Goal programming approach for hedging a portfolio with financial futures: An empirical test

Abstract

Purpose of Study: The primary purpose of this study is to test the Sharda and Musser goal programming hedging model in a portfolio environment employing real world data. The model is modified to accommodate a portfolio of securities, refined to include priorities and previous week's hedging information, and is also condensed to exclude constraints pertinent to past week's hedging activities. Results of the model are compared to those obtained from implementing the static hedge ratio models, the original GP model, the GP-naive model as well as to the best case scenario using perfect forecasts. Performance evaluation is based on four criteria; ending portfolio value, riskiness of the strategy, risk return tradeoffs, and the number of positive quarters.

Findings and Conclusions: Findings from the study reinforce conclusions from the earlier works which employed the goal programming approach. The condensed GP model was far superior than the original GP model and the GP-naive model in providing consistent net values. It also outperformed all the other ratio-related strategies in almost all of the criteria concerned. When actual, historical data were used, the model's performance improved substantially. Forecasting inaccuracies remain the major factor in impeding the model's potential performance. Putting it aside, the goal programming approach to hedging appears to perform remarkably well even in a portfolio environment.