Cointegration and Detectable Linear and Nonlinear Causality: Analysis Using the London Metal Exchange Lead Contract

Abstract

This study applies linear and nonlinear Granger causality tests to examine the dynamic relation between London Metal Exchange (LME) cash prices and three possible predictors. The analysis uses matched quarterly inventory, UK Treasury bill interest rates, futures prices and cash prices for the commodity lead traded on the LME. The effects of cointegration on both linear and nonlinear Granger causality tests is also examined. When cointegration is not modelled, evidence is found of both linear and nonlinear causality between cash prices and analysed predictor variables. However, after controlling for cointegration, evidence of significant nonlinear causality is no longer found. These results contribute to the empirical literature on commodity price forecasting by highlighting the relationship between cointegration and detectable linear and nonlinear causality. The importance of interest rate and inventory as well as futures price in forecasting cash prices is also illustrated. Failure to detect significant nonlinearity after controlling for cointegration may also go some way to explaining the reason for the disappointing forecasting performances of many nonlinear models in the general finance literature. It may be that the variables are correct, but the functional form is overly complex and a standard VAR or VECM may often apply.