Multivariate forecasting of mineral commodities prices: Implications for natural resource scarcity.

Abstract

A comprehensive methodology is developed in this study to assess the scarcity of nonrenewable natural resources. The commodities examined in the study are: aluminum, copper, lead, zinc, tin, molybdenum, steel, oil, and coal. Multivariate state space and extensive learning model methods are used to project real prices of mineral resource products to the year 2011. The methodology is conceived in multivariate dimension to accommodate economic theory of price formation; namely price of a commodity is dependent on prices of complement and substitute products in equilibrium market conditions of supply and demand. Contrary to previous studies which examined each commodity price separately, this study employs simultaneous forecasting of economic variables theorized to be determinants of commodity prices. The multivariate state space technique uses annual data and thus focuses implicitly on disaggregated economic relations. The extensive system of learning models uses cumulative data, and thus facuses on major trends and patterns in economic relations. These techniques are used to assess economic scarcity of mineral resources by means of "holistic" forecasts of commodity price trends. On the basis of projected trends to the year 2011, the study found that all commodities prices are either declining or flat; thus indicating a general failure to support a claim of increasing scarcity of mineral resources. However, observation of the entire price series for some commodities reveals increasing trends in prices, and thus increasing economic scarcity. Conflicting scarcity signals are examined, interpreted, and qualified in light of economic realities of each commodity market, such as the possible influence of market structure and non-market forces interferences on price trends.