IMPROVING ALUMINUM PRICE PREDICTABILITY FROM FORECAST COMBINATIONS, MODEL CONFIDENCE SETS AND SARIMAX

João Bosco Castro, Alessandra Montini, Emerson Marçal

Resumo


Industrial metals led the commodity space higher in 2017, closing the year up 28%. Among industrial metals, aluminum is the second most used metal in the world after steel. As aluminum is a key input for a variety of industrial applications, large swings in prices can have a major impact on the terms of trade. Therefore, improve forecasts on aluminum prices is of critical importance to economic policy decision makers, producers, industry consumers and investors. This work proposes a novel approach to obtain an optimal combination for aluminum price forecasting by utilizing Forecast Combination, Model Confidence Set and Sarimax. Five individual models were estimated The results showed that the best model was the forecast combination including Sarimax and Sarima models. The work indicated that aluminum inventories and 3-month aluminum forward prices improved price forecast accuracy, as key covariates to improve policy decisions.

Key-words: Commodity prices; forecast combination; model confidence set

Texto completo: PDF