Metal markets obviously vary greatly among each other and over time in terms of market structure, the substitutability and complementarity of other metals for various uses, the speed and specifics of technological change, and the political and resource-based constraints on the metal's availability. Models of many different types have been employed for various purposes related to analysis of metal markets. In a 1984 report, the United Nations Department of International Economic and Social Affairs categorizes these different modeling techniques as follows: Qualitative methods, cost and reserve-based methods, trend extrapolation, time-series methods, econometric market models, reduced-form methods, and global model systems. The report notes that each of these approaches has characteristics that make it more appropriate for some purposes but not others. For assessing feedbacks in metal markets and for policy analysis, econometric modeling is one of the best techniques. Some authors have also pointed out that composite models using econometric information supplemented with data from time series, futures markets, experts' opinions, and other information may be even more useful in commodity market modeling than any of these alone. A shortcoming of econometric models is that they cannot predict political events, which often have major effects on metal markets.

The tendency in the literature on econometric models of metal markets is toward ever more precision and detail in model specification. Since the early 1970s, many such models have been built privately, for use in predicting future price trends (or even for advising the GSA on how to manage the strategic stockpile). The most current of these models are not in the public domain, since their builders—consulting firms such as Wharton Econometrics and Charles River Associates—use them in their forecasting work for paying clients. Thus, the "state of the art" in terms of detailed econometric models of metal markets is no longer (if it ever was) in the academic realm.

There are, however, substantial doubts on the part of some experts about the ability of even quite detailed econometric models to provide satisfactory price explanations or predictions. Greater model size and complexity does not necessarily imply significantly greater forecasting ability. This may be due to the vulnerability of metals markets to a wide range of external influences, and the difficulties inherent in including accurate estimations of all of them in the models. These potential influences include technological breakthroughs, substitution effects between and among metals, recycling, environmental and resource use policies of governments, the behavior of publicly and privately held stocks and international buffer stocks, speculation, market cornering attempts, cartel formation, international trade and commodity agreements, strikes and political change in both producing and consuming areas, interest rates, and energy prices.

Moreover, as one recent study emphasizes, price formation in minerals markets is determined not only by market conditions (e.g., supply and demand factors), but also by market structure (i.e., the configuration of market actors and their relative power), and the implications of market structure for bargaining and the division of gains from trade. The price actually paid for metals often differs from published "producer prices" or prices on metal exchanges by a premium, which may vary for each individual transaction depending on transport costs, convenience, and other undefined factors. Accurate price modeling, especially in monopolistic or oligopolistic markets, can be quite difficult in the face of theory and data limitations. Humility and caution are called for in interpreting the results of modeling exercises.

Despite these recognized difficulties, hundreds of attempts have been made to model specific metal markets, mostly for the purpose of price forecasting but also to predict demand, analyze changes in market structure, and explore the process of technological change. These models vary widely in their details and specifications, and results are often fairly sensitive to how market characteristics, such as stock mechanisms, are modeled.

As evidenced by these models, econometric market modeling usually consists of modeling supply and demand and inventory behavior, along with their roles in determining price. Price, in conjunction with exogenous variables, affects the supply, demand, and stock variables, which in turn determine the equilibrium price and quantity. The different models are distinguished by the technological and institutional variables relevant to each particular industry, or by their emphasis on particular market forces or the behavior of specific decision makers.

Second, price has been found to be insignificant in the determination of both demand and supply for some metals (e.g., tin and zinc); also prices sometimes do not appear to clear the market, and/or two or more different prices exist simultaneously (e.g., tin, aluminum, zinc, and copper). Third, market modelers have noted difficulties with representing stocks/inventories and the widely differing production techniques which are used simultaneously in some industries (e.g., tin). The following sections discuss several existing metal market models in more detail, especially with regard to their representations of the effects of the strategic stockpile. Many of the published models of metal markets include the U.S. stockpile as a factor, in some capacity. Often the stockpile is employed as an instrumental variable in the price equation. Successive iterations to achieve a price equation with a good fit in these models generally show the stockpile as an important variable. The models are discussed below in alphabetical order by metal.

Patricia E. Perkins “World Metal Markets”