Optimal dividend and risk control policies in the presence of a fixed transaction cost (Q2223849)

The optimal dividend problem for a diffusion approximation to a risk model \(R_{1,t} = x + \mu t + \sigma W_t\) is considered. Introducing proportional reinsurance where the premia are split fairly between insurer and reinsurer, the underlying process becomes \[ R_{2 t} = x + \int_0^t a_s \mu\; d s + \int_0^t a_s \sigma \;d W_s \;. \] In addition, the authors introduce costs \(I > 0\) for the first underwriting of reinsurance at the stopping time \(\tau^\pi\). This modifies the surplus to \[ R_t^\pi= x + \int_0^t a_s^\pi \mu\; d s + \int_0^t a_s^\pi \sigma \;d W_s - L_t - I \mathbb{I}_{t \ge \tau^\pi}\;, \] where \(a_t^\pi = 1\) for \(t < \tau^\pi\), \(a_t^\pi \in [0,1]\) for \(t \ge \tau^\pi\) and \(L_t\) denotes the accumulated dividend payments. That is, \[ R_t^\pi = R_{1,\tau^\pi \wedge t} + R_{2 \tau^\pi \vee t} - R_{2 \tau^\pi} - L_t - I \mathbb{I}_{t \ge \tau^\pi}\;. \] The goal is to determine the optimal strategy \((L_t, \tau^\pi,a_t)\). The solutions \(V_1(x)\) to the problem with no reinsurance and \(V_2(x)\) to the problem without costs (\(I = 0\)) are known. The problem stated is therefore an optimal stopping problem and is solved in the paper. For me, the question arises, why future changes of the retention level is free of costs. It would be interesting to solve the problem where each change of the retention level triggers the fixed costs \(I\).