In this paper we show that the exchange rates of some commodity exporter countries have the ability to predict the price of spot and future contracts of aluminum. This is shown with both in-sample and out-of-sample analyses. The theoretical underpinning of these results relies on the present-value model for exchange rate determination and on the tight connection between some commodity prices and the currencies of some commodity exporter countries. We show results using traditional statistical metrics of forecast accuracy: Mean Squared Prediction Error and Mean Directional Accuracy. We also explore different ways in which we can jointly take advantage of the predictive information contained in all of our commodity-currencies. While LASSO and a model equipped with the first principal component of our currencies perform well, the best combination strategies involve the pre-selection of the two best performing individual currencies: The Chilean Peso and the Icelandic Krona.
- Exchange rates
- Out-of-sample comparison
- Univariate time-series models