7.

4 Market Models and Forecasts

A mineral-market model describes the relationship between the attributes of a particular
market (such as production, consumption, prices, or inventories) and the determinants
of these attributes (such as resource endowments, market structure, government
policies, technologies of production and use, income, and many others).
Factors that are determinants in some models are described by other models; for
example, price is a determinant in many production models and yet is itself modeled
in models of price formation. Some models only partially describe the market under
study (such as a model of copper production in the United States), whereas others
more completely describe a market (such as a model of the world copper industry).
Some models are formal in the sense that they explicitly and quantitatively describe
the important relationships in a particular market, whereas other models are largely
informal, intuitive, or judgmental. This chapter focuses largely on formal models.
Models are used by a variety of people and organizations for a variety of purposes.
Private companies active in mineral development, for example, use models as
one of several tools for making decisions concerning mineral production, marketing,
pricing, exploration, and investments in new mines and processing facilities; often
these models are used to forecast or predict various factors that influence the
economic attractiveness of a project, including prices, capital and operating costs,
world production, and consumption. Government organizations active in mineral
development
use models for these same purposes. Government organizations - whether
or not active in mineral development - also use models for other purposes, such as assessing
the potential impact of different fiscal or concessionary regimes on investment,
and evaluating the likely impact of a supply disruption in an important producing
country on the availability of a particular mineral in the home country.
More generally, mineral-market models have at least three important purposes:
(1) analysis, to understand the relationships between the attributes of a market (for
example, the price of a mineral) and the important determinants of these attributes
(for example, factors of supply and demand, including raw material costs, technology,
income, government policies, and consumer tastes and preferences), (2) simu.lation,
to assess the potential market impact of a change in one of the underlying determinants,
and (3) forecasting, to predict future values of, for example, prices, production,
or consumption.
Given this mwtiplicity of purposes for mineral-market models, it is not surprising
that many different types of models exist. The next section reviews several of the important
types.
7.4.1 Types of Mineral-Market Models
Most mineral-market models belong to one of two families of model types (adapted
from Labys eta!' 1985). Some models incorporate characteristics of both model
types. The first family of mineral-market models is econometric models. This type of
model describes the relationships among supply, demand, prices, and inventories, as
well as the determinants of these market attributes (cf. Fig. 7.5). Prices adjust to
changes in supply, demand, and inventories, which in turn respond to the change in
prices. A typical econometric model consists of equations for supply, demand, inventories,
and prices, as shown in the illustrative example below:
Mineral Markets
Set) = s[P,G,GP,MS]
D(t) = d[P,PSC,Y]
let) = i[l(t-l) + Set) - D(t)]
pet) = p[l(t)/D(t)].
181
Four important attributes of a market are explained in terms of a number of
explanatory variables. The quantity of mineral supplied in any year t, Set), is a function
of four explanatory variables: the mineral's price (P), geologic factors (G), government
policies affecting mineral production (GP), and market structure (MS). The
rationale for including price in the model is that an increase (decrease) in price provides
an incentive for a producer to increase (decrease) production, other factors remaining
the same. Geologic factors importantly influence the availability of a mineral
resource for mining, and production is likely to rise or fall with changes in the
geologic favorability for mining. Government policies influence the level of mineral
production by encouraging or discouraging mining. Market structure also influences
the level of production; monopolistic markets tend to supply smaller quantities at
higher prices than competitive markets, whereas oligopolistic markets tend to have
quantities and prices intermediate between those of monopoly and competition (see
Sect. 7.1).
The quantity demanded by consumers depends, in this example, on three
explanatory variables: the mineral's own price (P), the prices of substitute and complement
materials (Psq, and income or industrial activity (Y). The quantity demanded
is expected to be inversely related to a mineral's price; the higher the price,
the less consumers will demand, other factors remaining the same. The relative price
of a substitute material reflects the attractiveness of substituting one material or minerai
for another in a particular use (such as aluminum for copper in electric-power
transmission). As the price of a substitute falls relative to the mineral in question,
consumers will demand smaller quantities. A complement is a good or service that is
used along with the mineral in question (for example, copper and zinc in brass). As
the price of a complement rises (falls), it is expected that consumers will consume less
(more) of the mineral in question (again, other factors remaining the same). Income,
sometimes measured as industrial activity, is also expected to importantly influence
the quantities demanded by consumers: as income or industrial activity rise and fall,
so, too, is the demand for minerals used in this industrial activity likely to rise or fall.
Inventories in any year are simply the inventories from the previous year, I(t-l), plus
current supply minus current demand. Finally, the system is closed by relating prices
to inventories and demand. The relative importance of each explanatory variable in
each equation, or in other words the sensitivity of the attribute being modeled to
changes in particular explanatory variables, is determined by regression analysis
using historical data. All four equations then are solved simultaneously to determine
market-equilibrium values for each attribute.
As noted earlier, this example is merely illustrative. Not all models include
exactly the same equations or variables, and some models are much more complicated,
while others are simpler. The wide variety of mineral-market econometric
models should not be surprising. First of all, mineral markets differ in a number of
important respects, including the extent of competition among producers, the nature
of government policies influencing costs of production, the relative importance of the
various sources of supply (main product, by-product and co-product, and scrap), and

the nature of end uses. Second, there are different models for different purposes; for
instance, a model designed for policy analysis will not necessarily be appropriate for
a more purely academic study striving to understand the past.
There are many examples of econometric models of mineral markets, illustrating
the differences among models (although it is beyond the scope of this study to compare
and contrast models). There are numerous models for the copper market, for instance.
Fisher etal. (1972) is the seminal work in copper. Their model, estimated for
the period 1948-1968, divides the world into two separate, but linked, markets. The
first market is the United States, where administered or producer prices dominated
during the period under study. The second market is the rest of the world, governed
by free-market prices based largely on the prices of the London Metal Exchange.
Seven equations describe copper supply: five equations for primary production (in
four important producing countries, and the rest of the world) and two equations for
secondary - or scrap - production (the United States and the rest of the world). Four
equations describe copper demand in the United States, Europe, Japan, and the rest
of the world. Although an important purpose of the model was to assess the impact
of potential increases in Chilean output on copper prices and Chilean revenues, the
model has perhaps been more influential as a standard against which to compare subsequent
models of the copper industry. Charles River Associates (1978), in one subsequent
model, incorporate exploration and discovery of new reserves into a copper
model. Lasaga (1981) examines the role of the copper industry in the Chilean
economy. Takeuchi etal. (1987) use an econometric model to forecast future consumption,
mine capacity and production, production costs, and prices. For additional
copper models, as well as models for other mineral markets, see Labys etal. (1985),
Mikesell (1979), and the references cited therein.
The second important family of mineral-market models is engineering models.
This class of model includes a wider variety of models than the econometric family
and thus can only be defined loosely. What distinguishes engineering models is their
mathematical description of the production process largely in technical, rather than
in more purely economic or behavioral, terms.
Engineering models vary significantly in the extent to which they incorporate the
economic determinants of the attributes of a market. The general form of this type of
model is illustrated in Figure 7.6. National economic activity determines the demand
for final products, which in turn determines the derived demands for inputs of raw
materials (including minerals), energy, and labor. An important strength of an engineering
model is its technical description of how inputs are transformed into outputs
of final products. The description typically is dis aggregated into several transformations
intermediate between the initial use of raw materials and the production of
final goods.
Most engineering models are designed either to determine the optimal combination
of inputs given an initial set of constraints (technical, economic, or other) or to
simulate what would occur under a given set of assumptions. One specific purpose of
optimization techniques is to determine the best location for a new plant or processing
facility given what is known about the costs, technologies, and location of existing
and expected future plants or processing facilities. Another purpose is to choose
the optimal combination of inputs for an assumed mix of final products. Important
examples of optimization techniques include linear programming, process optimizaMineral
Markets and spatial optimization. Simulation or system techniques, the second broad
class of engineering models and including system-simulation and system-dynamic
models, can also be used for a variety of purposes, including: to estimate the effects
of changes in costs, government policies, or technologies on supply and demand; to
assess the impact of mineral demands on national or regional economic activity; and
to forecast a variety of factors such as future production, consumption, and the geographic
distribution of production and consumption.
Engineering models typically are best suited for analyzing the important
technological and to a lesser extent policy determinants of mineral supply and demand.
Because engineering models emphasize the technologies of mineral production
and use, they are often useful for long-term forecasts, which depend critically on
technological change. But engineering models tend to be less useful for predicting the
short and medium term, periods generally short enough so that major unanticipated
changes in technolgy do not occur. (Over the short and medium term, mineral markets
tend to be more strongly influenced by changes in the level of industrial activity
and other largely economic factors.) Similarly, engineering models tend to account
better for material substitution than econometric models. Engineering models can
also capture the impact of policy changes to the extent that they properly describe the
effects of policy changes on costs.
Econometric models, on the other hand, tend to be strong in exactly those areas
where engineering models are weak, namely in accounting for the important
economic determinants of mineral supply and demand. This type of model is valuable
184 Mineral Economics
for assessing the responsiveness of production and consumption to changes in
economic variables such as price and income, a concept known to economists as elasticity;
for instance, a mining company may want to know whether a given percentage
change in price will cause demand for its product to change by a greater, lesser, or the
same percentage. Econometric models are often used for short and medium term
forecasts, periods in which technological change is more likely to be smaller and more
gradual than in the longer term. But because econometric models are based largely
on historical data reflecting old and existing technologies, they are not able to capture
significant changes in production and consumption technologies. Similarly,
econometric models are less able to account for significant changes in government
policies and other institutional factors that importantly influence mineral markets.
Given these complementary strengths and weaknesses of econometric and engineering
models, a number of researchers have developed hybrid models that attempt to
take advantage of the strengths of each type of model.
An important lesson from this brief review of mineral-market models is that there
are different types of models for different purposes. The appropriateness of any particular
type of model depends on the purpose of the modeling exercise. There is no
all-purpose type of model. Finally, a word of caution is in order: a model is only as
good as the input data and, perhaps more importantly, the underlying rationale or
theory behind the model. Bad data fed into a theoretically sound model and good
data used to test a model based on bad economic or engineering thinking both yield
bad results.
7.4.2 Forecasting
An important use of mineral-market models is forecasting. Although many forecasts
are derived from formal or quantitative models similar to those described above,
others are based on less formal and more qualitative methods. Regardless of the
method, forecasting is indispensable for any individual or organization that must
make decisions today whose correctness will be determined by future events. For
mining companies considering an investment in a new mine or processing facility, or
banks considering a loan to the project, estimates of future profits hinge critically on
estimates of future demand, prices, and production costs. For government organizations,
the impacts of stockpiling, land-use, environmental, and other policies need to
be considered with some sort of forecast.
Most forecasts belong to one or - more frequently - a combination of three
families of forecasting techniques (Tilton 1983). The first family includes statistical
methods, which essentially extrapolate from past trends of prices, production, consumption,
or other market attributes. These methods range from simply projecting a
straight-line trend of past data into the future to sophisticated statistical methods of
extrapolation (such as the Box-Jenkins technique). Compared to other forecasting
methods, statistical methods tend to be simple and inexpensive, and require little
data. They are often good enough for short-term forecasts. However, these methods
implicitly assume that there are no significant changes in the fundamental determinants
of price, demand, supply, or whatever factor is being predicted. In other words,
these methods by their very nature will not foresee major turning points in the market. For
example, many mines and processing facilities were built in the middle 1970s
largely on the basis of projecting historical trends in metal consumption (compare, for
instance, actual copper consumption since 1973 with a simple projection of earlier
trends in Figure 7.7).
The second type of forecasting method is causal or behavioral. These methods
identify cause and effect, and predict the future on this basis. Important examples include
the econometric and engineering models described previously, as well as several
others. Causal methods can better predict the future, especially major turning
points, if they correctly identify cause and effect. However, they are usually more
time consuming and expensive and tend to require larger amounts of data, compared
to statistical methods.
The third type of forecasting method is judgmental. In contrast with the first two
types of methods, judgmental methods are not explicit in their analysis of either cause
and effect or of historical data. The methods are largely qualitative, usually based on
the informed judgments of experts, who presumably have some sort of implicit causeand-
effect model in their minds. If the experts have correctly identified the causeand-
effect relationships, they may be better able to foresee major changes in mineral
markets than statistical and causal methods of forecasting. Judgmental methods can
adjust current trends for anticipated changes in policies and technologies, changes
which in many cases are likely to be unstatable in an explicit or formal forecasting
model. Judgmental methods suffer, however, from their lack of quantitative rigor.
The general problems of forecasting can be appreciated by considering two types
of forecasts: short-term forecasts of this year and perhaps the next several years, and
long-term forecasts of the next several decades. Over the short term the major influence
on mineral markets is the general state of the economy, which in turn determines
the demand for final goods and services and thus the derived demands for minerals,
as well as mineral prices and production. Thus, short-term projections require
186 Mineral Economics
a good macroeconomic model. Over the longer term, changes in technology - including
material substitution - are at least as important as overall economic acitivity as
driving forces for change in mineral markets. Technological change and material substitution
are both exceedingly hard to predict. Thus, an important message is that
forecasting is inherently difficult. Nevertheless, those who make decisions concerning
mineral development and policy have no choice but to use forecasts - whether formal
or informal, quantitative or qualitative - in making their decisions.