, has to be observed from the claims history in a Bayesian setup. Moreover, a claim exceeding a threshold \(L\) leads to a proportional drop of the risky asset. The goal is to maximise the expected terminal exponential utility. For this purpose, the insurer can decide continuously about the amount invested into the risky asset and how much proportional reinsurance is bought. After determining the dynamics of the process of the Bayesian probabilities of the parameter of the claim size distribution (filtering), the corresponding Hamilton-Jacobi-Bellman equation is determined. Because it is not clear whether the solution is differentiable, the derivative has to be interpreted as Clarke's generalized subdifferential. A verification theorem is proved and the existence of a solution is verified. Some special cases and a numerical example are explicitly discussed.