Supporting financial and management decisions for insurance companies with the SDDP method

This work proposes a novel asset and liability management (ALM) model for insurance companies that not only provides investment policies under solvency/liquidity requirements, but also includes the effect of business and financial decisions on their earnings. It does so by measuring the exposure taken when selling different volumes of contracts, assessing the amount of reinsurance needed to cover the losses from possible claims, and incorporating debt financing as an alternative to expanding business operations. The model includes a quadratic reinsurance cost as a function of the reinsurance taken and a flexible structure to incorporate different types of claims. The stochastic dual dynamic programming (SDDP) method is applied to cope with the long-term nature of this multistage stochastic programming (MSP) model. A proven open-source package is used to implement the SDDP. Using empirical data on asset prices, financial statements from the insurance industry, and a database of vehicle insurance policies, the method finds solutions for 30-year instances in 1 h on average. Tests show that the solutions obtained are sensible in the way that the asset allocation and financial leverage change under different economic regimes, insurance severity, and the degree of deviation from the equity-growth target. Sensitivity analysis shows how the methodology can find the right combination of solvency ratios and reinsurance levels to deal with the trade-off between the profits from selling more insurance and the resulting increased exposure to future claims.