Chapter 13 - Stochastic programming models for strategic and tactical asset allocation—A study from Norwegian life insurance

Abstract

This chapter describes the development and use of two decision support models for asset allocation used within the Gjensidige-NOR Group. For strategic, long-term, asset liability management, the life insurance company within the Group uses an asset and liability management (ALM)-model. GN Asset Management, the asset management company within the group, uses a different model for shorter term tactical asset allocation in a hedge fund. Both models are based on stochastic programming. The models have been developed in close cooperation between GN Asset Management and the Norwegian University of Science and Technology. This chapter describes the institutional setting, with an emphasis on structure, relevant parts of the legal framework and the competitive situation. It outlines the models and motivates the chosen modeling frameworks. It discusses the experiences regarding interactions with users during the developing process, the time spent in different phases of the project, possible pitfalls in the developing phase, and how the models have changed the organization. A model is no better than its input. Hence, it provides a detailed description of the scenario-generation procedure, from data collection and the formation of market expectations, to the final scenario tree needed by the stochastic programming models. It provides a case study which examines the whole investment process, from establishing market expectations and generating scenarios, to constructing, implementing, and managing the portfolio. It concludes by presenting the track record of the hedge fund.