Modeling Agricultural Commodity Price Volatility using GARCH Model with Structural Break

Abstract

A time series data has the usual assumption of constant mean and variance over time i.e., stationary. In real situation this assumption seems to be violated often. Mostly, the series dealing with financial aspects such as prices of various commodities hardly holds the assumption of constant variance. Moreover, few series shows an interesting behaviour of stationarity for some time, then suddenly the variability of the error term changes, it stays constant again for some time at this new value, until another change occurs (Ang and Bekaert, 2002). If in the analysis of a series, this factor is not accounted for, then we may land up having poor results (Clements and Hendry, 1998). The change in variance of the series at certain time epochs must be identified whether be a single or multiple. The very first work in this direction was started by Hsu et al. (1974), who used this method as an alternative to the Pareto distribution to model stock returns.