Triphonas Kyriakis, Cormac Lucas, Gautam Mitra
     Brunel University
     Department of Mathematical Sciences

Markowitz mean variance model is a static model; also it is well accepted that it does not capture and adequately represent decision-making in a dynamic environment. Further, this model incorporates a symmetric measure of risk that nowadays is not considered appropriate in real life financial applications. Stochastic programming model (two-stage and multistage) applications with multiple planning periods can capture the dynamic nature of decision-making. We report our investigation of a two-stage and multistage stochastic programming model for asset and liability management. The stochastic programming model is augmented to include (a) downside risk and (b) VaR constraints. Thereby we are able to qualify the financial decision-making with risk measures. This model is rolled over five time periods and back tested using historical data. The performance of each risk measure is compared with a mean variance model, rolled and back-tested using the same historical data set.