DOI: 10.36347/sjebm.2014.v01i10.003

Pages: 442-461

Predicting commodity price dynamics is very important for speculators as well as investors. This paper analyzes the Indian commodity price as well as production data with a view to ascertain whether a linear or nonlinear approach is best suited to model their behaviour. Investigations are carried out to ascertain the stationarity of price and production data corresponding to four agricultural commodities, four metal commodities, and the Indian Gross Domestic Product (GDP). Standard and advanced unit root tests and variance ratio statistics are used in the analyses. Cointegration tests are carried out to ascertain whether the commodity production data are co-integrated with the Indian GDP. Further, parameters of the Lewis model, as formulated by Deaton and Laroque (2003) under the assumption of a linear commodity price process, are estimated using a full information maximum likelihood (FIML) technique to verify how this model fits the data in the Indian commodity market. Later, with the help of advanced unit root tests, it is shown that the commodity prices under study are better represented by a nonlinear process.