In the medium term, factors that shock commodity markets can also be of a political or cataclysmic nature, but they tend to be more related to national economic conditions or to market forces themselves. Fluctuations in market forces tend to be observed in the demand and supply conditions and underlying market equilibrium. Fluctuations in national economic conditions, commonly observed in the form of business cycles, can cause changes in industrial production and hence mineral demands or in interest rates and ultimately commodity investments. Variations in weather conditions induce changes in agricultural supply and hence in product prices. The formal analyses of the impact of these kinds of shocks has appeared in modeling studies, such as for example Labys (1973,1998) and Rausser and Hochman (1979).

In the short term, market shocks come primarily from financial factors, particularly those related to speculation and hedging on commodity futures, options and other derivative markets. The resulting price behavior has been termed random, because it reflects the flow of randomly appearing information. In fact the price behavior can be identified more specifically as being stochastic or following a nonlinear dynamic or stochastic process (e.g., autoregressive conditional heteroscedastic), or even a chaotic process. It can also be related more specifically to financial shocks such as in interest rates or exchange rates. A substantial literature exists attempting to explain this short-term behavior, examples of which include Barkoulas et al. (1997,1999), DeCoster et al. (1992), Holt and Aradhyula ( 1990), Hudson et al. ( 1987), and Teysseire et al. ( 1997).

In summary the described price fluctuations, which vary frequently and extensively, have made market modeling and forecasting an extremely difficult task. Linear and nonlinear time series methods which have been applied to stock prices and financial markets have been employed in commodity price analysis. Business cycle, econometric and time series methods that have performed effectively in macroeconometric forecasting have also been applied in commodity market forecasting, i.e. see Diebold ( 1998). Nonetheless, commodity market forecasters have become frustrated in their search for improvements in the performance of these methods. My intention is thus not only to trace improvements in present research but also to suggest certain new developments at the frontier of applications.

The recent analysis of commodity markets has been largely occupied with the explanation of the temporal or time series behavior of prices. Most of this work has dealt with tests and models that have investigated linear and nonlinear price fluctuations based on mean reversion and variance measures. The data of interest have included spot and futures prices ranging from intraday ticks to annual averages. Tests applied to these price series include chaos, random walk, fractional integration, stationarity, hetereogenity, etc. Models constructed range from ARMA and ARFIMA to GARCH and Neural Networks. Examples are Barkoulas et al. ( 1997,1999), Cheung and Lai (1993), Cromwell et al. ( 2000), Dam ( 1998), Deaton and Laroque (1992), DeCoster et al. ( 1992) Kyrtsou et al. (2002), and Teyssiere et al. (1997).

The most basic type of commodity model from which econometric and modeling methodologies have developed is the competitive market model. Such a model initially neglects market imperfections and assumes that commodity demand and supply interact to produce an equilibrium price reflecting competitive market conditions. Such a model may consist of a number of combined regression equations, each explaining separately, a single market or sector variable. Market models or the equivalent industry models are applicable to all agricultural, mineral or energy production and use categories. Their greatest utility is in providing a consistent framework for planning industrial expansion, forecasting market price movements, and studying the effects of regulatory policies.

There is no doubt that commodity modeling evolved from agricultural economics. Discussions of the structural modeling of agricultural markets appear in Brookfield (1991) or Labys ( 1973). The development of econometric models suitable for analyzing mineral markets possessing competitive behavior has evolved more recently. One of the first was the Desai ( 1966) tin model which explained tin price fluctuations on a world basis. The copper market has also been subject to several modeling efforts. Most notable, Fisher, Cootner and Bailey ( 1972) built a world copper model which was recognized as one of the first major econometric mineral modeling efforts. CRA (1978) later extended the supply sector of the Fisher, Cootner and Bailey copper model to include long run adjustments in exploration and discovery as well as subsequent mining capacity formation. This long run adjustment process was combined with a short-run inventory adjustment process in a distinctly disequilibrium form of copper model by Labys (1980a). Such an approach to modeling the copper market was suggested by Richard (1977) with his continuous time, differential equation approach. More recent developments on the structural modeling and forecasting of mineral and energy markets appear in Labys (1999 b).

Applications to energy markets have not been as extensive because of the difficulties of dealing with regulatory and non-competitive influences on market behavior. Verleger ( 1993), however, has shown that it can be applied to explain disruptive shortages. His model links econometric equations for oil spot prices, consumer demand, and supply shortage conditions. MacAvoy and Pindyck (1975) have built an econometric model of the natural gas industry which has been used extensively to analyze the effect on the industry of federal regulation of the wellhead price of gas and of permissible rates of return for the pipeline industry. Labys, et al. ( 1979) have modeled the U.S. coal market using this approach to forecast future levels of coal demand, supply, prices and inventories. Most recently, Trieu et al. (1994) have reported their modeling of the world spot uranium market.

The several econometric approaches taken to model noncompetitive market configurations are essentially similar. For example, the monopoly case involves one dominant (monopolist) producer and many (perfectly competitive) consumers. The single producer thus maximizes his own profits given the aggregate demand function for the commodity of interest and the supply response of the other firms in the industry. Examples of applications to OPEC and the crude oil market include Pindyck (1978 a, b) who developed a model to determine optimal price and quantity paths that would result from cartel behavior on the part of producer’s organizations in the copper and bauxite markets.

The analysis of commodity prices independent of other market variables had been popular long before structural modeling made its appearance. This form of price analysis which essentially relates to a single economic sector lends itself well to reduced form or nonstructural equation methods. Recall that most any structural commodity model consisting usually of a multi-equation market equilibrium formulation can be reduced to single equations with endogenous variables appearing on the left-hand side and only exogenous variables on the right-hand side. This form of model can be specified based on econometric regression equations or on econometric time series equations. The latter can be univariate in which a single variable is explained in terms of its past statistical history or multivariate in which the past statistical history of several variables is combined. A number of variations exist on the univariate and multivariate themes. Over time several of these have evolved as being useful for analyzing and forecasting commodity markets.

The explanation and forecasting of commodity prices using univariate and multivariate methods depends on whether the researcher is interested in long run as compared to medium run or short run price behavior. The modeling of long run behavior involves basic linear or nonlinear trend models as shown earlier. The explanation of medium run behavior can involve models capable of generating some form of price cycles; examples include ARIMA, ARFIMA, exponential smoothing, VAR or structural time series (STS) models. The deciphering of short run behavior often is concerned with stochastic or random processes, such as those associated with the discovery of futures price movements. Empirical applications of these methods to commodity prices is extensive and can only be briefly reviewed here. A first direction has been to discover the time series generating processes underlying price behavior using advanced methods such as spectral analysis, e.g. Labys and Granger (1970). Today more emphasis has been placed on discovering the order or fractional order of integration using special testing procedures, e.g. see Barkoulas et al. (1997,1999).

Another constructive direction that commodity price analysis has followed concerns commodity price fluctuations in the form of waves and cycles. Greater attention has been directed concerning macroeconomic influences in the form of national and international business cycles that come to bear on commodity markets. The stimulus for this research has come from several directions. First, macroeconomists have developed better approaches for defining and explaining “real” business cycles. Some of this work has dealt with trend-breaks in an attempt to measure structural breaks in major cyclical upturns and downturns. Second, cyclical identification techniques have improved. For example, spectral analysis of the periodicity of price cycles has reemerged in the form of wavelet analysis as discussed in Davidson et al. (1997). Third, “structural” or “nonobservable component” models were advanced by Harvey (1985,1989) to directly model the amplitude and frequency of cycles in a dynamic and stochastic context employed senusoid and Kalman filter analysis. Examples of application to commodity prices appear in Labys et al. (1998,1999) and Kouassi et al. (1999). Analysis of the impacts of business cycles on commodity markets can be found in works of Cashin et al. (1999 a), Cristini (1999), Fama and French (1988), Labys et al. (1999), Moore (1988), Davutyan and Roberts (1991), and Saadi (1997).

In an attempt to summarize this and other related research of this period, I would emphasize that a transition occurred from structural models that emphasized demand, supply, and other price determinants towards nonstructural models that explain only the price variable itself. Some of this transition resulted from advances in the nonstructural modeling of financial markets. Other influences included advances in the econometric tests and models that can be employed for this purpose. Not to be forgotten is that the costs of performing commodity market research can be greatly reduced by only analyzing price variables. In retrospect, this overemphasis on univariate time series analysis has left us with a weak heritage, as we attempt to improve commodity market analysis in the near future.

The challenges presented to commodity price analysts and forecasters have been discussed and are reflected in the respective structural and nonstructural forecasting methods which have been reviewed. Among the more difficult challenges is the considerable uncertainty which pervades the markets : speculative runs, exogenous shocks, political interventions, and structural changes. This uncertainty is also related to endogenous instability such as that caused by price inelasticities, excessive market speculation, and links to business cycles. Also important are supply-side changes such as those due to biotechnology or genetic adjustments. Influences on the demand side include the competitive development of synthetic and engineered materials. On a global scale, market destabilizing factors range from multinational market power and government controls to international economic and financial transmission and feedback effects. Below some new possibilities that could augment commodity market analysis are presented.

We are just beginning to explore the influences of globalization on commodity markets. Here globalization is referred to in an economic context that ranges from financial market integration to labor markets and employment. A good example of this globalization is that gold can now be traded practically 24 hours per day as spot and futures exchanges open and close from New York through London and Sydney until Hong Kong, Tokyo, and again New York. The importance of such phenomena has been preliminarily explored by Gilbert (2002) for the case of cocoa and by Labys (2000 a) for the case of crude oil. Other implications for commodity markets to be investigated include shifts in trade towards increased transport of processed materials, reduction in pollution from the redeployment of energy commodities, and greater attention to geopolitics with respect to crude oil and refined products.

A second important domain of research concerns how commodity market and price cycles are interrelated with national and international business cycles. The important economic and econometric results achieved by macroeconomists dealing with business cycle analysis should be applied and advanced in the case of commodity markets. Issues of how inflationary adjustments, monetary and exchange rate policies in the developed economies are linked to commodity market fluctuations should be more carefully explored, including simultaneous interreactions. Longer term analysis should be performed relative to periodic waves, trend-breaks, and regime-switching. Shorter term analysis should be conducted relative to the impacts of financial market shocks, economic uncertainty, and geopolitical disruptions. Researchers need more stock-flow knowledge of longer term investment and capacity adjustment cycles and of shorter term inventory and stocking cycles. Examples of studies to be advanced in this area include Badillo et al. ( 1999) or Perron (1989).

Many commodity price decisions are made when outcomes are uncertain. This uncertainty evolves from the market structure of the industry, the international instability of commodity markets, environmental problems, changes in technology and a host of other factors. Uncertainty has only recently been recognized as a feature in policy analysis and a complexity for commodity policy modeling. Important advances that deserve attention in policy analysis and modeling recognize choice under uncertainty. For mineral market forecasting of this nature, see Torries (1998). The correspondence between decision making at individual and market levels and the relatively strong assumptions on the utility functions required also are important.

The theory underlying a commodity price model will suggest the functional form for the structural equations. The typical relationships are estimated either in linear form or in a form that is linear in the parameters, for example, loglinear or some other transformation. While these forms generally conform to a theoretically formulated model, the functional form only broadly approximates the theoretical specification. This reflects the difficulty and cost of non-linear estimates when short sample periods are involved. While simplified functional forms, even linear approximations, may be suitable for many simulations covering a “normal” period, the difficulty is that the abnormal or extreme periods are precisely the points when the non-linearities of the model should become effective.

Short-term price analysis is more related to hedging, speculation and trading on commodity related futures and derivative markets. Research in this area has been stimulated by the time series and econometric methods that have been developed for analyzing and forecasting financial market and price movements, i.e., see Diebold ( 1998). Some of this work concerns random walks, chaos and fractile behavior (DeCoster et al., 1992; Frank and Stengos, 1989). Other works dwell on the possibility that ARFIMA models might better be applied to forecasting when long memory is present in prices by using a fractional rather than an integer value for the degree of integration (Barkoulas et al., 1997,1999; Cheung and Lai, 1993). While this form of model emphasizes nonlinearity in price means, modeling based on nonlinearity in price variance or volatility has continued in the form of the variety of autoregressive hetereoscedastic models e.g., Kouassi et al., 1999).

There is also the need to untangle more carefully the nature of commodity market structures and the role played therein by different forms of market interventions such as regulated prices, subsidies, taxes and trade controls. The potential of multidisciplinary analysis in this area might be limited, unless the disciplines of quantitative political analysis and of industrial organization advance in this direction. Attention to market structure in mineral and energy models has been expanded by Pindyck (1978 a, b), Kolstad ( 1982), Lord (1991), and others. We have thus seen the embodied structure of the petroleum market advance from simple monopoly to rather complex forms of oligopoly. Greater attention to commodity structure appears in Labys (1980b) and Lord (1991).

A lesser amount of work has taken place regarding the role of market interventions and disruptions. Commodity models are thus likely to improve in the direction of offering a more realistic picture of these two related aspects of market structure.

Another domain of research deals with developing country linkages. Still with us are problems of commodity export quantity and revenue stabilization stemming from international commodity price instability. Some research has commenced as to how to deal better with commodity price derivatives and financial instruments in the form of commodity risk management (Privolos and Duncan, 1991 and Claessens and Duncan, 1993). Better understanding is needed of commodity investment formation, supply control and disruption, and the impact of import fluctuations on developed economy markets.

A most important research frontier concerns the behavior of the core of the commodity market mechanism. Here I refer to the recursive and simultaneous adjustments of commodity markets. Important market variables in this respect besides prices are production, consumption, inventories, imports and exports.

The shocks that markets receive include not only business cycles on the demand side and climate cycles on the supply side, but also a variety of random shocks. This would suggest new investigations of models that are dynamic, nonlinear, and stochastic in nature. Recent examples include Granger and Teravista (1993), Labys (1999 b), Labys and Yang ( 1991), and Lord (1991). The structure of such models should include imperfect competition, government regulation, and special characteristics. The core of such models should reflect cyclical and disequilibrium adjustments, such as those occurring when demand does not equal supply and imports exceed exports. Model applications should be improved in the direction of employing newer approaches to market and policy simulation, and model forecasting should embody new probabilistic approaches for evaluating risk impacts. A basis for further work on forecasting can be found in Chen and Besseler (1990), Gerlow et al. (1993), and Goodwin (1994).

Finally, it should not be neglected that all of this research requires improved sources of data. Practical commodity data research has been neglected to the point where no current estimates are available of commodity demand and supply elasticities for most commodities. The stock or inventory data found today are not as complete as that compiled twenty years ago. Typical but intransigent data problems need to be conquered, such as short-length series, missing observation, measurement error and stochastic behavior. Not the least of problems is decreased public and private expenditures on basic data collection and publication.

A more lofty but equally important goal is to pursue research funding and to establish research institutions that would finally provide a continuity to the progress of commodity market research. We still are suffering from the inability of the reconstructionists to fulfill Keynes’ plan. For 50 years, economists have witnessed a continuity in financial research via the International Monetary Fund, in economic development research via the World Bank, but no commodity research via “COMMOD.” Greater available research funding must be directed to this need if we are ever to develop a more profound understanding of these extremely complicated and diverse commodity markets.

***