Essays on commodity dynamics, volatility, and interaction
Date of Issue2017-03-24
College of Humanities, Arts, and Social Sciences
This thesis consisting of three self-contained essays devotes to the volatility behavior of agricultural commodity, price transmission between energy price and grain price and the extreme returns interaction and response among commodities. The first essay investigates the persistence and permanent-transitory structure in the volatility of agricultural commodity continuous future contract returns with sample period from Jan 3rd, 1994 to Feb 18th, 2015. Lo’s Modified R/S Test and GPH Test confirm the existence of volatility persistence in the whole sample period with both standard GARCH and component GARCH models with normal distribution, student-t distribution, generalized error distribution and skewed generalized error distribution. Component GARCH model finds strong evidence on the permanent-transitory structure. Out of sample forecasting with rolling estimation is performed together with Value at Risk backtesting. Out of sample forecasting superiority of component GARCH model is confirmed by both Diebold-Mariano test besides traditional loss functions and MZ regression. Risk management application also appears to favor component GARCH model. These findings suggest the necessity of modeling long memory in volatility of agricultural commodity for volatility forecast purpose and risk management purpose for policy makers and investor. The second essay examines the asymmetry in price transmission between global grain and energy indices from January of 2007 to June of 2015. It applies nonlinear cointegration methods and its error correction model proposed by Sun (2011) together with the traditional linear methods. All methods confirm the existence with co-integration with high statistical significance. Nonlinear cointegration method also finds evidence on asymmetry and threshold effect in the price transmission. The existing literature has mixed finding regarding the co-integration between grain and energy prices and this paper contributes by suggesting the failure to establish cointegration may lie in the model misspecification in ignoring asymmetry and threshold effect. The third essay focuses on the extreme returns and responses in the commodity future market from an information flow perspective. It aims to examine whether there is heterogeneity in the interaction between extreme returns and those with the full samples. It applies modified event studies with bootstrapping to track the behavior of commodity future returns of Brent oil, corn, ethanol, natural gas, sugar and WTI oil around the extremely good and bad days defined by the returns of these commodities from Jan 1st, 1994 to Dec 31st, 2015. It also compares the result with those from generalized impulse response functions (IRFs). It finds evidence suggesting that the tail behaviors are distinct from those in the whole return spectrum and even among those tail behavior there is heterogeneity of many kinds.