JOURNAL OF ECONOMK THEORY 19, 28-37 (1978)

Resource Extraction, Substitute Production, and Monopoly*

MICHAEL HOEL

Unhersity of Oslo, Box 1095 Blindern, Oslo 3, Norway

Received November 9, 1976; revised June 12, 1978

1. INTRODUCTION

The development of price and extraction of a nonrenewable resource

owned by a monopolist has been analyzed by Kamien and Schwartz [4],
Lewis [5], Stiglitz [8], Sweeney [9], and others. The present paper analyzes
how the extraction of a resource owned by a monopolist is affected by the
existence of a perfect substitute for the resource.
To focus on the existence of a substitute produced by a “backstop
technology” (see [6]), we disregard changes in the demand function over time
(see [S]) and costs depending on current extraction (see [9]), or on remaining
reserves (see [2, 71). Section 2 gives the monopolistic extraction paths in
three cases; (1) no substitute exists, (2) the resource owning monopolist
controls the production of the substitute, and (3) the substitute is supplied
competitively. Section 3 compares these three cases for a general class of
demand functions. In Section 4 the special case with a constant demand
elasticity and zero extraction costs is treated, before some conclusions are
given in Section 5.

2. THE THREE MONOPOLY CASES

In all three cases a monopolist is assumed to maximize discounted profits

W, where

w= I (r e-rt[F(x(t) + y(t)) - bx(t) - q(t)] dt, (1)

0

* An earlier version of this paper was written during a one year visit at MIT, and present-
ed at The European Meeting of the Econometric Society in Vienna, September 1977.
Financial support from The Bank of Norway’s Fund for Economic Research and The
United States Educational Foundation in Norway is gratefully acknowledged. I would
also like to thank Donald Hanson, Hal Varian, and two anonymous referees for useful
comments.
28
0022-0531/78/0191-0028$02.00/0
Copyright 6 1978 by Academic Press, Inc.
All rights of reproduction in any form reserved.
 RESOURCE EXTRACllON AND MONOPOLY 29

subject to
S(t)= -X(f), S(O)
= so9 (2)
so>3 0, x(t)3 0, y(t)3 0. (3)

Here x and y stand for resource extraction and substitute production by the
monopolist, while S is the remaining resource stock (S, is given). The discount
factor (v) and the unit cost of producing the substitute (c) are both positive,
and the unit cost of resource extraction (b) satisfies 0 < b -C c. The function
F(x + JV)gives the monopolist’s total revenue.
In the case with no backstop technology we have the additional constraint
u(t) = 0, and the revenue function is

F(x) = xf(x), (4)

where f-‘(p) is the demand function and p is the price of the resource.
Throughout our analysis we assume that

L$df(X) = cc (5)

so that we can disregard the possibility of zero resource extraction when there
is no substitute. We also assume that f’ < 0 and that X$(X) is strictly
concave, so that the marginal revenue function declines with x everywhere.
Under these assumptions it is well known [5, 81 that the monopolist’s optimal
extraction path makes the marginal current profit rise with the rate r, with the
initial extraction determined so that
co

I0
x(t) nt = s, . (6)

Defining
q = g(x) = f(x) + xf’(x) - b (7)
as the marginal current profit, we therefore have

h(t) = 4dO) ert, (8)

where the subscript 1 refers to case number 1. The value of ql(0) is determined
so that (6) holds when x = g-l(q) is used.
If substitute production is possible and controlled by the monopolist there
are no additional constraints to our problem (l)-(3), and F(x + y) =
(x + y)f(x + y) like before. Since c > b and r > 0, it is obviously not
optimal for the monopolist to start producing the substitute before the
resource is exhausted. Furthermore, after the date of exhaustion we must
have marginal revenue equal to marginal cost, i.e., q2(f) + b = c from (7)
(the subscript 2 refers to case number 2). Like before, we have q2(t) rising
30 MICHAEL HOEL

with the rate Y as long as resource extraction takes place, and it is easy to
verify that q2(t) must reach its long-run value (c - b) exactly when exhaustion
occurs (see, for instance, Dasgupta and Stiglitz [I] or Hoe1 [3]). The solution
in this case is therefore given by
q2(t) 1 (c - /J) .@+),

where T,
is the date of exhaustion. The value of
that (6) must hold with x(t) = g-l(q2(t)) for t <
T,
T,
is determined by the fact
and x(t) = 0 for t > T,
inserted.
Our third monopoly case is slightly more complicated than the two
previous cases. Here we have the constraint y(t) = 0 like in case number 1
(remembering that y is the quantity of the substitute produced by the
monopolist). Furthermore, the monopolist cannot sell any resources at a
price exceeding the competitive substitute price c. To simplify the mathe-
matics, we assume that the monopolist can sell as much as he wishes, up to
the total demand f-‘( p), at p = C. Since r > 0, it will not be profitable for
the monopolist to have 0 < x(t) <f-‘(c) for any period, since he at the
price c could obtain a higher discounted profit by letting x(t) = f-‘(c) first
and x(t) = 0 later in such a period. We can therefore solve the optimization
problem by introducing the additional constraint

x(t) 3 f-‘(c) or x(t) = 0. (10)

For the same reason as 0 < x(t) < f-‘( c) never would be optimal it is clear
that it cannot be optimal to have x(t) > 0 after a period where x(t) = 0.
The optimization problem (l)-(3) with the additional constraints y(t) 7: 0
and (10) therefore has the same solution as maximizing
r3
s e-‘6[F(x(t))
- bx(t)]
dt
0

subject to (2) (3), and x(t) > f-l(c), where the date of exhaustion
control variable. The Hamiltonian corresponding to this problem is
T, is a

H(x, A, t) = P”[F(x) - bx] - xx, (11)

where h is a positive constant since S does not enter the function H. The
optimal extraction path x3(t) must satisfy

f&(x3(t), A t> < 0,

where < implies x3(t) = f-‘(c). Together with (4), (7), and (11) this means
that

(12)
 RESOURCE EXTRACTION AND MONOPOLY 31

where < implies x(t) = f-‘(c). The horizon T3 must satisfy H(x,(T.&
h, TJ = 0, i.e., from (4) and (11)

[e-‘r3(f(~,(T3)) - b) - A] x3(T3) = 0. (13)

From (12) we see that x3(t) is nonincreasing. Furthermore, the constraint

x(t) > f-l(c) must sooner or later be binding, otherwise the present case
would be identical case 1, and since (8) implies that x1(t) <f-‘(c) for a
sufficiently large value of t, this would give us a contradiction. These two
properties of the solution x3(t) imply that x,(T,) = f-l(c), so that (13) gives
us
h = eeTT3(c- b). (14)

The solution for q3(t) = g(xS(t)) follows from (12) and (14):

q3(t) = Min[q, (c - b) e-‘(rZ-t)], (15)

where
q = g(f-l(c)) < (c - b). (16)

Like before, T, is determined by the fact that (6) must hold with x(t) =
g-l(qg(t)) for t < T3 and x(t) = 0 for t > T3 inserted. In other words, the
monopoly solution in this case has two phases: In the first phase the marginal
current profit rises with the rate r until the resource price reaches its upper
limit c in the end of this phase. In the second phase the resource price is
constant and equal to c, and this phase lasts until the resource is completely
exhausted. Notice that the second phase must always occur in a solution
(cf. (15) and (16), which imply that q3(t3) = Zj for t sufficiently close to T,).
However, the first phase need not exist. This is true no matter how large S,,
is if 4 < 0. Since b > 0, it is clear that a suficient condition for 4 < 0 is
that the demand elasticity (=f(-xi’)) is
. smaller than one for all p < c. If
the second phase does not exist (due to 4 < 0 or S, sufficiently low to make
e-rT3(c - b) 3 q), the monopolist’s solution is simply to set x(t) = f-l(c)
until the resource is exhausted.

3. COMPARISON OF THE THREE CASES

We now compare the three monopoly solutions. In particular, we compare

the initial price and extraction, remembering that p is strictly increasing in q
(p = f(g-l(q))) and that x is strictly decreasing in q (x = g-l(q)).’
The first difference we notice is that while the natural resource never is
completely exhausted when no substitute exists, it is exhausted in finite time
in both the cases in which a substitute exists.

64211911-3
32 MICHAEL HOEL

Since the first two monopoly cases both are characterized by q(t) increasing
with the rate r whenever extraction takes place, q2(t) - ql(t) must have the
same sign for all t < T, . But since the resource is exhausted in finite time
when a substitute exists, and x is higher the lower q is, we must have q2(t) -=c
ql(t) for all t < T, . From the relationship between q(t), p(t) and x(t) we
therefore can conclude that

P&> < PI(t) for all t,

x20>> xl(t) for all t < T, .
In other words, the resource is extracted at a faster rate before exhaustion
occurs in the case with a substitute than in the case without a substitute. The
price of the resource and substitute is therefore always lowest in the case with
a substitute.
Let us now compare the date of exhaustion in the two cases with a sub-
stitute. Assume first that T, 3 Tz . From (9) and (15) we see that this implies
q3(t) < q2(t) and x:(t) > x,(t), with a strict inequality holding for t sufficiently
close to T3 . But since the same total stock S,, is assumed to be extracted by
the date of exhaustion in both cases, this inequality must imply T3 -=cT, .
This contradiction proves that T3 > T, cannot be true, i.e., we have

In other words, the resource is exhausted at an earlier date when the substitute
is supplied competitively than when it is controlled by the resource owning
monopoly.
From (9) it is clear that we may well havep,(t) > c for all t. Sincep,(t) < c,
this means that pz(t) > pa(t) for all t is possible. However, we see from (9),
(15), and (18) that
PdO)> P,(O),
(19)
x3(0) -=cx,(O) if p,(O) < c.

This result is rather surprising. The case in which the substitute is supplied
competitively is in a sense more competitive than the case in which the
monopolist controls substitute production as well as resource extraction.
Intuitively, one might expect the price alway to be lowest in the most com-
petitive case, at least for some well-behaved demand functions. The result
above shows that this is never true unless the constraint pa(t) < c is binding
for all t: The initial price is higher, and initial extraction lower, in the case
with a competitively supplied substitute than in the case in which the
monopolist controls the substitute production, as long as pa(O) < c.
 RESOURCE EXTRACTION AND MONOPOLY 33

Since T3 < T, , we obviously must have x3(f) > x&) for some t < T3,
i.e.,
PdO< A(t),
(20)
40 > x2(t)
for t smaller than but sufficiently close to T3 . This result helps explain why
~~(0) > ~~(0) is not so surprising after all: The existence of the constraint
p3(t) < c when the substitute is supplied competitively prevents the mono-
polist from charging as high a price sometime in the future as he would have
if he controlled the production of the substitute. It then seems reasonable
that some of this reduced future price will be compensated for by raising the
price early in the period of extraction.
From (17) and (20) we see that

P&) < P*(t),

(21)
x3(f)> x10>
for t smaller than but sufficiently close to T3. However, with the genera1
assumptions used so far it does not seem possible to sign pa(O) -p,(O) or
x,(O) - x,(O) unambiguously even for the case in which p&O) < c. In the
next section we see that this can be done for the special case with zero extrac-
tion costs and a constant demand elasticity.

4. ZERO EXTRACTION COSTS AND A CONSTANT DEMAND ELASTICITY

In this section we assume that b = 0 and that

f(x) = x-a, where O<a<l (22)

This implies that

q = g(x) = (1 - a) x-5 and ij = (1 U)C. (23)

From (8) and (23) we get

x1(t) = x,(O) e-‘+‘t, (24)

which inserted into (6) gives, after integrating,

x1(t) = k &e-(~i=)t. (25)

It is well known [5, 81 that the extraction path described by (25) is identical
to the extraction path under perfect competition.
34 MICHAEL HOEL

From (9) and (23) we get

x2(t) = ( 1 ; “)“” e(rlaHT+)*

Inserting this equation into (6) and integrating (remembering that x,(t) = 0
for t > T.J gives

so that (26) may be rewritten as

q(t) = [bSo + (L-$-f-)1’“] e-(Tia)t,

which holds for all t < T, . Comparing (27) with (25) of course confirms
our general result (17).
Dasgupta and Stiglitz [l], Hoe1 [3], and Nordhaus [6] have shown that the
competitive extraction path when a substitute exists is given by (9), except
that q2 is substituted by pz - b. In the present case this is the same as sub-
stituting c with c(1 - u)-l in (9). This means that the competitive solution
is given by (27) with (1 - u)/c substituted by l/c (see [ 1, Sect. 2.11). We
therefore have the following result for our present case: While the compe-
titive and monopolistic extraction paths are identical when no substitute
exists, the presence of a substitute gives the competitive path a higher rate of
resource extraction up till the date of exhaustion than the rate of extraction
is when a monopolist controls resource extraction and substitute production.
Let us now turn to the case in which a monopolist extracts the resource
and the substitute is supplied competitively. From our result (19) and what
we said above, it is clear that unless ~~(0) = c, this intermediate market
structure does not have an initial price and initial extraction rate which lies
between the prices and extraction rates of full monopoly and full competition.
To study the present market solution in more detail, we find an explicit
solution for x3(t).
From (15) and (23) we get
q3(t) = c * Min[l - a, e-T(T33-t)],

which together with (23) gives

xx(t) = c-lla * Max[l, (1 - @la e(rla)(Tz-t)]. (28)

The case in which pa(t) = c for all t is uninteresting, as it simply gives

x3(t) = c-l/~ for all t until exhaustion occurs. Let us therefore assume that
 RESOURCE EXTRACTION AND MONOPOLY 35

S,, is sufficiently large to make pa(O) < c. Denoting the time when p3(t) = c
is reached by t, , we see from (28) that
ebla)(r3-t3) = (1 _ a)-ll"
(29)
or

T3 - t, = - f In(l - a). (290

In other words, the period with constant extraction is independent of c and

S 0'
Inserting (28) in (6) and using q(t) = 0 for t > Ts gives us

w - 4 l/a,(fm] lt’ e-‘7/a’t

dt+ (T3- to)= cwso,
which after integrating, using (29) and (29’) and rearranging terms gives

1/a e(rla)T, = ; SOCllU + a + ln(l - a)

(1 -4 a

By inserting this expression into (28) we obtain

a + Ml - 4 C-l/a e-(r/a)t
fSo + a (30)
I I

which is valid for all t < T3 . From our assumption ~~(0) < c we have

x3(0)= 6 so+ a + Wa - 4 C-l,a

Since 0 < a < 1, we have a + In(1 - a) < 0. Comparing (31) with (25) we
therefore can conclude that in our present case

(32)
if ~~(0) < c.

In other words, the initial price is higher and the initial extraction is lower
when a substitute is supplied competitively than when no substitute exists.
This rather surprising result can be given an interpretation similar to the
one we gave after (20): Introducing a competitively supplied substitute
affects the monopolist in the same way as introducing a maximal price c.
If such a maximal price depresses the monopolist’s future price, some of
this price reduction may be compensated for by raising the price early in the
period of extraction.

642/19/1-4
36 MICHAEL HOEL

When a substitute exists, one expects the price to be lower the lower this
cost is. This is indeed the case when the monopolist controls substitute
production: From (27) we get ax,(t)/& < 0 for all t, which means that the
extraction will be higher, exhaustion will occur earlier and the price will be
lower the lower is the cost of producing the substitute. From what we said
above about the competitive solution, it is clear that the same results hold
under full competition. It is also easy to see that these results are valid also
for the more general case treated in Sections 2 and 3.
From (31) it is clear that as long asp,(t) < c we get ax(t)/& > 0. In other
words, the initial price is higher and the initial extraction is lower the lower
the cost of producing the substitute is. For larger values oft we havep,(i) = c,
so that the opposite result is true for this part of the extraction period.
Clearly Z&(O)/& < 0 must imply at,/& > 0 by the definition of t3 (i.e.,
p3(t3) = c). But since T, - t, is independent of c (cf. (29’)), this means that
the resource will be exhausted earlier the lower is the cost of producing the
substitute.
We saw above that the initial resource price is higher the lower the cost of
substitute production is, as long as this cost is relatively high. For lower costs
of producing the substitute (giving ~~(0) = c), the initial resource price is
lower the lower this cost is. One may be interested in knowing more precisely
how low c may be before ~~(0) = c. From (30) it is clear that the critical
value of c, denoted by C, must satisfy

Z-lb = f so + a + 1nU - 4 E-l,a,

a
i.e.,
---It@ -a)
E=
[ rS, 15’ (33)

From (33) we see that 2 is higher the lower the initial resource stock S, is
and the lower the demand elasticity I/a is (i.e., the closer to zero 1 - a is,
remembering our assumption a < 1). In the limiting case when a = 1 we get
E = + co, i.e., ~~(0) = c no matter what values S,, and c have.

5. CONCLUSIONS

We have seen that for quite general demand functions, the intermediate
market structure in which a monopolist controls resource extraction, but a
substitute is supplied competitively, gives a higher initial price and a lower
initial resource extraction than we get when a single monopolist controls
both resource extraction and substitute production. This conclusion holds
as long as the cost of producing the substitute exceeds the initial price charged
 RESOURCE EXTRACTION AND MONOPOLY 37

by a single monopolist in control of both resource extraction and production

of the substitute.
In the special case treated in Section 4, we found that introducing a
competitively supplied substitute would increase the initial price the resource
monopolist would charge, as long as the cost of producing the substitute
exceeds this price.
A market structure with resource extraction fully monopolized and
substitute production perfectly competitive is of course a very special
situation. If our results were only valid for this special market structure, their
significance would be rather limited. However, there is nothing in the
preceding analysis indicating that these results hold only for the market
structure studied. There probably also exist more realistic market structures
giving similar results, for instance some cases in which the market conditions
both of resource extraction and substitute production lie somewhere between
perfect competition and full monopoly, with the resource extraction being
more monopolized, in some sense, than the substitute production.

REFERENCES

1. P. DASGUPTA AND J. E. STIGLITZ, “Uncertainty and the Rate of Extraction under Alter-
native Institutional Arrangements,” Technical Report No. 179, Institute for Mathematic-
al Studies in the Social Sciences, Stanford University, May 1976.
2. G. HEAL, The relationship between price and extraction cost for a resource with a
backstop technology, Bell J. Econ. 7 (1976), 371-378.
3. M. HOEL, “Resource Extraction under some Alternative Market Structures,” Mathe-
matical Systems in Economics 39, Veriag Anton Hain, Meisenheim am Glan, 1978.
4. M. I. KAMIEN AND N. L. SCHWARTZ, A note on resource usage and market structure,
J, Econ. Theory 15 (1977), 394-397.
5. T. R. LEWIS, Monopoly exploitation of an exhaustible resource, J. Environmental
Econ. Management 3 (1976), 198-204.
6. W. D. NORDHAUS, The allocation of energy resources, Brookings Papers Econ. Activity
3 (1973), 529-576.
7. R. M. Sorow AND F. Y. WAN, Extraction costs in the theory of exhaustible resources,
Bell J. Econ. 44 (1976), 359-370.
8. J. E. STIGLITZ, Monopoly and the rate of extraction of exhaustible resources, Amer.
Econ. Rev. 66 (1976), 655-661.
9. J. L. SWEENEY, Economics of depletable resources: Market forces and intertemporal
bias, Rev. Econ. Stud. 44 (1977), 125-142.