The sustainability of Malaysia's defined contribution pension system : implementation of deterministic linear programming

This article investigates, under deterministic linear programming model, asset allocation decision and optimal investment strategy for Malaysia's defined contribution pension (DC) scheme -Employees Provident Fund (EPF). The model requires generation of scenarios and probabilities to represent future assets and liabilities streams. We employed Vector Autoregressive (VAR) model to generate future returns of five asset classes i.e. equity, money market instrument, Malaysia government bond with 1 and 10 years of maturity date and property. Future liabilities factors were derived from two sub-models; population and salary. In population model, the future status of the EPF members was determined using a Markov Chain model. Then, the random factors of assets and liabilities were used in the asset liability model (ALM) based on linear programming (LP) and fixed mix (FM) strategy. The results of the research are grouped in two levels. First, we briefly discuss the finding of the random factor model and then we analyse the optimal investment strategy for the EPF. In terms of finding an optimal investment strategy, the FM strategy generated higher expected terminal wealth than the LP strategy. This finding suggests that FM strategy is superior to the LP strategy. In addition, we find that the higher dividend distributed to the members may result in decreasing of the expected terminal wealth of the fund for both strategies. This portrays that dividend distribution policy may affect the financial soundness of the EPF in the long run.