Price Adjustment in Long Term Coal Contracts- MIT (1987)
Massachusetts Institute of Technology March, 1987
Published on: Mar 4, 2016
Transcripts - Price Adjustment in Long Term Coal Contracts- MIT (1987)
Digitized by the Internet Archive
2011 with funding from
Boston Library Consortium
PRICE ADJUSTMENT IN LONG TERM CONTRACTS:
THE CASE OF COAL
Paul L. Joskow
Number 44 4
Revised March, 1987
50 memorial drive
Cambridge, mass. 02139
PRICE ADJUSTMENT IN LONG TERM CONTRACTS:
THE CASE OF COAL
Paul L. Joskow
Revised March, 1987
Price Adjustment In Long Term Contracts: The Case of Coal
Paul L. Joskow
A sample of coal contracts between electric utilities and coal supplier!- is
used to analyze mechanisms for determining prices in long term coal contracts
Alternative methods for determining prices in long term contracts are
discussed and the actual adjustment mechanisms specified in a set of actual
coal contracts presented. The vast majority of long term coal contracts use a
base price plus escalation or cost-plus adjustment formula. Base price
equations and subsequent transactions price equations are estimated. The
analysis shows that on average long term contracts are flexible in the sense
that prices adjust to major changes in the costs of supplying coal. However,
some pricing rigidities are found which appear to reflect the economic
conditions prevailing at the time the contracts were executed. Furthermore,
some contracts track changes in market values very poorly.
Revised March 1987
Revised Draft (March 1987)
Price Adjustment in Long Term Contracts: The Case of Coal
Paul L. Joskow-'-
In two previous papers^
analyzed empirically the role of relationship
specific investments in the choice of contracts of various durations or
vertical integration to govern transactions between electric utilities and
coal suppliers. In this paper
examine how the parties to a long term coal
supply contract provide ex ante for adjustments in transactions prices over
utilize a sample of about 250 coal contracts to analyze the structure
of formal price adjustment provisions in coal contracts,
to determine the
factors that affect initial negotiate contract prices, and to examine actual
transaction price behavior over time.
There has been a lot of recent theoretical work and some related empirical
work that has focused on the benefits of long term contracts when relationship
specific investments are important.-^ There has also been some theoretical work
that focuses on the contractual arrangements that may be chosen to mitigate
opportunism problems and guard against certain types ex post inefficiencies in
am aware of very little systematic empirical work that
has examined how the parties to a long term contract provide ex ante for
adjustments in the terms of trade as market conditions change through the life
of the contractual relationship.-' Furthermore, despite the importance of
assumptions about price rigidity in macroeconomics (and to a lesser extent in
microeconomics) and the association of price rigidity with explicit or implicit
long term contracts,
there has been very little systematic empirical analysis of
the rigidity of actual transactions prices.
This paper seeks to expand our
empirical knowledge in these areas using data for coal contracts.
Coal supply arrangements are interesting for examining how long term
contracts provide for and actually work in adapting to changing market
conditions for several reasons. Electrio utilities routinely enter into very
long term coal supply relationships via contract. Contracts with specified
durations of 20 years or more are frequently utilized. Creating price
adjustment provisions in such contracts that do a good job adapting to
changing market conditions
while simultaneously preserving other benefits of
long terms contracts would appear to be a formidable task. Yet coal supply
relationships are rarely terminated prematurely
and utilities have continued
to rely on long term contracts as market conditions have changed.
reasonable to hypothesize that some way has been found to keep the costs of
long term contracts low relative to their benefits.
The paper proceeds in the following way.
The next section discusses
alternative methods for price adjustment in long term coal contracts when such
contracts have been chosen to ameliorate contracting problems that may emerge
when relationship specific investments are important. Section three describes
the structure of price adjustment provisions found in a sample of long term
Section four presents an empirical analysis of the factors
that determine initial coal contract prices at the time contracts are
negotiated and adjustments in these prices over time for a sample of about 250
coal contracts. The final section contains some concluding remarks.
Pricing Considerations In Long Term Coal Contracts
A consummation of a coal supply contract between an electric utility and
coal supplier involves agreement on numerous contractual terms and conditions.
Among the most important are the price, the quantity, the quality
characteristics of the coal, the duration of the contractual commitment and
provisions for adjusting one or more of these contractual provisions over
have argued elsewhere^ that both the duration of coal contracts and
the decision to internalize coal production through vertical integration are
heavily influenced by the importance of relationship specific investments of
the types described by Williamson.
The importance of relationship specific
investments in coal supply relationships varies widely. As a result, there is
also wide variation in the negotiated duration of coal supply contracts and
the incidence of vertical integration. Historically, roughly 15% of coal
purchased by electric utilities involves spot market transactions. Another 15%
involves integrated supply. The rest involves coal supply contracts with
contractually specified durations that vary between one year and fifty years.
Let us focus on a long term coal supply contract that involves deliveries
of prespecified quantities of coal over a period of several years.
for a supplier to agree to provide supplies,
the present discounted value of
expected future revenues must be greater than or equal to the present
discounted value of expected future costs, including the opportunity costs of
any future sales foregone by the seller by commiting to a long term supply
agreement. If coal markets are competitive, as they appear to be, buyers on
average will pay no more than the
present discounted value of expected future
production costs (including rents and opportunity costs)
We can think of this expected present value price per unit supplied as
being composed of several components. These include the present value of
expected future operating cost (labor, material and supplies), capital costs,
and any expected economic rents that may accrue to infra-marginal coal leases
at each point in time over the life of the contract. 10 So we expect the
payment provisions in a coal contract to satisfy:
L + M + K + R
P = the
present value of expected future revenues per unit of coal supplied
L = the present value of expected future labor costs per unit of
M — the present value of expected future materials and supplies costs
per unit of coal supplied.
K - the present value of the expected cost of capital invested per
unit of coal supplied.
R = the present value of expected economic rents or opportunity costs per unit
of coal supplied which clears the market and accrues to the owners of
mining rights to inframarginal coal deposits (R > 0)
It is reasonably straightforward to relate coal production associated with
a specific contract to the "variable" costs of labor, materials and supplies,
and perhaps economic rents. It is more difficult to relate a specific contract
to costs associated with capital investments required to provide supplies
since the capital investments may have an economic life that is longer than
the term of a specific contract. This is further complicated when the supplier
makes relationship specific investments.
Let us assume that the buyer and
seller deal with the the relationship specific investment problem when they
establish the duration of the contractual commitment and associated notice and
y e can then think of their being a time stream of
expected annual rental costs for capital equal to the opportunity cost of
investment funds plus economic depreciation which is independent of observed
differences in contract duration.
examine whether or not initial contract
prices are independent of the duration of the contract negotiated at the time
it is executed in the empirical work reported below.
There are many different payment profiles that could satisfy equation (1)
In principle, there exists a fixed price contract that satisfies (1)
reflecting current and future costs and market price expectations at the time
the contract is negotiated.
If coal markets are competitive a "market price"
contract in which the buyer and seller simply agreed to base future payments
on the "market price" for comparable coal supplies at the time of delivery
could be relied upon to satisfy equation
as well. A contract that
specifies a base price to reflect current production costs, the current annual
rental cost of capital investments, and economic rents combined with
escalation provisions to reflect changes in production and opportunity costs
over time could also be designed to satisfy (1)
Numerous mixtures of fixed
price plus escalating price formulas could be designed to to satisfy (1) as
We can narrow down the likely structure of mutually satisfactory price
adjustment provisions in long term contracts by recognizing that the parties
are likely to want to structure price adjustment provisions to achieve certain
objectives in addition to satisfying equation (1). These include (a) a desire
to guard against "opportunism",
"hold-up" or haggling problems associated with
the presence of relationship specific investments,
a desire to minimize
the incentives the contractual provisions themselves give the parties to
breach their contractual promises,
a desire to provide enough flexibility
to facilitate efficient adaptations to changing market conditions and,
since the price the buyer (a regulated electric utility) can charge customers
for the final product is in this case regulated by state and federal
regulatory agencies, to avoid pricing provisions that might lead a regulatory
agency to disallow a fraction of the coal costs as being "imprudent"
Fixed Price Contracts
In a world in which nominal production costs are expected to increase over
time, a fixed price contract that reflects ex ante expectations of future cost
increases is likely to have bad properties from all of these perspectives.
A long term fixed price contract necessarily "front loads" the revenues and
expected profits of the seller relative to the flow of costs when nominal
costs are expected to increase over time. A fixed price that satisfies (1)
will involve an initial price that is high relative to current spot prices,
high relative to prices in older fixed price contracts, and high relative to
current production costs. If actual cost changes equal the ex ante expected
changes in costs, at some point later in the term of the contract the price
will be below then current spot market prices and below the prices in new
fixed price contracts. If the expected rate of cost increase is fast enough
and the contract long enough, the fixed price could fall below then current
costs of providing incremental supplies at some future date.
In either case,
the seller will have strong incentives to breach on quantity or quality
promises both because he can sell his supplies elsewhere at a higher price and
possibly because the additional direct costs of meeting his commitments may be
greater than the revenues he will receive. If production costs and market
prices rise more quickly than anticipated the seller will face even stronger
incentives to breach.- -^ While the buyer can always appeal to the courts to
enforce the contract or to award damages this route is costly and the results
While almost any price adjustment provision could lead to large
disparities between contract prices and "market prices" if certain
contingencies arise and thereby provide incentives for either the buyer or the
seller to breach, a fixed price contract almost guarentees that these problems
will arise even if there is no uncertainty about how costs and market values
will change over time."
Of particular relevance to regulated electric utilities is the fact that
an agreement to a fixed price contract which involves initial prices that are
far above current "market" prices is not likely to be treated very favorably
by their regulators especially since fuel costs are generally passed through
directly to customers through fuel adjustment mechanisms.
For all of these reasons it is unlikely that an electric utility buyer and
coal supplier would find a fixed price long term contract mutually
satisfactory in a world where nominal costs and prices are expected to rise
over time or where there is a possibility that uncertain events will occur
that will lead to cost or price increases or decreases that are significant
different from expected values
Market Price Contracts
A potentially attractive alternative to a fixed price contract would be a
"market price" contract. Such a contract would involve the parties simply
agreeing that prices will be adjusted to reflect changes in the "market price"
of coal with identical quality attributes available from (approximately) the
same location as the coal that has been contract for.^° If coal markets are
competitive we would expect that the expected PDV of future market prices
A market price provision eliminates any incentives the
buyer or seller may have to breach the agreement as a result of better
alternative opportunities during the term of the agreement. The pricing
provision is in theory easy to state and easy to enforce by the courts and
therefore is potentially attractive to guard against ex post opportunistic
behavior. If the supplier produces efficiently it is unlikely that the price
will fall below his incremental cost of meeting his contractual commitments
and if it does he could meet them by buying rather than producing (and this
would be efficient). This price adjustment mechanism would be potentially very
attractive for electric utilities concerned about prudence reviews based on
comparisons between contract prices and some measure of the "market price."
The primary problem with a "market price" contract is defining an
appropriate market price norm to use for this purpose. Coal is not a
homogeneous commodity. There are wide variations in heat content, sulfur
content, ash content, moisture content and chemical composition all of which
affect the value of coal to buyers and market prices. Because transportation
costs are an important component of the delivered price of coal, prices at the
mine also vary from area to area to reflect proximity to coal consumers and
the costs of transportation.
While the government reports the average FOB
mine price per ton by producing district (with a considerable lag)
no breakdown between spot and contract prices FOB the mine
and only limited
information that would make it possible to accurately adjust for differences
in coal quality or variations in the "quality" of the coal supply
relationship. Even within producing districts that have reasonably homogeneous
coal deposits there is a very wide variation in FOB mine prices observed at
any point in time. In short,
there does not appear to be a single simple
number that a contract can rely on as a good indicator of the relevant market
price of coal.
The use of a market price norm in long term contracts where relationship
specific investments are important is further complicated by the way damages
are likely to be assessed if there is a breach of contract. The essence of a
contractual relationship involving relationship specific investments is that
there is an expected cost to one or both parties of premature termination that
associated with the investments that they have sunk in anticipation of
performance. Ideally, the damages that would be assessed if either party-
breached the agreement should reflect these "relationship specific" costs. But
if the parties simply agree that they will trade at the "market price," a
breach may lead to damages that bear no necessary relationship to the true
costs of a breach. If the seller breaches the buyer would be expected to
replace the coal at "the market price" which is also what the buyer agreed to
pay the seller, so arguably there may be no damages to the buyer. Similarly,
if the buyer breaches,
the seller would be presumed to dispose of the supplies
at "the market price," which is arguably what he would have been paid anyway. 21
Aside from costs associated with lost sales or the use of more costly
substitute generating capacity due to lags in finding a new buyer or seller
this could lead,
from the breach.
If some other method is used to specify the transactions
to the a conclusion that there were no damages
price in the contract that does not depend directly on observed market prices,
however, the difference between the contract price and what the buyer must pay
to replace the supplies or the seller receives when he finds other buyers
becomes a natural basis for comparison for purposes of damage computation.
All things considered, therefore,
would not expect the parties to rely
overtly on a simple "market price" provision for determining transactions
prices in long term coal contracts even if the parties desire to establish a
pricing provision that at least roughly tracks changes in market values over
time. When such a provision is included,
would expect that it would specify
exactly how the relevant "market price" is to be determined, in order to avoid
haggling over the proper benchmark and to serve as a basis for damage
Escalating Price Contracts
But for the problems associated with coming up with an appropriate norm
and the problems associated with protecting relationship specific investments
from premature breach or termination, a simple market pricing mechanism has
the attractive property that it minimizes incentives to breach that arise
because of differences between contract prices and market values. It would be
desirable to devise a pricing formula that satisfies equation (1) and (a)
provides for a clear specification of the agreed upon contract price that does
not depend directly on "market prices",
which, over the term of the
contract, would come reasonably close to tracking the market value of the
coal, and (c) which provides a basis for properly assessing damages associated
with the loss of relationship specific investments.
Over the long term we expect that coal prices will change as the costs of
production (including rents and opportunity costs) rise and fall, other things
It is natural therefore to think of starting with a base price that
reflects current supply and demand conditions and then allowing it to vary
with changes in the costs of producing coal. This might be accomplished by
structuring a contract that establishes an initial or base price equal to the
seller's current productions costs plus an economic rent component and then
provides for prices to change along with the seller's actual costs of
production (e.g. some type of cost plus contract). There are at least three
potential problems with a cost plus contract, however. First, it has bad
incentive features from the perspective of inducing minimum cost supply by the
seller. Second, even if a particular supplier makes his best efforts to supply
efficiently, a specific mine may turn out to be significantly higher cost or
lower cost than the typical mine and contract prices may consistently be above
or below the market value of the coal. Third,
it will not be sensitive to
unanticipated changes in market supply and demand conditions that would affect
market values more or less than changes in the supplier's costs of
production. ^2 In each case, costly haggling and renegotiation problems
resulting from large differences between contract prices and prevailing market
values may make this contracts unattractive.
An alternative to a cost plus contract
contract that specifies a
base price reflecting supply and demand conditions when the contract is signed
and which then provides for adjustments in the base price using a formula that
incorporates a weighted average of exogeneous input price indices reflecting
the anticipated input/output mix of the supplier, combined with exogenous
indices of changes in labor and capital productivity reflecting "general"
changes in production opportunities. Instead of using the actual costs
incurred by a specific seller to adjust prices, exogenous indices reflecting
general market opportunities are at least partially used. These contracts are
generally called "base price plus escalation" (BPE) contracts.
As long as the market value of the coal moves along with changes in input
prices, general productivity changes affecting comparable mines, etc. this
approach seems superior to a cost plus contract. It helps to solve the first
incentive problem associated with pure cost plus contracts, since prices are
at least partially decoupled from the actual costs incurred by a specific
it solves the second problem associated with a mine that
it unusually costly or unusually efficient.
It does not solve the third
problem, however. Indeed, there is no obvious way to solve the third problem
without tying contract prices to market values directly in some way. With
either a cost plus contract or a BPE contract it is impossible to track large
short run changes in the market value of coal associated with demand side
shocks. Short run supply side shocks would be more easily captured, but
certainly not perfectly. If one of these adjustment mechanisms is chosen, the
parties would have to recognize that serious haggling problems may emerge if
markets are subject to unanticipated demand or supply side shocks which lead
to large increases or decreases in the expected market value of coal over the
term of the contract.
Clearly, none of the price adjustment alternatives is ideal. Long term
(nominal) fixed price contracts are simply not credible in a market like this
would not expect them to be used extensively ^3 Market price contracts
are attractive but both the difficulty of defining an appropriate market price
norm and the problem of providing for damage penalties that appropriately
reflect the quasi rents associated with the difference in value of
relationship specific investments between the intended use and alternative
uses implies that they will not be relied on very much in coal markets except
in special circumstances
Both BPE and cost plus contracts have attractive
features in that both could do a reasonably good job of tracking market values
in the long run as long as changes in market values move closely with changes
in the average cost of production. Large unanticipated demand or supply side
shocks could lead to problems, however.
Typical Pricing Provisions In Long Term Coal Contracts
have general descriptive information on the method used to adjust prices
for 158 of the contracts" in the data base discussed in the Appendix.
in the course of my research on coal contracts
have had the
opportunity to review over 80 actual coal contracts (plus amendments) in more
detail. 26 Table
breaks down these 158 contracts by method of price
adjustment. The vast majority of the contracts use a BPE adjustment formula.
About 15% of the contracts are listed as cost-plus (often with an incentive
fee provision indicated) and about 7% of the contracts provide for negotiated
prices or prices tied to some market basket. Six of the eleven contracts in
the latter category have durations of five years or less.
Price Adjustment Mechanisms Found in 158 Long Term Coal Contracts
In Force in 1979
Base Price Plus Escalation:
Negotiated or market price:
The typical long term coal supply contract thus uses a base price plus
escalation (BPE) pricing formula in which an initial base price is set when
the contract is negotiated and then adjustments are made to the base price
using a weighted average of changes in external input price and productivity
indices and changes in actual costs. For example,
contract might specify of
base price of $30 per ton (with coal quality attributes specified elswhere in
the contract) and then break down the base price into several different
Materials & Supplies:
The contract then includes provisions for adjusting each of these
components for changes in input prices or productivity based on either an
exogenous index or the supplier's actual cost experience. For example, the
labor component might be adjusted for changes in actual labor costs or changes
in union wage settlements or sometimes an index of local area mining wages
Especially when a wage index is used the contract is likely to include a
manning table which fixes a base manhours per ton (adjusted for wage
differences among different types of workers) figure. The latter is normally
subject to further adjustment for changes in work rules and labor productivity
mandated by changes in union wage agreements or government regulations
The materials and supplies component (explosives and electricity are
often broken out as separate components) is typically indexed to a weighted
average of several sub-indexes of the Producer Price Index (e.g. tractor
tires, rubber belts and belting, construction machinery and equipment,
lubricating oil, fuel oil, etc.)- The explosives component is generally
adjusted for changes in the wholesale price index for explosives. Finally, the
electricity component would be adjusted for changes in the applicable retail
rate charged by the local electric utility.
There is substantial variation among contracts in how much the "other"
category is broken down. Sometimes separate "profit",
royalty components are specified. More typically there is just a residual
component called "other." This component is generally either partially or
fully indexed to changes in a general price index like the wholesale price
index or the consumer price index.
The contracts that
have reviewed almost always provide for additional
adjustments to reflect the costs of meeting new government regulations,
changes in property and excise taxes, changes in third-party royalty payments,
etc. As inflation was recognized as being a more severe problem in the mid and
late 1970 's it also appears that scheduled price adjustments have became more
frequent. Before the early 1970'
annual or quarterly adjustments were
typical. Quarterly adjustments and then monthly adjustments appear to have
been have been used more after 1974. Indeed, the speed of adjustment has been
one item that has sometimes been renegotiated by the parties. Furthermore,
appears that as inflation became more important, base prices have been broken
down into finer and finer components, more sub -indexes have been incorporated
in the adjustments to the M&S component and the "other" category is much more
likely to be fully rather than partially indexed.
Finally, many contracts include reopener provisions based on "gross
inequity" to the buyer or the seller and force majeur provisions to protect
the buyer and seller from certain unanticipated events. These provisions are
most likely to be applicable when changes in production costs get far out of
line with changes in adjusted prices, when government regulations are imposed
that have significant effects on the costs of mining and were not anticipated
by either party when the contract was executed, or when government regulations
on emmissions constrain the ability of the buyer to make use of the quantities
of coal that have been contracted for. Deviation of contract prices from
"prevailing market prices" alone is not generally an excuse for invoking
either type of adjustment clause, however.
It thus appears that prices in long term coal supply contracts are
"flexible" in the sense that prices can change over time in response to
changes in the costs of production. However, they appear to be "inflexible" in
the sense that they do not respond to changes in market values that are not
reflected in changes in the average costs of production as provided for by BPE
or cost plus adjustment formulas. There is no reason to believe,
that long term contract prices will move up or down in lockstep with spot
The Behavior Of Coal Prices For Coal Sold Under Long Term Contracts
In this section
examine empirically the actual behavior of coal prices
for coal sold under long term contracts.
am particularly interested in
examining the factors that determine the initial or base prices that the
contracting parties agree to and how these prices have adjusted over time. The
approach is very simple. As discussed in more detail in the Appendix,
constructed a data base that contains information for nearly 300 coal
contracts negotiated between the late 1950'
and 1979. 28
For most of these
contracts the data base includes the initial (nominal) base price FOB the mine
(per million BTU) specified in the contract as well as for the actual average
price FOB the mine for coal supplied pursuant to these contracts in 1979, 1980
have information on two important quality attributes of the
its btu content and its sulfur content,
information on where the coal is
mined, information on when the contract was executed, and information on the
negotiated contract duration specified at the time the contract was executed.
estimate an equation to explain the observed base prices for
coal. This equation allows us to identify the factors that determine the
initial negotiated base prices in coal contracts
prices for new contracts signed during a particular period with the average
price prevailing for all coal transactions in a particular year, and to see
how base prices have varied over time with changing market conditions. Next,
estimate a set of equations to explain actual transaction prices resulting
from these contracts in three different years subsequent to the formation of
the contract and the specification of a base price. These transactions prices
presumably reflect the workings of the price adjustment provisions discussed
above plus mutually beneficial renegotiation of contract terms. These
equations make it possible to compare transactions prices at a particular
point in time pursuant to contracts negotiated at different points in time.
Before proceeding with the statistical analysis it is useful to discuss
the general patterns of coal prices and changing market conditions over the
period for which we have base price and transactions price data (1960
provides information reported by the Department of Energy (DOE) for
the average price of coal at the mine (per million BTU's) sold in the U.S. by
year between 1960 and 1981. The prices reflect both contract and spot market
transactions. The first column gives the nominal prices and the second column
gives the real prices ($1972) using the GNP deflator. The third column is an
index of average labor productivity at coal mines. Nominal prices are fairly
Average Transactions Price For Coal
1960 - 1981
Source: Statistical Abstract of the United States
constant between 1960 and 1965 and real prices fall slightly. Both nominal and
real prices then begin to rise. Real prices are about 20% higher in the 19711973 period than they were in the mid-1960'
Real prices take a big jump in
1974 and 1975 and then decline slightly. Nominal prices rise throughout the
The general pattern of average price changes that we observe is not
terribly surprising. The period covered in Table
is one of
change in coal markets. Coal price movements reflect a combination of supply
side changes affecting the costs of producing coal and demand side changes.
After a period of declining or stagnant coal production lasting into the early
1960's, coal production increases gradually but significantly between 1965 and
1973 (about 20%).
Especially by 1970, contemporary discussions of the market
indicate that supply was tight. After 1973, coal production increases more
rapidly, and by 1981 production is about 35% higher than in 1973. The
increases in coal production reflect primarily increased utilization of coal
by electric utilities.
Large increases in oil prices in 1974 and 1975 made coal a much more
attractive fuel for generating electricity and almost certainly increased the
expected demand for coal in the long run significantly. The short run effects
on coal consumption by electric utilities appear to have been modest,
however. 29 Large increases in oil prices in 1978 and 1979 had a larger effect
on short run coal demand. The large price increase accelerated conversions of
plants from oil to coal in the East and led to increases in purchased power
transactions ("coal by wire") between the midwest and the east and between the
Southwest and the Pacific regions.™
At the same time, long run projections
of coal demand began to be reduced as it was becoming clear that electricity
demand would grow more slowly than anticipated and
utilities began to slow
down capacity additions of all kinds. By this time oil and gas prices had
risen far enough that no new oil or gas fired generating capacity was
anticipated in any event. *!
The primary effect of the oil price increase in
1978-89 was probably to increase the short run demand for coal at a time when
long run projections of coal demand were declining and a great deal of new
production capacity was becoming available.
Federal mine safety regulation began to affect the costs of supplying coal
by about 1970.-^3 Additional safety and health regulations affecting mining
were introduced during the 1970'
and are also thought to have increased
costs. 4 Federal and state environmental regulation passed during the 1970 's
affecting land reclamation procedures and the disposal of mine wastes are also
likely to have increased costs during the 1970'
Coal mine productivity
declines from about 1969 to 1977 and then begins to increase. Federal and
state environmental regulations affecting power plant emissions shifted the
demand for coal toward low- sulfur and ash coal which in turn greatly increased
the demand for western coal. After 1973,
the coal industry was affected by the
high rate of inflation in input prices as was the rest of the economy.
The most significant change in the real prices occured between 1973 and
1975 after the first oil price shock. Both demand side and cost side shocks
are consistent with the behavior. On the cost side,
average labor productivity
(See Table 3) declined by nearly 20% between 1973 and 1975
this occurs between 1974 and 1975)
(essentially all of
Real wages for coal miners increased by
about 10% during this two year period as well. The real prices for equipment
and materials and supplies used in coal mining (construction equipment, fuel
electricity, explosives) also increased dramatically during these two
years.-"3 A significant portion of this increase would probably have been
picked up by many BPE adjustment formulas and reflected in prices under
existing contracts, especially those adjustment formulas that had labor
Base Price Equations
Clearly, any effort to analyze base prices and subsequent transactions
prices must take account of the changing market conditions that characterized
the demand and supply of coal during this period. Let me turn first to a
discussion of the base prices in BPE and cost-plus type contracts.
Base prices FOB the mine should vary directly with the quality of the
coal (Btu and sulfur content)
the contract was negotiated.
the supply region that it comes from and when
expect that the higher the Btu content of the
coal the higher will be the price, other things equal. High Btu coal is more
valuable than low Btu coal for several reasons. First, the thermal efficiency
of electric generating plants varies directly with the Btu content of the
Second, the cost of building a generating unit with a specific design
temperature and pressure varies inversely with the Btu content of the coal it
is designed to burn.
the cost of transporting coal per Btu of useable
energy must be lower for high Btu coal than low Btu coal.
also expect that high sulfur coal will be more valuable than low sulfur
after approximately 1970,
as a consequence of changes in sulfur
emissions regulations applied to electric utilities. The FOB mine prices for
coal should be significantly lower for coal produced in the Western region,
after accounting for differences in Btu and sulfur content, to reflect the
much higher average transportation distance to utilities compared to coal
supplied from mines in the Midwest and East. While
allow for price
differences between Midwestern and Eastern supply regions as well in the
empirical work reported below, these differences should be much smaller
because there is little difference in average transport distance.
Since we are looking at nominal base prices there should be differences in
negotiated base prices over time as well. These differences should reflect
both changes in nominal input prices and real changes in supply and demand
conditions. In what follows
generally group the contracts into four time
periods. The first period covers contracts negotiated before 1971. The second
covers contracts negotiated between 1971 and 1973, after federal sulfur
emissions restrictions were announced and the first major new federal mine
safety legislation began to take effect, but before oil prices increased
dramatically and before inflationary concerns became severe. The third period
is 1974 through 1977,
after the initial large increases in world oil prices
and more stringent environmental and health and safety regulations were
imposed on the coal mining industry and when concerns about rapid inflation in
input prices probably got more attention in coal contracting. The fourth
period is 1978-1979 when oil prices rose again, the rate of inflation
increased, coal markets began to "cool down" somewhat do to declining
projected rates of growth of electricity demand and increases in coal
The contracts in the data base vary widely in the duration for which the
parties agreed to abide by the terms and conditions of the contract when it
contrary to my expectation, coal contracts have a
significant fixed price component which "front loads" future expected costs,
then those contracts with longer terms should have higher base prices
things equal. Similarly,
if the buyer or the seller tried to use the level of
the base price to as a sort of financial "hostage" ,^® rather than relying on
variations in the term of the agreement to amortize relationship specific
investments, base prices could vary systematically with the negotiated term of
the agreement. Accounting for variations in the agreed upon duration of a coal
supply agreement at the time it was signed in the base price equations allows
us to determine whether or not these effects are present.
begin by estimating the following linear base price relationship which
measures regional and time related effects relative to Eastern coal contracts
signed in the 1978-1979 period. 41 Variable definitions and summary statistics
for each variable are reported in Table
BASE PRICE - a + b]Tl+ b 2 T2+ b 3 T3 + b 4 MIDWEST + b 5 WEST + b 6 BTU
+ b 7 SULF*Tl + b 8 SULF*T2 + b 9 SULF*T2 + b 10 SULF*T4
+ b 11 DURATION+ v
v = iid error term
If base prices generally reflect prevailing supply and demand conditions
at the time a contract is signed
would expect to observe the following
pattern of coefficient estimates. The coefficients of Tl
T2 and T3 should all
be negative and decline monotonically in absolute value since nominal coal
prices generally rise throughout this entire period. The coefficient of WEST
be negative reflecting the relatively long average transport distances
for Western coal. The coefficient of BTU should be positive reflecting the
higher value of high BTU coal. The coefficient of SULF should be negative to
reflect the relatively lower values associated with high sulfur coal due to
environmental regulations. The coefficient of DURATION should be zero unless
base prices include a significant fixed price or financial hostage components.
if we use the estimated relationship to "predict" base prices for new
contracts in each period the resulting values should track changes in
prevailing market conditions fairly closely, although prevailing market
conditions will be difficult to measure precisely.
Variable Definitions and Summary Statistics
Initial Base Price
specified in each
price in 1979
Price in 1980
Price in 1981
Average Btu content
per pound of coal
Average % sulfur
content of coal
Duration at contract
Dummy Variable that
equals 1 if coal is
supplied from midwestern region
Dummy Variable that
equals 1 if coal is
supplied from western
Dummy variable equals
1 for contracts signed
Dummy variable for
contracts signed 1971-73
Dummy variable for
contracts signed 1974-77
Dummy variable for
Conracts signed 1078-79
All prices are in cents per million Btu's of coal.
OLS estimates of the coefficients of (2) may be biased because of the
nature of the sample of coal contracts that
make use of. As discussed in
sam pie is based on contracts in force in 1979, rather than all
contracts written over time. We observe contract prices only if contract
duration (y) is greater than or equal to (1979
have a censored sample.
execution date, so we
where t^ is the contract
This implies that the random
error (v) in the base price equation (2) may be correlated with the random
error (u) in a second contract duration equation^ as a consequence of the
sampling procedure. If this is the case, the random error (v) in the price
equation will be a function of the independent variables in such a contract
duration equation. OLS estimates of the coefficients of the independent
variables in the price equation
would then be biased if they are
correlated with the independent variables in the duration equation. In
particular, independent variables that appear in the contract duration
equation may appear to be significant when introduced as independent variables
in equation (2) when in fact they are not 4 ^
appear in the price equation
Since three variables that
also appear in the duration equation that
estimated in my earlier paper this is a potential problem here.
Very simply, the problem with OLS estimation of
potential presence of
is that it ignores the
"missing variable" which is equal to the mean of v
conditional on the sample selection rule and the specification of a contract
duration equation. The conditional mean of v given the sample selection rule
and assuming that both equations are linear and the joint density of u and v
bivariate normal is given by:
yi > (1979
ti )) =
s uv Hi
L ± = (1979
the covariance of u and v, y^ is the duration of contract
was negotiated, g((.) is the standard normal density function
and G(.) is the standard normal distribution function.
We can obtain consistent estimates of the coefficients of (2) by making
use of information obtained from the ML estimates of the duration equations
estimated in Joskow (1987) to generate
consistent estimate of H for each
observation, adding the estimated values for this variable to (2) and then
running OLS to the augmented model. The coefficient of H is then
estimate of the covariance of u and v.
Table 4 report OLS estimates of the coefficients of equation (2) in column
and estimates of equation (2) with H included in column (2)
the coefficients of equation (2) with the coefficient of SULF constrained to
be equal across periods are reported in column (3)
Finally, estimates of
equation (2) without DURATION or H are reported in column
Note first that the coefficient of H is insignificant and the coefficient
estimates are not sensitive to its inclusion, so that there does not appear to
be a serious censoring problem associated with the data used to estimate the
base price equations. Second, constraining the coefficient of SULF to be
identical across periods does not change the estimated coefficients in
important ways. It is quite clear, however, from an examination of the
estimated dummy variables in this case that nominal base prices have risen
significantly over time.
will return to this in a moment. In any event, we
can reject the hypothesis that the effects of sulfur on coal prices is the
same in the pre-1971 period as it is in the later periods at the 5% level
(F(l,229) - 4.25)
Base Price Equations
(standard errors in Parentheses)
Dependent Variable: BASE PRICE
Focusing on the first column of Table
we find that the estimated
coefficient pattern is just what we expected. High quality (BTU) coal has
higher price per million btu's than does low quality coal. There is a
significant penalty associated with sulfur content (i.e. a negative
coefficient on SULF)
after 1970, but not earlier. Coal supplied from the
Western producing region has a significantly (numerically and statistically)
lower price than coal produced elsewhere. There is no significant difference
in prices for coal supplied from mines located in the East and Midwest. The
estimated coefficient of DURATION
and is very small.
insignificantly different from zero
The fact that DURATION has no effect on base prices is consistent with the
view that base prices generally reflect prevailing market conditions and do
not embody significant fixed price components reflecting the front loading of
future expected nominal cost increases. To explore this phenomenon futher
can use the estimated equation reported in Table 4 to generate base price
predictions for contracts signed in different time periods. Base price
predictions using column (4) of Table 4 are reported in Table 5.^' The first
column of Table
provides such estimates using the sample means for the
independent variables in equation (2)
The second column is the average price
for all coal transactions reported by the DOE during each time period based on
the figures in the first column of Table
The figures in this column
included both contract and spot market purchases.
It is clear from Table
that the predicted average base price in new
contracts tracks contemporaneous average transactions prices fairly closely.
To the extent that the average transactions price for all coal transactions is
a good indicator of "prevailing market conditions" this result suggests both
that base prices are established to reflect prevailing market conditions and
Predicted Base Prices And Average Transactions Prices
(cents per million Btu, FOB Mine)
Predicted Base Price
Tr ansaction Price
Source: Predicted base prices are calculated from Table 4 column (4) using the
sample means of the independent variables. Actual average transactions prices
for all coal sold during each period (contract + spot) calculated from Table 2.
that contractual price adjustment provisions combined with mutually-
satisfactory renegotiation, do a reasonably good job in tracking changes in
Subsequent Transactions Price Equations
We can get better insights into price rigidity by more carefully
analyzing contemporaneous transactions prices associated with contracts
negotiated in different coal market "eras."
To do so
turn next to an
analysis of the actual transactions prices that emerged in three different
years subsequent to the time the contracts were executed and an associated
base price and adjustment mechanism agreed to using information on actual
transactions prices by contract for the subsequent years 1979, 1980 and 1981
(see Table 2)
estimate exactly the same relationship as (2) above,
without DURATION, 50 using actual transactions prices for 1979, 1980 and 1981
rather than base prices.
report separate estimates for the transaction price
relationship using transaction price data for each of these three years and
also pool the three years of transaction price data. In the latter case, dummy
variables for transactions recorded for 1979 and 1980 are included (PRICE79
and PRICE80) as well to account for adjustments in nominal prices over the
three year period.
If the formal BPE contracts are meaningful in the sense that the initial
base price and the associated adjustment provisions specified in the contract
do affect future transactions prices then we should observe some specific
patterns of price rigidity. In particular, it is likely that we will find
significant differences in transactions prices between pre-1974 contracts and
later contracts. As
indicated above, the structure of the typical price
adjustment provisions of long term coal contracts appear to provide, in one
way or another, for price adjustments that reflect changes in the average
costs of producing coal. They do not appear to provide for adjustments that
reflect changes in the average economic rent that a supplier might earn due to
unanticipated changes in demand expectations or supply side changes that
increase the costs of producing from marginal properties more than the average
cost of production. These are exactly the kinds of changes that are likely to
have occurred after 1973.
After 1973, the expected demand for coal increased substantially and it is
reasonable to hypothesize that the present discounted value of future economic
rents increased on average. Zimmerman's estimates suggest that the effects of
long run increases in future demand on prices in 1980 should be fairly small
since the long run supply function for coal is quite elastic.
demand shocks may have increased prices in the short run much more, however.
Furthermore, cost increases due to input price changes, productivity changes
and environmental and safety regulations increased more rapidly as time went
While contractual provisions are generally in place to protect the
supplier from these changes, they are likely to be imperfect, and are probably
more imperfect for pre- 1974 contracts than for later contracts. The post- 1977
contracts are likely to be the best protected of all. The pre-1970 contracts
are less likely to have anticipated the possibility of major increases in
costs due to environmental, health and safety regulations and the general
deterioration in labor productivity. The pre-1974 contracts are less likely to
have anticipated the very high rates of input price inflation of the mid and
and probably provided for less detailed and less frequent
adjustment of base prices to cost changes. To the extent that these cost
increases affected the costs of producing from new properties more than from
existing mines, there would be an associated potential economic rent
attributable to older contracts that would not actually appear in higher
prices due to contractual rigidities 52 These considerations make it likely
that "rigid" pre-1974 contracts yielded transactions prices during the 1979-81
period that were lower than contemporaneous transactions prices for post- 1974
contracts. Differences between the contracts negotiated from 1974-1977 and
those negotiated in 1978 and 1979 are likely to be distinguished primarily by
the frequency of adjustment and perhaps the use more detailed adjustment
formulas, but should probably have fairly comparable prices.
provides estimates of transactions price equations for 1979, 1980
and 1981 transactions along with an estimate of a transactions price equation
using the pooled data and allowing for different intercepts for 1979, 1980 and
Estimates with (OLS/H) and without H (OLS) are reported. 53 in SO me cases
the H variables is now significant and has a systematic effect on predicted
transactions prices reported below.
Turning to Table
we see that the basic structure of the transactions
price equation estimates is very similar to the structure of the base price
equation estimates. There is a premium for high quality (Btu content) coal and
a penalty for high sulfur coal associated with post- 1970 contracts.
penalty for contracts written between 1971 and 1973 is smaller and less
significant than in the base price equations, however. As we shall see,
contracts negotiated during this period of time
yield transactions prices
significantly below those for post- 1973 contracts and the sulfur penalty is
probably getting picked up in part by the time dummy for this period. Coal
from the Western producing region carries a significantly lower FOB mine price
than coal supplied from mines in the East and Mid-west. Thus, price adjustment
provisions appear to preserve in later transactions prices the relative market
value weights that are important for determining initial base prices.
To compare transactions prices in 1979,
1980 and 1981 for coal supplied
Transactions Price Ecuatiocs
(standard errors in Parei theses)
Dependent Variabl e: Actual Transactions Price in 1979, 1980 or 1981.
1980 Tra nsactions
KOTE: One outlier has been dropped froE 1981 transactions price equation (5).
All three observations for the associated contract have been dropped froa the
pooled regressions. Observations for this contract do not appear in the
equations with E because B »as not available for these observations.
pursuant to contracts written at different points in time, it is useful to use
the estimated relationships in Table
transactions prices in
each year (1979, 1980 and 1981) and on average for the 1979-81 period for
contracts written in each of the four market periods. These comparisons are
presented in Table
for equations estimated with and without H.
it is clear that contract prices are
"flexible" in the sense that they change over time, and as first order
approximation, moved along with transactions prices in the market generally
during this time period. Despite the very large differences in the initial
base prices in contracts negotiated in different years, by the 1979-81 period
mean transactions prices for contracts of different vintages are within a
range of 10-15% of one another. The differences are largest when we make the
censoring correction, which is not surprising since the greatest incidence of
"missing" contracts is associated with relatively short contracts executed in
the pre-1974 period.
there are indeed some long run rigidities in price adjustment
observed during the period of time covered by my data. Transactions prices
observed in 1979, 1980 and 1981 are lower for pre-1974 contracts than for
post-1974 contracts, as was expected. If we use the transactions prices for
contracts negotiated in 1978 and 1979 as more closely reflecting market
conditions in the 1979-81 period, the difference between pre-1974 and post1974 contract transactions prices is on the order of 10-15%. The difference
between transactions prices associated with 1974-77 contracts and later
contracts is much smaller numerically and this difference is not statistically
significant. The historical market conditions that characterized price
formation in the pre-1974 period appear to affect transactions prices many
Predicted Transaction Price By Contract Execution Period
(cents per million btu's)
Single Year Transaction Data
Source: Predicted Values from Equations in Table 6 using the sample means for
the independent variables. Coal characteristics and reaional mix are thus held
constant for purposes of generating the comparative information above. OLS/H
refers to those equations that include H.
The pricing behavior that we observe is perfectly consistent with the use
of BPE and cost plus adjustment provisions during an era in which demand grew
and production costs increased. We observe both the flexibility and the
rigidity that these adjustment mechanisms are likely to exhibit. The price
difference observed for 1979-81 is probably too large to be completely
explained by changes in the long run price trajectory pre and post 1974, due
to changes in demand expectations.
Cost side rigidities and short run price
effects probably explain part of the difference as well. ^
These results suggest that the written provisions of contracts are in
fact meaningful since we do find systematic differences in transactions prices
between contracts negotiated in different periods of just the type we would
expect to observe given the structure of price adjustments provisions and the
nature of changes that took place in coal markets during this period of time.
If formal contract provisions were not particularly meaningful and the parties
simply renegotiated prices each year (for example)
would have expected to
find no systematic differences in transactions prices associated with the
period of time that contracts were negotiated.
While the results for the pre-1974 contracts as a group are consistent
with expectations, the results for the pre- 1971 contracts are a little bit
1979-81 transactions prices are lower for these contracts
than for the post-1974 contracts, the difference is smaller than that for the
contracts signed immediately before 1974. My expectation was that these
contracts would have performed especially poorly given what was expected when
they were written.
The most likely explanation for this result is that these contracts
really did perform quite poorly after 1973 in tracking cost changes as costs
of all kinds increased at rates and in ways not anticipated when they were
written. Indeed they may have performed so poorly that the sellers could
credibly appeal to gross inequity and force majeur provisions or credibly
threaten to breach through bankruptcy to force renegotiation of escalation
provisions sometime in the mid or late 1970'
In some cases, buyers may have
been quite willing to renegotiate to reduce take or pay requirements where
emissions regulations reduced the demand for previously contracted for coal.
Such renegotiations would probably have resulted in escalation provisions
similar to those being negotiated in contemporary contracts and some
adjustment in base prices to reflect past cost increases that had not been
reflected in prices. There is no reason to believe that buyers would have
agreed to move prices up to then prevailing market levels, however, if this
was not necessary to make continued performance economical for the seller.
There is no obvious way to test systematically this hypothesis without
obtaining complete histories of contracts and contract renegotiation for the
observations in my data base. This would be impossible.
have put together a
reasonably complete contractual history for one coal contract negotiated about
believe is informative. The contract involves the supply of coal
to the Four Corners Generating Station in New Mexico
(a mine -mouth plant)
first contract was executed in 1960 and covered deliveries for three
generating units to be built at this cite, with options to extend supply
commitments for additional units
A second related contract was executed in
1966 to cover coal supplies for two additional (and much larger) generating
The 1966 contract uses a BPE adjustment formula. The base price is broken
down into six different components. All but one of these components is indexed
in some way. The non-escalating component includes profits,
rents and some
depreciation. The contract includes a gross inequity provisions, but does not
include a specific provision for recovery of costs associated with new
government regulations. Prior to 1975, there were several technical amendments
to the contracts, but these did not affect prices levels or the price
adjustment formula. In 1975, the contract was amended to allow for the
recovery of costs associated with land reclamation expenses resulting from the
New Mexico Coal Surface Mining Act. In 1977, the contract was amended again to
allow for the recovery of new severence taxes imposed by the State of New
Mexico. On January
1981 the escalation provisions in the 1966 contract were
revised. The seller appealed to the portion of the contract that provided for
revision of methods of price escalation if there should occur extreme or
radical changes from economic factors and conditions which existed at the time
.which seriously distorted or rendered clearly inequitable the
application of the methods of escalation set forth in the Agreement."
The new pricing formula added a complicated
component to the price which provides for additional adjustments for changes
in production costs that were sustained since mid-1979.
It also provides for
monthly adjustments of those components that had previously been adjusted
annually. Recovery of costs associated with compliance with new environmental
regulations is provided for, is laid out in great detail and the provision
appears to include recovery for cost incurred prior to 1981 if they have not
already been incorporated in the base price through previous amendments.
Finally, the amendment gives the buyer or seller the right to seek a
lower/higher price after five years if the seller's rate of return on
investment significantly exceeds or falls short of the "normal" rate of return
for a similar business.
To the extent that other contracts executed during the 1960's went through
similar adjustments in the late 1970
as did the Four Corners Agreements,
this would help to explain the pricing patterns observed for pre -1971
While coal contracts written between 1960 and 1978
appear to have
adjusted fairly well on average to changing market conditions, as revealed by
comparative transactions prices in 1979, 1980 and 1981, this does not mean
that all of the contracts did
adjustment formulas can lead prices
To get some sense for how far off
have examined a set of outlier contracts
associated with both the base price and later transactions price
relationships. An outlier was defined as an observation that had a Studentized
residual with an absolute value of more than 1.96.5°
Seven of the
equation there were 13 contracts that met this criterion.
contracts have negative and six positive residuals. The large negative
residuals are primarily associated with contracts written in 1974.
that these contracts were initially negotiated before market conditions
tightened. The large positive residuals are primarily associated with earlier
contracts. Only one of these 13 contracts, however, shows up as an outlier in
the transactions price equations. This raises the possibility either that
these contracts had atypical adjustment provisions, or that the base prices
were subsequently renegotiated.
When we turn to the transactions price relationships
contracts that met the outlier criterion in at least one of the equations.
The outliers come from all regions, all contract execution periods, and are
associated with contracts of varying durations. The absolute value of the
residuals is roughly 50% of the expected price.
For 20 of these contracts we
can match the 1979-81 transactions prices with the associated base price. With
one exception these contracts were not outliers in the base price equation,
they appear to have had adjustment provisions that led "typical" base prices
to become "atypical" transactions prices over time,
other things equal.
Furthermore, the distribution of outliers is not symmetric. Sixteen of the
twenty-two contracts have large positive residuals, meaning that actual
transactions prices are much higher than predicted. Only six have large
negative residuals. This suggests that the adjustment provisions were more
likely to get far off on the high side than on the low side. Most of the
sixteen contracts with large positive residuals are either post-1973 contracts
or earlier contracts that used a cost-plus adjustment mechanism (2)
the six contracts with large negative residuals,
only one was signed before
1974. Three of the six contracts were signed by the same utility at about the
Since only one of the outliers associated with the transactions price
equations also appeared as an outlier in the base price equation, it does not
appear that these outliers are associated with special coal characteristics
that are not being measured properly in the price relationships. However, of
the fifteen contracts with large positive transactions price residuals for
which we can match base price information, twelve had positive, though not
necessarily large, residuals in the base price equation as well. This suggests
that for some reason these contracts had prices that started out higher than
expected and got further and further from their expected value over time.
(There is nothing particularly noteworthy about the three other contracts with
positive residuals.) Of the five contracts with large negative residuals which
we can match with base prices,
two also had negative,
though not necessarily
residuals in the base price equation. The prices in these contracts got
lower and lower compared to the expected value over time. The remaining three
contracts with negative transactions price residuals had positive residuals in
the base price equation. These contracts were written at about the same time
by the same utility, but for coal from different supply regions. This suggests
that this particular utility may have chosen to rely on contracts which
incorporated a significant fixed price component in the base price.
While the contracts as a group adapted to changing market conditions
it is clear that individual long term contracts that specify the
way prices will be determined over time ex ante can eventually yield
transactions prices that are way out of line with prevailing market
Relationship specific investments often make it desirable for electric
utilities and their coal suppliers to enter into long term contracts. A major
challenge in structuring such contracts involves the specification of price
adjustments provisions that both guard against opportunistic behavior and do
not lead to serious adaptation problems as the contractual relationship plays
itself out. The kinds of price adjustment mechanisms that are typically relied
upon in long term coal contracts appear to have been reasonably successful in
responding to this challenge during the 1970'
and early 1980'
The contractual adjustment provisions in long term contracts provide for
price flexibility at least in response to changes in the costs of production
but also embody some potential long run rigidities. Between 1970 and 1981
nominal coal prices increased by a factor of about four. Real coal prices
doubled. Yet transactions prices pursuant to long term contracts tracked
prevailing market conditions during the 1970 's quite closely. While we find
some significant rigidities in contract prices, the price disparity between
contracts of different vintages resulting from these rigidities is on the
order of only about 10- 15% on average by the 1979-81 period. While a 15%
difference in average transactions prices is far from being trivial it is
still relatively small compared to the overall movement of prices in the
market over time. There
is also some
evidence that some individual contracts
did very poorly in tracking changing market conditions during this time period
and exhibited substantially more rigidity in this sense than the average
contract. At least during the time period studied here, the contractual
provisions in long term coal contracts did a fairly good, but far from
perfect, job of adapting to changing market conditions.
It will be interesting
to see in future research how these contracts performed as time went on and
the coal market softened considerably.
^-Professor of Economics, Department of Economics, MIT, Cambridge, MA 02139.
The research for this paper was conducted while the author was on sabbatical
leave from MIT at the Center For Advanced Study in the Behavioral Sciences.
Support from MIT and the Center is gratefully acknowledged. Leslie Sundt
provided valuable research assistance. Keith Crocker, Oliver Hart, Victor
Goldberg, Jean Tirole and a referee read an earlier version and provided
2 Paul L. Joskow, Vertical Integration and Long Term Contracts: The Case of
Coal, 1 Journal of Law, Economics and Organization 33 (1985) and Paul L.
Joskow, Contract Duration and Relationship Specific Investments: Empirical
Evidence From Coal Markets, 77 American Economic Review
^The literature is discussed in Joskow (1985) and (1987) supra note 2, and
Paul L. Joskow, Long Term Vertical Relationships and the Study of Industrial
Org mization, 141 Journal of Theoretical and Institutional Economics 586
^See especially Oliver Williamson, Credible Commitments: Using Hostages to
Support Exchange, 73 American Economic Review 519 (1983), Benjamin Klein,
Robert Crawford and Armen Alcian, Vertical Integration, Appropriable Rents and
the Competitive Contracting Process, 21 Journal of Law and Economics 297:
(1978), and Oliver Hart and Bengt Holmstrom, The Theory of Contracts, Working
Paper #418, Department of Economics, MIT (March 1986). Keith Crocker and Scott
Masten, Mitigating Contractual Hazards: Unilateral Options and Contract
Length, Working Paper #449, Graduate School of Business, University of
Michigan (1986), examines the duration of natural gas contracts in the context
of a model that tries to measure both the costs and benefits of long term
contracts. Scott Masten and Keith Crocker, Efficient Adaptation in Long-Term
Contracts: Take-or-Pay Provisions For Natural Gas, 75 American Economic Review
1083 (1985), discusses take-or-pay provisions in natural gas contracts.
Victor Goldberg's papers are the only ones that I am aware of which explore
this question with information from real contracts. Victor Goldberg, Price
Adjustment in Long-Term Contracts, 1985 Wisconsin Law Review 527 (1985), and
Victor Goldberg and John Erickson, Long Term Contracts for Petroleum Coke,
Working Paper No. 206, Department of Economics, University of California at
"Dennis Carlton's recent paper is a notable exception. Dennis Carlton, The
Rigidity of Prices, 76 American Economic Review 637 (1986).
Of a sample of 160 coal contracts in force in 1979 and that had contractual
termination dates in 1983 or later, I found that all but three continued to
operate in 1983 despite significant changes in coal markets between 1979 and
Joskow (1985) and (1987), supra, note
Williamson, supra note 4 at 526. As in my early papers on coal supply
arrangements, I ignore risk sharing explanations for the reliance on and
structure of long term contracts. See A.M. Polinsky, Fixed Price vs. Spot
Price Contracts: A Study In Risk Allocation, 3 Journal of Law, Economics and
for a discussion of the role of risk sharing
considerations in the choice between fixed price and market price contracts.
long run supply function for coal to U.S. utilities is upward sloping,
although it appears to be fairly elastic. As a result, Hotelling type "user
costs" associated with resource depletion are relatively small. See Martin
Zimmerman, The U.S. Coal Industry: Economics of Policy Choice, MIT Press,
Cambridge, MA (1981) at 91-93
^These economic rents will not necessarily accrue
to the coal supplier since
the rights to mine the coal often must be secured from third parties
-^For example the contract is long enough to pay off fully the relationship
specific component of investments made by the seller and/or that notice,
termination and damage provisions otherwise provide for the recovery of losses
due to premature termination.
l^The terms and conditions of coal contracts negotiated between a coal
supplier and an electric utility are not regulated directly by state public
utility commissions. The costs of coal purchased under long term contract
would generally be treated like any other fuel cost and passed on in final
electricity prices through a fuel adjustment clause. Regulatory agencies can
deny cost recovery, however, if they determine that the utility has signed an
"imprudent" contract. Regulatory agencies have sometimes disallowed fuel
purchased pursuant to long term contracts. Regulatory agencies are likely to
become especially interested in prudence issues when contract prices appear to
be higher than what other buyers are paying for coal.
-^Court awarded damages aside, the buyer is not in a good bargaining position
since contract prices may be significantly below the price of his next best
The seller will attribute any shortfalls in quantity or quality to factors
outside his control or try to exploit ambiguities in the terms of the
contract. Especially if cost increases result from changes in government
regulation or union work rules and prices are less than identifiable
production costs, the seller may try to rely on vague doctrines of commercial
impracticability or force majeur to prevail in court. Bankruptcy may very well
be an attractive strategy for the seller in some cases.
A simple numerical example is useful for understanding the nature of the
performance problems that emerge with fixed price long term contracts. The
average cost of production from a typical mine is composed of operating costs
that are variable in the short run and capital related costs which are not.
The breakdown is roughly 75% short run variable costs and 25% costs that are
fixed in the short run. With reasonable assumptions about expected inflation
rates and interest rates during the 1970' s, a 20 year fixed price contract
which satisfies (1) will have a price that falls below the then current
variable costs of production after between 10 and 15 years if inflationary
expectations are realized. It would be economical for the supplier to shut
down at this point assuming that all capital related costs are sunk in year
zero. If we recognize that capital investments actually are made fairly
continuously through the life of a mine, the shutdown point could be much
sooner. A twenty year commitment will simply not be credible if the seller is
expected to have strong financial incentives to walk away from the deal in the
middle of the contract even if all expectations about costs and market values
A set of numerical simulations that amplify on this
are realized exactly.
discussion are available from the author.
^With a fixed price contract and expected increases in nominal production
costs and market values over time, in order to satisfy (1) the contract price
is likely to start out being higher than the then current average transactions
price and end up lower. If the regulatory agency compares contract prices with
average transactions prices in the early years of the contract the difference
in prices may lead to an unjustified disallowance. This would not be made up
with a reward in later years. Using the same numerical example as reported
above, the initial base price for coal pursuant to a twenty year fixed price
contract could be 50% to 70% above the then prevailing average cost of
'-"Since transportation costs are a large fraction of the buyer's ultimate cost
of coal and neither supply nor consumption is uniformly distributed across the
country, prices at the mine will vary from location to location.
'•'The econometric analysis reported below confirms these assertions regarding
the market value of coal quality and location.
^°There is for delivered prices, but these include transportation costs.
^The buyer could argue of course that he must pay more than "the market
price" to replace the contract and the seller could argue that he must accept
less than "the market price" to dispose of the supplies, but this just gets us
back to the problem of defining a meaningful market price.
For example, if there is an unexpected short run increase in demand, short
run market values may increase faster than average production costs in the
short run. If the shift in demand is permanent the expected trajectory of
future prices may rise as well if there is a significant long run economic
rent component in the market value of coal. Or the demand for coal could fall
suddenly, there could be a lot of resulting excess capacity and prices would
fall to reflect this situation. If the reduction in demand is permanent, the
expected trajectory of future prices could shift down if there is a
significant economic rent component in the average contract price.
Fixed price contracts become more viable from this perspective as the agreed
upon duration of the agreement gets shorter and the magnitude of changes in
expected nominal production costs and market values over the term of the
agreement gets smaller.
Market price provisions will be more viable when the quasi-rents associated
with relationship specific investments are small.
These contracts generally have durations of greater than four years.
26 Over fifty
of which are also represented in the data base
discussed in the
The source that I relied on for the information in Table 1 did not generally
provide more detailed information than a simple categorization of the price
adjustment formula. Based on actual contracts that I have reviewed it is
likely that at least some of the BPE contracts have "reopener" provisions that
allow for renegotiation at some point in time or when certain contingencies
^Although I have all of the information required for the econometric analysis
reported below for only 247 of these contracts.
^Between 1973 and 1978 coal accounts for a slightly declining fraction of
electricity production. Oil's share is about constant. Nuclear' s share
increases by a factor of almost three as nuclear facilities committed before
1973 were completed. It is only after 1978 that fuel oil consumption declines
dramatically and utilities rely much more on coal for generating electricity.
•^Oil consumption by electric utilities does not begin to decline
significantly until after 1978.
31-The Fuel Use Act of 1978 restricted the use of oil or gas in new utility
boilers as well.
-^Setting the stage for a subsequent collapse in spot and new contract prices
after roughly 1983. This later period is the subject of ongoing research.
33 Federal Coal Mine Health and Safety Act of 1969.
The Federal Mine Safety and Health Act of 1977 and related state laws.
•"e.g. The Surface Mining Control and Reclamation Act of 1977, Federal Water
Pollution Control Act, Solid Waste Disposal Act, Safe Drinking Water Act, Soil
and Water Resource Conservation Act and related state laws.
input prices were no doubt affected by increased demand for the same
inputs by the oil and gas drilling sectors and by uranium mining which also
expanded at the same time.
•^'See Paul L. Joskow and Richard Schmalensee, The Performance of Steam
Electric Generating Plants in the U.S.: 1960-1980,
Journal of Applied
JO See Richard Schmalensee and Paul
Joskow, Estimated Parameters as
Independent Variables: An Application to the Costs of Steam Electric
Generating Units, 31 Journal of Econometrics 275: (1986)
The Clean Air Act of 1970 and associated regulations affecting existing
sources and new sources are especially important.
^ u See Williamson,
supra note 4. For example, suppliers might try to recover
relationship specific investments quickly by seeking higher base prices but
lower escalation rates rather than dealing with differences -in the importance
of relationship specific investment by adjusting the term of the agreement as
I suggested earlier. Buyers might try to protect themselves with lower base
prices and higher escalation rates.
All prices are FOB the mine and are expressed in nominal cents per million
The Theory and Practice of Econometrics, Wiley,
George Judge et. al
James Heckman, The Common Structure of Statistical Models
(1985) at 610-613,
of Truncation, Sample Selection and Limited Dependent Variables and a Simple
Estimator for Such Models, 5 Annals of Economic and Social Measurement 475:
(1976), and James Heckman, Sample Selection Bias As A Specification Error, 47
Econometrica 153: (1979).
number of observations varies somewhat from equation to equation because
of missing data. I do not have information on contract duration for four
observations. Values for H are not available for six observations.
^"There does not appear to be a contract quantity effect either. When a
variable measuring contract quantities is introduced, its coefficient is not
significantly different from zero.
^'The results are not sensitive to the relationship chosen from Table
^"The average transactions price at a particular point in time is of course an
imperfect indicator of prevailing market conditions as they affect prices
pursuant to new long term contracts The average transactions price includes
both spot transactions and contract transactions Spot market transactions
were unusually high in 1974 and 1975 and spot prices are likely to have jumped
considerably in these two years as well. There does not appear to be a
separate series for spot prices FOB the mine. However, 24% of coal delivered
to electric utilities in 1974 and 18% in 1975 was spot coal, compared to an
average of less than 15% for 1976-1982. See Joskow (1985, p. 53). The average
transactions price figure also averages prices pursuant to contracts
negotiated in many different years. If there are price rigidities, we would
expect to find some differences between base prices in new contracts and
contemporaneous average transactions prices.
^'Obviously I do not have transactions price information for each contract in
each of these years. Aside from reporting gaps, some of the contracts in the
data base came to an end in 1979, 1980 or 1981.
-^DURATION is never significantly different from zero in the transactions
price equations, so I have not bothered to report estimates with it included.
-'-The long run supply function for coal sold in the U.S. appears to be quite
elastic. As a result, the long run price effects of even large increases in
the expected demand for coal should be modest. Zimmerman's calculations
suggest that large changes in the expected demand for coal would change
prevailing market prices by only 10% to 20% in the year 2000. The long run
Hotelling rent component of prices in 1980 associated with such an increase in
coal demand over the long run should be quite small. See Zimmerman, supra
note 10, at 91-98.
To the extent that adjustment provisions include adjustments for average
changes in productivity rather than changes attributable to a particular mine,
some of the increases in economic rents may be captured by the pricing
provisions in existing contracts.
5^When the equation for 1981 transactions was first estimated, I detected one
contract that was a very large (six standard deviations) outlier. The
associated reported price is much higher than anything else in the sample and
much higher than it was in 1980. The observation was dropped from the 1981
transactions equation and the associated contract was dropped from the pooled
regression equations. This turned out to be a contract for which H is not
available, so that this contract does not appear in any of the equations
including H. This contract is not an outlier in the base price equation,
however. Outliers are discussed in more detail below.
->^When this work is extended to the post 1983 period
downward rather than upward price rigities.
expect to observe
-'-'Nor does it mean that large differences did not emerge later. Coal markets
softened considerably after 1981 and when the data are available it will be
interesting to see how transactions prices pursuant to these contracts
compared to prices charged in new long term contracts in (say) 1984 and 1985.
I plan to explore this further when the data become available to me.
David A. Belsley, Edwin Kuh, and Roy
Welsch, Regression Diagnotics
-''The transactions price equations were reestimated without these contracts as
well. The key results reported earlier were unaffected. The outliers
consistently have large residuals in each transaction year 1979-81.
Contract Data Base
The data base that
make use of to estimate the base price and
transactions price equations discussed in the body of the paper was
constructed for this analysis from information contained in The Guide To Coal
Contracts (Pasha Publications, Arlington, VA, 1981 and 1983 Editions). The
1981 Edition contains information for coal contracts executed prior to 1980
based on reports filed by investor -owned utilities with the Federal Energy
Regulatory Commission. The 1981 Edition picks up information on the initial
base price specified in each contract when this information was provided by
the utility to the FERC. Base price information is not contained in subsequent
Editions of The Guide To Coal Contracts
This source was also used to obtain
information on coal characteristics and the location of mines. The 1983
Edition reports transactions prices by contract for 1979, 1980 and 1981. To be
included in the data base, contracts had to appear in both Editions and all
information required for the analysis provided.
The 1981 and 1983 Editions of the Guide To Coal Contracts provided no
information on price adjustment provisions. The 1985 Edition (which was
released after the primary data set was constructed) did provide such
information for some contracts.
Contracts along with information
used the 1985 Edition of the Guide To Coal
obtained from a set of actual coal
contracts to obtain the information that is reported in Table
Carlton, Dennis W. "The Rigidity of Prices,"
(September 1986): 637-658
American Economic Review 76
"Mitigating Contractual Hazards:
Crocker, Keith J. and Masten, Scott E.
Unilateral Options and Contract Length," Working Paper #449 Graduate
School of Business Administration, University of Michigan,
Goldberg, Victor. P. "Price Adjustments in Long-Term Contracts,"
Wisconsin Law Review (1985): 527-543
Goldberg, Victor P. and Erickson, John R. "Long Term Contracts for Petroleum
Coke," Working Paper No. 206, (1984) Department of Economics,
University of California, Davis, California (forthcoming in the
Journal of Law and Economics as "Quantity and Price Adjustment in
Long-Term Contracts: A Case Study of Petroleum Coke")
Hart, Oliver and Bengt Holmstrom, "The Theory of Contracts," Working Paper
#418, Department of Economics, MIT (March 1986)
Heckman, James. "The Common Structure of Statistical Models of Truncation,
Sample Selection and Limited Dependent Variables and a Simple
Estimator For Such Models " Annals of Economic and Social
Measurement 5 (1976): 475-492.
Heckman, James. "Sample Selection Bias As A Specification Error,"
Econometrica 47 (January 1979): 153-161.
Joskow, Paul L. "Vertical Integration and Long Term Contracts: The Case of
Coal-burning Electric Generating Plants," Journal of Law, Economics
and Organization 1 (Spring 1985): 33-80.
Joskow, Paul L. "Contract Duration and Relationship Specific Investments:
Empirical Evidence From Coal Markets"
American Economic Review
Joskow, Paul L. "Long Term Vertical Relationships and the Study of Industrial
Organization," Journal of Theoretical and Institutional Economics 141
(December 1985) 586-593
Joskow, Paul L. and Schmalensee, Richard. "The Performance of Steam Electric
Generating Plants in the United States: 1960-1980," Journal of
Applied Econometrics (in Press)
Judge, George. G. et
The Theory and Practice of Econometrics
Klein, Benjamin, Crawford, Robert .G. and Alchian, Armen.A. "Vertical
Integration, Appropriable Rents and the Competitive Contracting
Process," Journal of Law and Economics 21 (October 1978): 297-326.
Scott E. and Keith J. Crocker. "Efficient Adaptation in Long-Term
Contracts: Take-or-Pay Provisions for Natural Gas," American Economic
Review 75 (December 1985) 1083-93
Polinsky, A. M. "Fixed Price vs. Spot Price Contracts: A Study in Risk
Allocation," Journal of Law, Economics and Organization 3 (1987),
Schmalensee Richard, and Joskow, Paul L. "Estimated Parameters As Independent
Variables: An Application to the Costs of Steam Electric Generating
Units," Journal of Econometrics 31 (April 1986) 275-305
Williamson, Oliver E.
"Transaction-Cost Economics: The Governance of
Contractual Relations," Journal of Law and Economics 22 (October
Williamson, Oliver E.
"Credible Commitments: Using Hostages to Support
Exchange," American Economic Review 73 (September 1983): 519-540.
The Economic Institutions of Capitalism
Zimmerman, M. The U.S. Coal Industry: Economics of Policy Choice
1981, Cambridge, MA.