Risk management and portfolio optimization for agricultural commodity futures returns: Multivariate heterogeneous autoregressive realized volatility (MHAR-RV) approach

Abstract

© Serials Publications Pvt.Ltd. The objectives of this paper are to construct the efficient frontier and optimum portfolio from the most commonly traded agricultural commodity futures, and to evaluate financial risk by Value at Risk. We evaluated alternative volatility forecasting and computed daily Value at Risk (VaR) based on Multivariate Realized Volatility (MRV) and Multivariate Heterogeneous Autoregressive (MHAR) approach. The intraday trade data of three agricultural commodity futures prices, namely corn, wheat and soybean traded in Chicago Board of Trade (CBOT) with three different frequencies that is 1 minute, 5 minutes and 15 minutes, were collected from Bloomberg database. The complete data set covered the period from November 2015 to December 2016. The empirical results showed that calculated realized volatility from Realized Covariance Measure (Andersen et. al. 2003) of corn, wheat and soybean futures returns have the long memory feature for every frequency based on R/S test and GPH test. The simulated returns from MHAR (Bauer and Vorkink, 2007) are applied to construct the efficient frontier and optimum portfolio. The optimum portfolio suggested to invest more than half in corn followed by soybean and wheat, respectively. The estimated VaR and ES of portfolio in period t+1 at 1%, 5% and 10% level are 8.98%, 6.33%, 4.92% and 10.46%, 7.95%, 6.75% respectively.