Mean-variance asset-liability management under constant elasticity of variance process

This paper investigates a mean-variance asset-liability management (ALM) problem under the constant elasticity of variance (CEV) process. The company can invest in n+1 assets: one risk-free bond and n risky stocks. The uncontrollable liability process is modelled by a geometric Brownian motion. The feasibility is studied and potential optimal portfolio is proven to be admissible. We derive the efficient frontier and efficient feedback portfolio in terms of the solutions of two backward stochastic differential equations (BSDEs), which do not admit analytical solutions in general. The closed form solutions are obtained under some special cases. Applying the Monte Carlo simulation, we provide several numerical examples to demonstrate how the efficient frontier is influenced by the relevant parameters.