Optimal dividend policy in an insurance company with contagious arrivals of claims (Q829004)

A Cox risk model is considered where the claim intensity process \(\{N_t\}\) follows a Hawkes process. That is, the intensity is modelled as \[ \lambda_t = \bar\lambda + (\lambda_0 - \bar \lambda) e^{-\alpha t} + \beta \int_0^t e^{\alpha(s-t)}\;d N_s\;.\] Thus, each claim arrival triggers an increase \(\beta\) of the intensity process. Between claims, the intensity decreases with rate \(\alpha\) towards the lower bound \(\bar\lambda\). The surplus of the insurer is then \[ X_t^0 = x + p t - \sum_{k=1}^{N_t} Y_i\;,\] where the claim sizes \(\{Y_i\}\) are positive iid variables independent of \(N\). The insurer can pay a dividend. The accumulated dividend process is \(\{L_t\}\), which should be a caglad non-decreasing process with \(L_0 = 0\). Hence, the surplus becomes then \[ X_t^l = x + p t - \sum_{k=1}^{N_t} Y_i - L_t\;.\] The time to ruin is \(\tau^l = \inf\{t > 0: X_t^l< 0\}\). The goal is to maximise the value of a dividend strategy \(V_l(x,\lambda_0) = \mathbb{E}[\int_0^{\tau^l} e^{-c s}\; d L_s]\). This yields the value function \(V(x,\lambda_0) = \sup_l V_l(x,\lambda_0)\), where the \(\sup\) is taken over all adapted strategies, such that ruin is not caused by a dividend payment. It is shown that the value function is the minimal viscosity solution to the corresponding Hamilton-Jacobi-Bellman equation. For some of the proofs a reference to other papers is given where one can find an analogous proof. One of the result states that \(V(x,\lambda_0)\) is concave in \(x\). I doubt that this is correct. Letting \(\alpha \to \infty\) yields in the limit the classical compound Poisson model with constant intensity \(\bar\lambda\). [\textit{P. Azcue} and \textit{N. Muler}, Math. Finance 15, No. 2, 261--308 (2005; Zbl 1136.91016)] have constructed an example where the optimal strategy consist of several bands and thus is not concave.