THE UNBIASED PRICING HYPOTHESIS AND THE PREDICTIVE CONTENT OF BASE METAL MARKETS

This study tested the unbiased pricing hypothesis, the theory of storage, and the ability of past futures’ prices to forecast future changes in spot prices in the copper, aluminum, nickel and lead markets for the period October 2011 through May 2021. Wavelets and a time-varying parameter model with Bayesian priors were the primary tools. The risk neutral version of the unbiased hypothesis was found to be true for aluminum, copper, and lead for 3-month futures. Nickel 3-month futures were found to be unbiased predictors in the risk adverse sense. The 12-month futures of all metals supported the risk adverse version of the hypothesis. The tests of the theory of storage were inconclusive as many residuals were serially correlated. Where this was not the case, the theory of storage did not hold. 3-month rolling wavelet forecasts were typically superior to ARIMA and random walk models for 16-32 days. For 32-64 days, the wavelet and ARIMA forecasts were better than the random walk forecasts. For 1-year rolling forecasts, the wavelet and ARIMA models were superior to random walk models for 32-64 days