COMPARATIVE RISK ANALYSIS BETWEEN THE MARKOWITZ QUADRATIC PROGRAMMING MODEL AND THE MULTIVARIATE COPULA MODEL FOR A SINGAPORE REITS PORTFOLIO

Abstract

While Markowitz Portfolio Theory (MPT) has been the main method in modern finance to construct portfolios with maximized returns for given risks, a more flexible model, the Multivariate Copula model, is introduced in recent years that enables investors and portfolio managers to obtain optimal portfolio even when the MPT assumption of returns that conform to a normal distribution does not hold true. This forms the impetus of this study, which seeks to carry out a comparative analysis to evaluate how the Multivariate copula model fare against the MPT in capturing risks for an S-REITs portfolio. The portfolio comprises seven pioneering S-REITs. Initially, an integrated Analytical Hierarchy Process (AHP) Strategic Asset Allocation (SAA) model is adopted to create a diversified portfolio. An efficient frontier is constructed and analyzed under the Markowitz Quadratic Programming (QP) Tactical Asset Allocation (TAA) model, constrained by tactical bands and formulated from the AHP-SAA model. Then the Multivariate Copula model is adopted as the alternative TAA model to obtain the optimal portfolio with maximized returns at the lowest risk. The two resulting TAA models are found to be comparable as the expected portfolio returns of both models are close and within random error range. From the comparative risk analysis, the Markowitz QP model is found to capture more specific risk to market risk while the Multivariate Copula model is able to capture greater market risk to specific risk. Both models are equally effective alternatives and each is rigorous in constructing and estimating TAA portfolio that enable investors to engage in active management of the investment and to take advantage of arbitrage opportunities in the market.