ABSTRACT
Much research has been recently directed to the forecasting of commodity prices using time series methods in response to the increasingly complex problems involved in commodity purchasing, selling and trading activities. In this chapter we consider the forecast performance of the structural time series (STS) approach employed to forecast short to medium term cyclical upturns and downturns in primary commodity prices (Labys and Kouassi, 2004). This gives specific acknowledgement of recent findings that commodity prices can display cyclical behavior, though obviously of a complex, stochastic nature. See for example the results in Chapters 4 and 7. Our approach is to evaluate two alternative STS models to assess their potential for explaining and forecasting these price movements. Further details on employing time series models to forecast cyclical movements can be found in Diebold (1998), Granger and Newbold (1986), Harvey and Jaeger (1993) and Mills (2003). Other useful books on time series forecasting include Bomhoff (1996), Clements and Hendry (1998), Franses (1998), and Taylor (1986).
