It is possible to model a wide range of portfolio management problems using stochastic programming. This approach requires the generation of input scenarios and probabilities, which represent the evolution of the return on investment, the stream of liabilities and other random phenomena of the problem and respect the no-arbitrage properties. The quality of the recommended capital allocation depends on the quality of the input scenarios and a validation of results is necessary. We propose scenario generation techniques and for output analysis in the context of defined contribution pension fund management. The application to the specific case of a Czech pension fund indicates the components that influence the recommended investment decisions and the fund's results. The initial position of the pension fund is important because of the accounting rules in the model and tracking both the market and purchasing valuation of assets.
展开▼
