Recommend    |    Subscriber Services    |    Feedback    |     Subscribe Online
 
 
 
 
IUP Publications Online
Home About IUP Magazines Journals Books Archives
     
 
The IUP Journal of Financial Risk Management
Modeling Time-Varying Volatility in Indian Commodity Futures Return: Some Empirical Evidence
:
:
:
:
:
:
:
:
:
 
 
 
 
 
 
 

The aim of this paper is to introduce several volatility models and use these models to predict the conditional variance of the rate of return in Indian commodity future market. This paper chooses the Generalized Autoregressive Conditional Heteroscedasticity (GARCH), E-GARCH, GJRGARCH and APARCH models to analyze the rate of return and considers using three different distributions on error terms: normal distribution, Studentís t distribution and skewed t distribution. So this paper mainly captures the forecasting performance with volatility models under different error distributions. Finally, by using AIC, the best model is chosen to predict the conditional variance. Forecasting performance is checked by using Mean Square Error (MSE), Heteroskedasticity- Adjusted Squared Error (HASE), Logarithmic Error (LE) and Mincer Zarnowitz Regression. This paper selects three Generic 1st Future Contracts traded on MCX (Multi-Commodity Exchange of India): Aluminum, Copper and Zinc. It is concluded that long memory is an important characteristic of the Aluminum, Copper and Zinc futures volatility returns and should be considered when addressing investment decisions.

 
 
 

There is ample literature that examines volatility modeling and forecasting, as volatility plays a key role in asset pricing, portfolio management and risk management (see, for example, a comprehensive review by Poon and Granger, 2003). As volatility is a latent variable, it needs to be estimated, and the standard method in the literature is to use daily squared returns. However, it has become increasingly popular to use Realized Volatility (RV) which is constructed from intraday data as a proxy of volatility since the seminal paper by Andersen and Bollerslev (1998). More recently, with the availability of high-frequency data, a strand of literature considers the informational advantages of high frequency data and compares volatility forecasts obtained from intraday data with those using daily data (see Pong et al., 2004; Fuertes et al., 2009; and Chortareas et al., 2011 among others). Jiang et al. (2015) compare the volatility forecasts using daily data and intraday data for the Chinese commodity futures market and found little information advantage in generating volatility forecasts using intraday data. There is a lot of debate on superior volatility forecasting methods which was started with the work of Taylor (1986). Since then, many researchers have tried to find the best performing method for different financial markets and time horizons by twisting around versions of the famous ARCH model, but there is still no consensus in the literature on the most adequate volatility specification. For example, in the work of McMillan et al. (2000), no method was unanimously proposed, because volatility techniques have been examined under different frameworks, such as statistical loss functions, sampling schemes, time periods, and assets. However, all suggested methods share a common characteristic: they account for volatility asymmetry.

 
 
 

Financial Risk Management Journal, Generalized Autoregressive Conditional Heteroscedasticity (GARCH), E-GARCH, GJRGARCH, Modeling, Time-Varying , Volatility, Indian Commodity, Futures Return, Empirical Evidence.