Economics Forecasting Metals Returns a Bayesian Decision

Turning points in commodity returns are important for decisions of policy makers, commodity producers and consumers reliant on medium term outcomes. However, forecasting of turning points has been a neglected feature of forecasting, especially in commodity markets. I forecast turning points in metals price returns using Bayesian Decision Theory. The method produces a probabilistic statement about our belief of a turning point occurring in the next period which, combined with a decision rule based on a loss function generates optimal turning point forecasts. This method produces positive results in forecasting turning points in metals returns, with the simple linear models investigated producing more accurate turning point forecasts than naive models across a number of different evaluation methods for the general case and for the specific example of a producing firm. 1 This paper is on my honours dissertation which was supported by the BHP Billiton Honours Scholarship. I would like to thank Kenneth Clements for his supervision and helpful comments. I would also like to thank Rachida Ouysse and other commentators attending the 2010 National Honours Colloquium, Edward Cripps and Arnold Zellner. This research was supported in part by the Australian Research Council and the UWA Business School.