Oscillatory Versus Quadratic Trends in Natural Resource Commodity Prices

In this paper we introduce a model for the description of natural resources�� price paths, which in contrast to the existing literature, captures non-linear trends by means of a simple trigonometric function. We then use a set of model selection criteria to compare our trigonometric trend model with Slade��s (1982) quadratic trend model, as well as a more general one, that nests both of these models. We estimate the models using price series of the main fuel and metal resources prices, analyzed by the relevant literature, and ��nd that in most cases the trigonometric trend model is selected as the one better ��tting the data. Our results have implications for the long-run projection of natural resources prices and, consequently, for the relevance of the Hotelling rule.