Big Data Analytics is an important and flexible tool available for data analysis and informed decision making. In this paper, we look at the use of Big Data Analytics in asset liability management and asset allocation in uncertain economic situations using stochastic linear programming (SLP). In particular, this paper is an extension of our earlier work and we contribute to the existing literature by conducting experiments on the stochastic model through DSS. In particular, for this SLP based DSS, we address issues like the optimal number of scenarios required for good results, and the impact of the change in the number of scenarios on the stability of the model. The paper also addresses the impact of the change in the number of scenarios on the policy holders’ as well as shareholders’ reserves. In particular, we show the relevance of employing a larger number of scenarios and also present the experimental design developed to test the relevance of this model. We also show that a stochastic model employing fewer scenarios produced marked improvements in both return side measures as well as risk side measures compared to a mean value model or a partial mean value model.
Big Data Analytics is an important and flexible tool available for data analysis and informed decision making. In this paper, we look at the use of Big Data Analytics in asset liability management and asset allocation in uncertain economic situations using stochastic linear programming (SLP). In particular, this paper is an extension of our earlier work and we contribute to the existing literature by conducting experiments on the stochastic model through DSS. In particular, for this SLP based DSS, we address issues like the optimal number of scenarios required for good results, and the impact of the change in the number of scenarios on the stability of the model. The paper also addresses the impact of the change in the number of scenarios on the policy holders’ as well as shareholders’ reserves. In particular, we show the relevance of employing a larger number of scenarios and also present the experimental design developed to test the relevance of this model. We also show that a stochastic model employing fewer scenarios produced marked improvements in both return side measures as well as risk side measures compared to a mean value model or a partial mean value model.
