Time-consistent mean-variance investment with unit linked life insurance contracts in a jump-diffusion setting (Q2234757)

In this paper, the authors discuss a time-consistent mean-variance portfolio optimization problem of an insurer with unit-linked life insurance contracts in the presence of mortality risk under a continuous-time modelling framework. Two key sources of risks, namely the longevity risk due to unexpected changes in mortality rates and the market risk attributed to fluctuations in asset prices, are incorporated into the model set up. It is assumed that both the price processes of financial assets and the force of mortality follow Markovian jump-diffusion processes. The total variance of terminal wealth is used to quantify risk. The optimal solution to the mean-variance problem is determined by a Nash subgame perfect equilibrium. In this game-theoretic framework, the value function and an optimal strategy are characterized by the solution of a system of nonlinear partial integro-differential equations (PIDEs), which is called the extended Hamilton-Jacobi-Bellman (HJB) system. The authors show that the game-theoretic equilibrium necessarily satisfies the extended HJB system in the jump-diffusion model set up. They also derive expressions for the solution to the extended HJB system and optimal trading strategies when the hedge of some terminal payoff is ignored by the insurer. Section 2 of the paper presents the model set up and some model assumptions. Specifically, the price processes of several risky stocks are assumed to be governed by geometric jump-diffusion processes. The force of mortality process and the dollar value process of a zero-coupon longevity bond are supposed to be governed by Markovian jump-diffusion processes. The mean-variance optimization problem is formulated in Section 3. The key idea of the problem is to minimize the variance of the terminal value of a portfolio and maximize its expected value at the same time. Using the mean-variance criterion, the insurer aims to hedge a terminal payoff through trading in the financial and longevity markets. The performance functional of the insurer based on the mean-variance criterion is presented in Definition 3.4. In the game-theoretic set up, an equilibrium control and the equilibrium value function are introduced in Definition 3.6. Specifically, it is supposed that the equilibrium control is a feedback control. The extended HJB system satisfied by the value function and the optimal trading strategies is specified in Definition 3.7. The main theoretical results are presented in Section 4, where two verification theorems, one for sufficiency and another one for necessity, are provided. Theorem 4.3 gives the sufficient result and states that the sufficient condition for the existence of an equilibrium control and the equilibrium value function is that the extended HJB system has a solution. Theorem 4.4 gives the necessary result and states that an equilibrium control is necessarily a solution to the extended HJB system. Section 5 gives an explicit solution to the problem under the assumption that the hedge of the terminal payoff is ignored. Specifically, Theorem 5.1 gives a decomposition for the equilibrium value function, a decomposition for the expected optimal terminal wealth as well as expressions for optimal investments in the risky stocks and the longevity bond. Section 6 presents numerical analyses of the results based on simulation. Specifically, Section 6.1 provides numerical analyses for the investment problem whose solution is presented in Theorem 5.1. Section 6.2 gives numerical results for an optimal investment with a terminal payoff obligation. Numerical comparisons with pre-commitment strategies are presented in Section 6.3.