A Stochastic Linear Programming Model for Asset Liability Management: The Case of an Indian Insurance Company

Abstract

Asset - Liability management is one of the most critical tasks for any financial institution, determining its cushion against the risk and the net returns. The problem of asset liability management for an insurance company requires matching the cash inflows from premium collections and investment income with the cash outflows due to casualty and maturity claims. Thus, what is required is a prudent investment strategy such that the returns earned on the assets match the liability claims at all points of time in future. Conventionally, the asset allocation has been done using the Mean Variance approach due to Markowitz (1952, 1959). While such a strategy ensures that the asset value always match or are greater than the liability for the next year, it does not maximise the net worth of the firm nor does it take care of all the cash inflows and outflows over a long term period. A stochastic linear programming model (on the lines of Pirbhai, 2004) maximises the net worth of the firm and also takes care of the uncertainties. While there are instances of stochastic linear programming being applied for ALM in financial institutions in developed markets, no such practical application has been reported in this area in Indian context as yet.

In this paper, we describe the development of a multi stage stochastic linear programming model for insurance companies. The multi-stage stochastic linear programming model was developed on the modelling language AMPL (Fourer, 2002).