Portfolio Optimization of Global Reits Returns: High-Dimensional Copula-Based Approach

Abstract

Pacific, Europe, USA, and emerging markets with multivariate t copula based on GARCH model, and to measure portfolio risk with value at risk (VaR) and component VaR (CVaR). The 1,454 REIT price index return observations were collected from 1 Dec 2009 to 29 June 2015 and calculated based on a continuous compound basis. The empirical results showed that the estimated equations of USA, Europe and emerging REIT index returns were ARMA(2,2)-GARCH(1,1), while ASIA-Pacific was ARMA(3,3)-GARCH(1,1). The coefficients of t distribution of these equations were also statistically significant at 1%, meaning the assumption of t distribution for ARMA-GARCH estimation was reasonable. Then, the multivariate t copula was used to construct an optimized portfolio for high dimensional risk management. The Monte Carlo simulation was applied in order to construct the optimized portfolio by using the mean-CVaR model at the given significance level of 5% and to obtain the efficient frontier of the portfolio under different expected returns. Finally, the optimal weights of the portfolio were obtained with the various expected returns in frontier.