Commodity prices and related equity prices

This paper shows that commodity-sensitive stock price indices have strong power in predicting nominal and real commodity prices at short horizons (one-month-ahead predictions) using both in-and out-of-sample tests. The forecasts based on commodity-sensitive stock price indices are able to significantly outperform naive no-change forecasts. For example, the one-month-ahead forecasts for nominal commodity prices reduce the mean squared prediction error by between 1.5% (for natural gas prices) and 20% (for copper prices). Moreover, the one-month-ahead directional forecast is found to perform significantly better than a 50:50 coin toss. As stock prices are not subject to revision, the proposed variable, which reflects timely and readily available market information, can potentially be a valuable predictor and thereby help to improve the accuracy of commodity price forecasts.