Coupling a memetic algorithm to simulation models for promising multi-period asset allocations

Abstract

The use of simulation models to manage and regulate property–liability insurers has gained in popularity over the last decade. This paper introduces a hybridized search optimization algorithm, also known as a Memetic Algorithm, for use with these insurer simulation models. The proposed algorithm combines the merits of both local and global search optimization techniques, and provides an efficient and robust approach for insurance model application. Our research investigated whether this enhanced optimization algorithm could further improve the results of a simulation model. As part of this investigation, a company-wide simulation model of a property–liability insurer was coupled with the proposed hybrid algorithm to tackle a typical multi-period asset allocation problem. The resulting asset allocations obtained by the proposed memetic algorithm coupled with the simulation model demonstrated better results than currently available investment strategies. The significant and robust improvements put forth in the present research demonstrate the great potential of our multi-phase hybrid algorithm in enhancing simulation model capabilities.