Dynamic metals demand model

Abstract

Studies by Nutting, Jacobson and Evans and Georgentalis et al. have all concluded, using panel data sets, that a number of metals appear to share a common demand curve that is stable over time. However, these studies have a number of theoretical and econometric limitations. This paper addresses these concerns and reassesses the hypothesis that some metals share such a common demand curve with the same price elasticity of demand. This is achieved within the framework of a random coefficients model. This model was applied to a dynamic metals demand function (DLR) and various estimation techniques were used including OLS, IV, Empirical Bayes (EB) and Instrumental Variables Empirical Bayes (IV-EB). It was found that each metal had its own individual short run demand function with statistically different own price and industrial activity elasticities of demand. In the long run, each metal appeared to be equally unresponsive to price changes, but had different industrial activity elasticities. The speeds of price adjustment to periods of market disequilibrium differed substantially between metals.