AN ECONOMIC THEORY OF PLANNED OBSOLESCENCE

JEREMY BULOW
Planned Obsolescence is the production of goods with uneconomically short
useful lives so that customers will have to make repeat purchases. However,
rational customers will pay for only the present value of the future services of a
product. Therefore, profit maximization seemingly implies producing any given
flow of services as cheaply as possible, with production involving efficient useful
lives. This paper shows why this analysis is incomplete and therefore incorrect.
Monopolists are shown to desire uneconomically short useful lives for their goods.
Oligopolists have the monopolists incentive for short lives as well as a second
incentive that may either increase or decrease their chosen durability. However,
oligopolists can generally gain by colluding to reduce durability and increase
rentals relative to sales. Some evidence is presented that appears to be generally
consistent with the predictions of the theory.

I. INTRODUCTION
Suppliers of darables in imperfectly competitive markets have
been suspected ofproducing goods with uneconomically short useful lives, so that consumers will have to repurchase more often.1
However, the theory behind ~planned obsolescence has been notably weak.2 Will customers not pay less for products that have
a shorter useful life? If the firm decides to sell customers any
given flow ofservices, does profit-maximizing behavior not imply
producing those services as cheaply as possible?3 These are the
questions with which an economic theory ofplanned obsolescence
must deal.

*Thanks to Paul Klemperer, Paul Pfleiderer, John Roberts, Myron Scholes,

and Larry Summers for valuable comments. Thanks also to the participants in
seminars at Chicago, Columbia, Michigan, MIT, Princeton, Stanford, and Wisconsin. This paper was begun at Stanford Business School and revised at the
Center for the Study of the Economy and the State, University of Chicago. I am
grateful to George Stigler for arranging my visit there.
1. See, e.g., Galbraith [1958, p. 352] discussing the wastefulness of chan~ng
automobile models. Probably the best known empirical work in this area is Fis er,
Griliches, and Kaysen [1962].
2. Martin [1962], Kleiman and Ophir [1966],Levhari and Srinivasan [1969],
and Schmalensee [1970] all concluded that monopolists would choose inefficiently
short asset lives, but did so via flawed analyses. Barro [1972]relied on consumers
and producers facing different discount rates to derive planned obsolescence.
3. See Swan [1972, 1977] and Sieper and Swan [1973] for a detailed presentation of this argument. These papers also showed the errors made by the first
several authors cited in footnote 2. For a good survey of the durability literature
until 1980, see Schmalensee [19791.

730

QUARTERLY JOURNAL OF ECONOMICS

This paper explains why, in a full information model4 with

rational customers, a firm might opt to give its products a shorter
than economically desirable useful life. Sometimes firms may even
pay more to produce a shorter lived asset. The key results of the
paper follow:
1. Except under unusual cost conditions5 a monopolist not
threatened by entry will produce goods with inefficiently short
useful lives. This result is closely linked to the observation that
a durable goods monopolist will prefer to rent, rather than sell
its output.6
2. An oligopolist, or equivalently a monopolist facing certain
entry in a subsequent period, also has a countervailing incentive
to extend durability. As a corollary, such firms have an incentive
to steer customers to purchase rather than rental contracts. This
same result also holds if future competition is to be over a related
but not identical substitute product. Therefore, while monopolists
will opt for inefficiently short useful lives, oligopolists may choose
either uneconomically short or long lives, depending on their technologies and market conditions. There is also an incentive to
increase durability to deter entry.7
3. There is generally an incentive for oligopolists to collude
to reduce durability, below noncooperative levels.
4. While antitrust policy requiring firms to sell rather than
rent their products will reduce profitability and any monopoly
power, it may also decrease welfare.
In Section II the basic model of a monopolist choosing a durability is developed. While competitive firms will choose the ef-

4. We mean to rule out equilibria where durability is unobservable prior to

purchase, firms have a cost incentive to produce shoddy products, and customers
therefore assume that low-durability products will be produced.
5. Specifically, a firms marginal cost curve having a steeper slope than its
demand curve around equilibrium.
6. See Bulow [1982] for an exposition on the advantages of renting for such
a monopolist. The link is that by renting the monopolist is selling off the nondurable services of his products and may thus achieve many of the advantages of
low durability without the costs ofinefficient production. However, there are many
durable product markets where rental markets have their own inefficiencies.
Consider, for example, the incentive problems in automobile rental. A good paper
discussing the lease versus buy question in a competitive environment is Wolfson
[1985].
7. There is an extensive literature on increasing the durability of investment
assets to deter entry. The analogous point on the demand side is that entry is less
ong lives.if customers are tied to long-term contracts or already own assets with

731

A THEORY OF PLANNED OBSOLESCENCE

ficient level ofdurability, the monopolists profit-maximizing rate

ofdurability is shown to equal the efficient level plus a term that
is generally negative.
Section III extends the basic model to oligopolies. Olrgopolists
must add an extra term to the monopolists durability choice,
generally causing them (in quantity competition) to decrease the
rate of obsolescence.
Section IV provides illustrations of monopoly and oligopoly
durability choice to indicate how an oligopolist or a monopolist
facing actual or potential entry will change obsolescence from
efficient levels. The monopolist facing entry may choose durability
to effectively become a Stackelberg leader in the post-entry period.
We also find that an oligopolistic industry can gain by colluding
to increase the rate of obsolescence beyond noncooperative levels.
In Section V we examine the strategic consequences to imperfectly competitive firms of renting versus selling. Increasing
the sales-rental ratio is the strategic equivalent of increasing
durability. This analysis suggests that prior to introducing a new
model in an oligopolistic market a firm may choose to sell rather
than rent its old units. This incentive exists even though a greater
sales-rental ratio with the stock of outstanding units unchanged
will lead to a lower sales price. The analysis also predicts that as
a monopolists markets become more competitive, the firm will
increase its sales-rental ratio. This prediction is loosely corroborated empirically.
Section VI summarizes and concludes the paper.

II. MONOPOLY AND PLANNED OBSOLESCENCE

We begin by considering a monopolist who is maximizing
profits over two periods. In the first period it chooses both a quantity q1 and a durability 5. Of the initial q1 units sold, (1

disappear at the end of the first period and Sq1 remain in period
2. The firm may also produce an additional q~ units in period 2.
The implicit rental price of a unit in period 1 is f1(q1) and the
implicit rental price in period 2 is (1 + r)(f2(Sqi + q2)), where r
is the interest rate. Total costs in period 1 are C1(q1,5), and period
2 costs are (1 + r)C2(q2). The present value ofsecond-period revenues is thus f2(5q1 + q2), and the present value of second-period
costs is C2(q2). The firm is required to sell, rather then rent, its

732

QUARTERLY JOURNAL OF ECONOMICS

output.8 Finally, we assume a perfect second-hand market: any

first-period purchaser of one unit can resell his remaining 5 units
in the second period at the market price ofnew output.9
The problem of the monopolist is to
max ~n = q
(1)

1f1(q1) + (Sqi + q2)f2(5q1 + q2)

ql,q2,5

C1(qi,5)

where the first two terms represent the present value of total
revenues and the last two terms are the present value of total
costs.
Left unconstrained, the monopolist faces the following firstorder conditions:
(2)

alT

aC1

=f1+q1f{+5f2(5qi+q2)Sf~-=0
aq1
~9q1

(3)

aq2

(4)

qif2 + (Sq1 + q2)q1f;2

as

aC1
as

Combining the last two conditions yields

1

aC1

C~.
Condition (5) states that the firm will choose an efficient durability: it equates the cost ofmaking its original units a little more
durable so that one more of the original units will still be useful
in the second period, ((lIqD(aC1/aS)), with the present value ofthe

8. We also rule out the monopolists engaging in financial contracts that place
him in the equivalent position of a renter. The simplest such contract would be
one to repurchase all scild items at a fixed price at the end of one perioi Anderson
[1984]suggests futures contracts where the monopolist, by holding futures, could
internalize changes in the general level of prices while making individual owners
responsible for changes in specific asset values due to differing levels of maintenance. One difficulty with such a contract is that it may be difficult for the
monopolist to make its net futures market position publicly observable.
9. This assumption has several implications. Combined with the assumption
of a downward sloping demand curve, it implies that first-period consumers are
made worse off by increased second-period production. Note that sometimes a
monopolist may wish to deter the second-hand market; consider a firm that offers
tied service contracts to customers which are only valid for the original purchaser
of a good. Imperfections in the second-hand market have the effectof reducing
the economic durability of a product (see Bulow [1982]).

A THEORY OF PLANNED OBSOLESCENCE

733

cost of replacing an extra unit in the second period, (Cs). This is

the conclusion of Swan [19721: for any given flow of services the
monopolist chooses to produce, profit maximization implies that
these services be produced as cheaply as possible.
However, this analysis is incomplete and consequently incorrect. The problem of the durable goods monopolist is that the
unconstrained solution to (1) is generally not dynamically consistent.10 When the second period arrives, the monopolist will
regard the first period as water under the bridge. It then faces
the problem,
max q

2f2(8q1 + q2)

q2

the solution of which implies that

(6)

q2f~+f2C~=O.

Note how (6) compares with (3): the difference is that in the
second period when the monopolist produces an extra unit it considers the effect ofthat unit on the price he receives for the other
q2 units sold (and thus sets MR = q2f~ + f2 = MC = C~) but does
not consider the reduction in the rental value of the units previously sold but now owned by others, 5q1f~. Of course, rational
consumers recognize that the monopolist will not consider their
interests in the second period and will adjust the price they are
willing to pay for first-period purchases accordingly. Thus, with
rational consumers the present value of the firms revenue will
equal the present value of the implicit rents its sales generate,
so the firm still wishes 11
to maximize
(1).(1)
It must
dotoso,(6),
however,
Maximizing
subject
by calwith (6) as a constraint.

10. Expositions of the dynamic problem of the durable goods monopolist are
provided by Coase [1972], Bulow [1982], KahnIIl982], and Stokey [1981]. Two
recent papers of interest in this field are Sobel [1984] and Conlisk, Gerstner, and
Sobel [1984].Gul, Sonnenschein, and Wilson [1985]are responsible for two major
advances in the field, being the first to model the dynamic monopoly problem as
a formal game and providing the first general proof of Coases original intuition.
11. Note that if the firm could rent its output instead of selling it, the monopolist could maximize the unconstrained problem. The reason is with rentals
the monopolist owns all of the outstanding units and is thus able to internalize
the capital loss on old units. However, in some markets rental is impractical. For
example, the auto rental market faces a serious problem in monitoringthedamage
caused by renters and therefore cannot force renters to treat the cars as if they
were their own. In some markets (e.g., computers, copiers, and shoe machinery)
the government has required a dominant firm to sell instead of rent some of its
output.

734

QUARTERLY JOURNAL OF ECONOMICS

culating d7r/d~ and recognizing that dq2/d5q1 is implicitly determined by the constraint, leads to the following condition on durability:
1 ac1
qi a5

(7)

C~

5qif~ d(5q1
d~q1
+ q2

The intuition behind the last term in (7) is as follows: because of

the constraint the monopolist suffers a loss of ~q1f!~on the marginal unit it sells in the second period, relative to its precommitment in the first-period output. The term 5q1f~ is simply the
difference between (3) and (6). When 5q1 is increased by one unit
due to increased durability, the choice of q2 is changed by dq2I
d~q1, which can be found by totally differentiating (6). On net,
the number ofunits outstanding in the second period changes by
d(5q1 + q2)/d5q1, which as we will see below is usually positive.
Increasing durability thus has the extra cost in this model of
increasing the units on the market in the second period, and those
extra units reduce profits. There is thus an incentive to reduce
durability below the efficient level.
By totally differentiating (6), we find that

d(5q1 + q2)
d~q1

(8)

2f~ + q2f~

which can be written as

(8)

d(~q1 + q2)
d8q1

slope of demand curve

slope of MC curve
slope of MR curve
slope of MC curve

The denominator of (8) must be negative: around the optimum the slope of the marginal revenue curve must be steeper
than the slope ofthe marginal cost curve. If the demand curve is
more steeply downward sloping than the marginal cost curve,
then the numerator is also negative, and (8) is positive. In this
case an increase in 8q1 caused by increasing durability also increases ~q1 + q2, and because ofthis effect the monopolist chooses
a lower durability, or ~planned obsolescence. If the marginal cost
curve is steeper than the demand curve, an increase in ~qi leads
to a decrease in (~q1 + q2), and 12
theFor
monopolist
therefore
chooses
the remainder
ofthe
paper
to
produce
too
durable
a
product.
we shall assume the ~normal case of the demand curve being
12. For example, if P2 = a
f3~q2 and MC
(I~ + s)q2, then
1(13 e) ~3I(I3
s)~qi, so an increase in Sq 2 = y
= (a. y)
1 of one unitwill decrease

q2

A THEORY OF PLANNED OBSOLESCENCE

735

more steeply downward sloping than the marginal cost curve, in

which case the monopolist unambiguously chooses an insufficiently durable product.13
Thus, the monopolist will generally choose a durability below
efficient levels.14 However, in the unusual case where an increase
in ~q
1 will decrease ~q1 + q~, which is the case when around
equilibrium the marginal cost curve slopes downward more steeply
than the demand curve, the monopolist will produce too durable
a product.
In reality, planned obsolescence is just one of several ways
in which a monopolist might mitigate the commitment problem.
While there is no room for it in our finite time, full information,
one-product model, firms may expend resources to establish a
reputation as suggested by Kreps and Wilson [1982]. They may
do so by developing a pattern for pricing a particular product over
time or through pricing related products that are produced sequentially. (A publisher can gain a reputation for enforcing a
substantial wait between the publication of hardcover and paperback editions, for example.) Also, a monopolists curtailment
ofcapacity will affect cost curves and thus be a way ofcommitting
to limit future outputs. The point of this paper is that as economists we would expect the firm to use all the means at its disposal
to reduce its commitment problem, and the envelope theorem
implies that a monopolist will use at least a little planned obsolescence. The two-period limitation (versus some larger finite
number of shorter periods) does not alter the qualitative impliq2 by F31(13 s) and thus decrease ~q1+ q~ by sI(13 s). Why does this happen?
Consider the experiment where &qi is increased by one, and we wish to discover
whether q~ is decreased by less or more than one. If 8q1 is increased by one and
q~ is decreased by exactly one, second-period marginal revenue,
q2f2(aql + q~) + f2(~q1 + q~), changes by fe, the negative of the slope of the
demand curve. Marginal costs change by C2, the negative of the slope of the
marginal cost curve. From the initial optimum where MR = MC, if the marginal
cost curve is steeper than the demand curve, then increasing 8qi and decreasing
q2 by a like amount will thus leave marginal revenue below marginal costs. So,
to return to MR = MC, q~ will have to be decreased by more than 5qi increased.
Therefore, if marginal costs are sharply decreasing, an increase in durability from
the economically efficient level will cause a decrease in ~ + q~ and thus cause
an increase in profits.
13. This restriction on marginal costs is fairly weak. It is worth emphasizing
that a firmcould have considerable economies ofscale in the ~econd-perioc1(average
costs decreasing with output) without having rapidly decreasing marginal costs.
14. There are also some tax reasons why a monopolist would prefer a less
durable product. By selling a durable product, the monopolist immediately pays
taxes on the present value of the future monopoly rents earned on the units sold.
Further, taxes must be paid on any income earned from reinvesting those profits.
With nondurable units, taxes are paid only at the time the monopolists services
are consumed.

736

QUARTERLY JOURNAL OF ECONOMICS

cations ofthe model and often causes the results to be understatecL

The reason is that a considerable amount of commitment is already implied when a firm that produces in period 1 cannot produce again until half the time horizon has elapsed. With shorter
periods the need to ~buy commitment via planned obsolescence
and other devices can increase.
A final note concerns the welfare implications of this model.
The perfectness constraint (6) reduces the monopolists profits;
that point is one of the major thrusts of the durable goods monopolist literature. But the constraint also causes inefficiencies
in production, such as planned obsolescence. It is entirely possible
that the constrained monopolist, while making low profits, will
reduce discounted consumer surplus relative to an unconstrained
monopolist.5 A simple numerical example in the next footnote
shows how the reduction in monopoly power cum planned obsolescence can be welfare reducing.6 Thus, for example, antitrust
policy requiring firms such as IBM and United Shoe to sell rather
than only rent equipment may well be socially costly even as it
reduces monopoly power.
The implication of small profits and large welfare losses is
similar to the results of textbook monopolistic competition. Because we know that in aggregate firms monopoly profits are relatively small,17 results like these are important for justifying the
substantial research in imperfect competition.
15. Bulow [1982,pp. 32728] shows that a constrained monopolist may build
artificially low capacity as a way of committing to a high future price, and this
in itself can lead to a reduction in discounted consumer surplus relative to the
unconstrained monopolist. While the point there was made in an infinite horizon
model, it is easy to construct two-period examples where the same result applies:
assume that a firm faces a demand curve of p = 100 q for each of two periods
and the interest rate is zero. Capacity costs are $20 per unit, and one unit of
capacity enables a firm to build one unit of output each period at zero marginal
cost. A monopolist renter will build forty units of capacity, and produce forty in
period 1 and ten in period 2. Total profits will be 4,100, and consumers surplus
will be 2,050. A monopolist seller will choose to build twenty-eight units of capacity
in period 1, and produce twenty-eight units in both periods. His profits will sum
to 3,920, and consumers surplus will be 1,960. Thus, the limitation on the monopolists power reduces both profits and consumers surplus.
16. Assume that the inverse demand curve for rental services in each of two
periods is p = 100 q. The interest rate is zero. First-period units cost 20 each
to build, regardless of their durability. Second-period units cost 10 to produce. A
monopolist renter would build forty-five units in period 1 with maximum durability (5 = 1) and rent them out each period. Producers surplus (gross of fixed
costs) would be 4,050, and consumers surplus would be 2,025 for a total of 6,075.
A monopolist seller, by contrast, would produce forty units in period 1 with a
durability 8 = He would add thirty-five new units in period 2, leaving a total
of fifty-five in the market in the second period. Simple arithmetic shows that total
profits drop to 3,725, consumers surplus rises to 2,312k, and thus total surplus
falls to 6,037k.
17.. See, for example, Salinger [1984].

~.

A THEORY OF PLANNED OBSOLESCENCE

737

III. THE OLIGOPOLIsTIC MAXIMIZATION PROBLEM

This section begins by outlining an oligopolists problem for
Cournot quantity competition. The formula for durability has one
extra term compared with the monopolists formula, relating to
the effect ofones durability on competitors second-period output.
With quantity competition a higher durability will usually cause
competitors to cut output, and this increases ones own profitability. Therefore, there is an advantage to high durability that
the oligopolist must trade off with the competing advantage of
faster obsolescence.5
We begin by considering an oligopolist who chooses q, and 5
in the first period and q
2 in the second period. It faces n competitors who will produce a total of ~, with a durability of 5 in the
first period and will produce ~2(5q1)in the second period, where
=
1 q~ and q~ is the second period output of the Cth competitor. The firms maximization problem, not much different from
(1), is
(9)

max q1f1(q1 + ~,) + (Sq, + q2)f2(Sq, + 5~. + q2 +

~2)

qi,8,q~

C,(5,q,)

C2(q2)

subject to
[2 +

q2f~

and subject to
[2 +

0,

where the first constraint is the perfection constraint imposed on

the oligopolist, and the final n constraints recognize that in choosing Sq, the oligopolist is also implicitly choosing his competitors
second-period outputs and that his choice in competitors outputs
is limited by the Nash requirement that each of those firms will
have MR = MC.
Durability choice can be derived just as in the monopoly case:
(10)

1 aC,
=

C~ + 5q,f~

d(Sq, + q2)
d5q,

(Sq1

____

+ ~2)f2~5q1.

18. The first significant paper discussing the oligopoly version of this problem
has just been written by Gul [1985].Using an infinite horizon supergame framework with oligopolistic price competitors, Gul shows that there are multiple equilibria, one of which (as the length of a period to which a firm is committed to a
fix price becomes small) is arbitrarily dose to the monopolistic precommitment
solution! Guls paper, like most ofthe previous literature, assumesno depreciation.

738

QUARTERLY JOURNAL OF ECONOMICS

All the terms in (10) should be familiar from the monopoly

case except the last one. It states that when durability is increased
so that the amount of output that lasts into the next period, 5q1,
increases by one, that competitors second-period output is changed
by d~2/d~q1 and this change affects the price the firm receives for
the 5q1 + q2 units of services that its output provides during the
second period. If, for example, an increase in ~iq1causes competitors to reduce second-period output, then the final term in (10)
will be positive and will act to increase durability, thus acting in
the direction opposite from the previous term in the relation. Note
that if a firm acts as a price taker, (10) shows that it chooses the
efficient durability.
A sufficient condition for the last term in (10) to be positive
and act to increase durability of a firm, say firm A, is that each
ofAs competitors second-period marginal revenue is decreasing
in industry output. Then an increase in As durability, which has
the same effect of increasing second-period industry output, will
9
cause all competitors to contract output and thus help A~s profits.
IV. ILLUSTRATIONS OF THE PROBLEMS OF DURABILITY CHOICE
The analysis of Sections II and III shows that the determinants ofdurability choice extend beyond efficiency considerations.
This section isolates these nonefficiency considerations by pro-

19. The condition in the paragraph above can be guaranteed by two plausible
assumptions. Define the total number of units on the market as
Q 8q~ + q2+8q~ + q~, and define the total value ofimplicit rents in the second
period as TR
Qf2(Q). Then the two assumptions are first that no individual
firm is producing more than ~Qin the second period alone; and second, assume
that a2TR
2/OQ = OMR2/aQ < 0, industry marginal revenueis decreasingin output.
The details are left to the reader. With competition involving some strategic
variable other than quantity, an oligopolist may have a strategic incentive either
to increase durability (as with quantity competition) or to decrease durability.
The crucial determinant is whether an increase in As durability will raise or
lower the marginal profitability of adopting a more aggressive second-period
strategy. For example, with quantity competition As increased durability reduces
the marginal profitability of extra output and thus causes competitors to become
less aggressive by reducing quantity. With price competition an increase in As
durability may cause competitors to charge a lower second-period price; a ~~more
aggressive strategy that has a negative effect on As profits. In the language of
Bulow,Geanakoplos, and Klemperer [1984], the analogy to the last term in (10)
in generalized origopolistic competition will cause A to increase durability if its
competitorsregard As output as a strategic substitute antiwilicause Nto4ecrease
durability if its competitors regard As output as a strategic complement. See also
the excellent paper by Fudenberg and Tirole [1984].

739

A THEORY OF PLANNED OBSOLESCENCE

viding illustrations of durability choice in examples where all

durabilities are equal in productive efficiency.
The assumption that all durabilities are of equal efficiency
in production is of special interest for its application to how a
manufacturer who could either rent or sell its outputs would choose
sales as a percentage of total placements. By decreasing its salesrental ratio, a firm is able to reduce the amount of future output
services owned by customers without any loss of productive efficiency. The analogy of rental-sales strategy to durability choice
is further detailed in Section V.
Both examples below involve quantity competition so that
firms facing potential or actual competition will have an incentive
to choose longer durabilities than a monopolist (see Section III).
The first example illustrates a monopolist who expects competition in the second period. It uses durability as a way to become,
effectively, a Stackelberg leader in the second period. The second
example illustrates quantity competition between symmetrical
firms. The optimal durability choice is derived as a function of
the number of firms in the market, with the monopolists choice
of S = 0 being the result for n 1.
The intuition of the examples provides the basis for a short
discussion on issues of entry and collusion. Finally, the problem
ofa firm planning to introduce a new but related product and its
strategy in choosing durability and a sales versus rental strategy
is analyzed.
Example 1: The Monopolist Facing Second-Period Entry
In these examples we shall assume that firms have a constant
marginal cost technology that allows them to produce units with
a durability of S at a cost of C(1 + 5). In the second period they
can produce units lasting one period at a constant marginal cost
of (1 + r)C. With this technology producing at all durabilities is
equally efficient. That is, the present value of total costs is equal
to q1C + (Sq1 + q2)C regardless of whether the firm chooses to
supply second-period demand with high durability and low q2 or
low durability and high q2.
In this first example assume that a first-period monopolist
chooses q1 and 5. In the second period the incumbent and an
entrant establish a Cournot equilibrium over the residual demand. The demand curve for rental. services is p = a 13q1 in
period 1 and p = (1 r)(cx
pSq1 ~3q~ Pq2) in period 2.
Because the second-period Cournot equilibrium is q2

740
=

QUARTERLY JOURNAL OF ECONOMICS

(a

135q1

0)1313, the monopolists first-period problem can

be thought of as
(11)
max q1(a

13q1

C) + (Sq1 + q2)(a

p(5q1

q2

+ ~2)

C)

qi,8

subject to
=

(a

135q1

C)/313.

(Because of the symmetry in the position of the incumbent

and the entrant in choosing second-period output, the monopolists
perfection constraint of second-period MR = MC (a
13q2
213q2
c = 0) and the Nash equilibrium requirement for the
competitors second-period output (a
135q1
13q2
213~2
c
= 0) can be simplified to the constraints above.) Solving (11) yields

r35q1

(a

C)1213,
=

Sq1
q~

(a

C)/4I3~5
C)/413.

(a

So the optimum depreciation rate, if all technologies are equally

efficient, is 1
S = 50 percent. Presumably if first-period technologies were differentially efficient, the monopolist would choose
a durability somewhere between 50 percent and the maximally
efficient level.
It is worth noting that in this problem the monopolist is able
to produce the monopoly quantity in the first period and by choosing its durability achieve the Stackelberg leader position in the
second period, where Sq~ + q2
(a
C)1213 and q2 = (a
C)!
413.20 This is a simple illustration of the tradeoffs in durability
where increased durability implies a lower second-period price
but a bigger market share for the incumbent.

Example 2: Symmetric Oligopoly

Using the same setup as in Example 1, we solve for the symmetric perfect Cournot-Nash equilibrium in a game with n firms
producing in each oftwo periods. We look at the problem ofa firm
choosing q1 and S in the first period and q2 in the second, knowing
that its (n
1) competitors will each produce ~ units with a

20. This Stackelberg result can be derived under reasonably general conditions.

A THEORY OF PLANNED OBSOLESCENCE

741

durability of ~ in period 1 and that a Cournot-Nash equilibrium

will be established on residual demand in the second period. The
firms problem is to
(12)
max q1(a

13q,

13(n

1)~~

C) + (Sq, + q2)(a

13(Sq,

qi,8

+ S(n

1)~, + q2 + (n

C)

1)~2)

C)/13(n + 1),

subject to
=

(a

13(Sq, + (n

1)S~,)

where the constraint is again implied by the second-period equilibrium. As with (11) symmetry enables us to combine the firms
perfection constraints on its own output with the constraints that
it is limited in its implicit choice of competitors second-period
output to their meeting the Nash equilibrium conditions. It is
easy to solve (12) for a symmetrical equilibrium:
=

aC
13(n + 1)
Sq1 =
aC
=
= 13(n2+ 1)

2+1)
(aC)(n1
13(n
n21

5n2+f

For a monopolist (n = 1) the optimal value ofS is zero: there

is a disadvantage to durability in that it causes low future prices;
there is no corresponding advantage to a monopolist in causing
competitors to cutback on future output.2 In this example S = 0.6
for a duopoly and approaches 1 as n increases. While oligopolists
do have a durability tradeoff, this result does not imply that with
technologies of differential efficiency that more and more firms
in the industry will lead to more and more excessive durability.
As more firms enter, profit margins decrease, and firms cannot
afford to stray so far from efficient production techniques.
21. Note that the optimal value for a monopolists ~ is zero; the firm would
not choose a negative ~ even if possible. What, economically, is a negative depreciation rate? Barry Nalebuff suggests that it has a role in a model of addiction:
for given 5, the more units sold in the first period, the greater the demand in the
second period. Ifour friendly monopolist were selling heroin, then it would prefer
that the product be neither satiating nor nonaddictive: intuitively, those who do
buy in the first period will be hooked and pay a big second-period price for the
product, but this will not compensate for the customers who never try the product
because of its addictive quality.

742

QUARTERLY JOURNAL OF ECONOMICS

Entry Deterrence and Collusion

Inspection of the examples also provides intuition about issues of entry deterrence and oligopolistic collusion. Increasing
durability reduces the demand for new units and thus reduces
the profitability of all firms including a new entrant in period 2.
In Example 1, the monopolist might choose a durability in excess
of0.5 if there are fixed costs ofentry and the increased durability
will prevent entry.22
Example 2 can be used to show the advantage to oligopolists
of colluding on the durability of their products. It is easy to ascertain that the higher the durability ofperiod 1 output, the lower
the equilibrium price in period 2. In this example if firms were
constrained to choosing a lower level of durability in period 1,
they would end up (weakly) increasing q
1 relative to what they
would choose at the Nash durability and reducing Sq1 + q2. For
example, in a collusive agreement to set S = 0, the firms would
all set q1 = q2 = (a
C)I(n + 1)~ and earn Cournot-Nash profits
for each of two periods. Formally, the maximization problem becomes the same as (12) with the added constraint that 5 5C, the
collusive 523 In the example, industry profits are increased, the
lower the collusive durability. Intuitively, without collusion firms
choose increased durability up to the point where the total derivative oftheir own profits with respect to durability is zero. But
at this point incremental durability, effectively increasing firms
second-period quantity, reduces competitors profits. So if it is
possible to collude on the rate of obsolescence but not on price,
firms will be able to increase industry profits by reducing durability from noncooperative levels.
If an oligopoly such as the American automobile industry
were colluding on durability, then, it would likely be in the direction of planned obsolescence. The entry offoreign competitors
to make the industry more competitive would then move durability toward efficient levels.

22. If the fixed costs also must be paid by the monopolist for operating in the
second period, the monopolist may even be better off. A high enough durability
may effectively permit the monopolist to precommit to not producing in period 2,
raising the price received in period 1 and the present2(n2
value
+ 1)].
of discounted
Then if~isfixed,
profits.
each23.
firmin
LetX~
Example
[(n +2 1)2
would
+ 8(n
choose1)]I[(n
q~ X(a
+ 1)~ C)113
+ ~ and q~ = (1
n~X)(a
r3(n + 1). Industry profits over the two periods will be H~ + 112 = (a c)21
13(nX(1 nX) + nI(n + 1)2(1 + 8X)(1 n8X)) whichturns out to be decreasing
in 8 over the range 0 ~ 8 ~ 1.

A THEORY OF PLANNED OBSOLESCENCE

743

V. SALES VERSUS RENTALS

Finally, we consider how a firm that has the choice between
selling and renting its output should balance its placements. Again
a simple modification of the two illustrative examples is of use.
Assume that all units produced in the first period last two periods
and cost 2C to produce. Units built in the second period last one
period and cost (1 + r)C to produce. The firms must choose firstperiod outputs, and decide what fraction of those units should be
sold and what fraction rented. Then firms will choose q1 just as
in the examples and choose
salesis,percentage
as monopolist
they chose
24 aThat
in Examplejust
1 the
S in the
will
sell original
half its examples.
output and rent half. In Example 2, each oligopolist will sell (n2 1)I(n2 + 1) of its first-period output. The
reason for the close analogy between a rental-sales ratio and a
depreciation rate is that when a firm sells S units with a life of
two periods and rents (1 5) units, it essentially selling one unit
of output for the current period and S units for the following
period, just as if it sold a depreciating asset. The sole difference
is that the production and rental ofdurables gives a firm marginal
costs ofzero in the second period until all ofthe first-period rentals
are disposed of. Because examples 1 and 2 give firms constant
marginal costs and firms always distribute more units in period
2 than they rented in period 1, this difference does not affect the
equilibrium in those examples.25
The analogy of rental-sales strategies to durability enables
one to make predictions about the evolution of a firms rentalsales ratio as the degree of competition it faces changes. A mo

24. This exact equivalence would not hold in other examples where the equilibrium number of units on the market in the second period would be less than
the number in the first period, for example because second-period demand was
weaker than first-period demand.
25. More formally, assume that O2C
2C
1/aq~ = a 2Iaq~ = Oso that marginal costs
are constant each period and
1 aC2 aC2
q1a8
0q2
so that there are no economies or diseconomies associated with a change in durability choice. Then the costs in the oligopolistic maximization problem~
Ci(8,qi) + C2(q2), can be rewritten as C(qi,8q1 + q~). In that case the maximization problem is precisely the same for a firm that faces an exogenous 8 and
chooses a sales-rental ratio and one that can choose 8 but can only sell, as long
as we have an interior solution (no units are rented in the first period and left
idle in the second). The general interpretation of 8 would be the fraction of firstperiod placements that are both serviceable in the second period and owned by
customers.

744

QUARTERLY JOURNAL OF ECONOMICS

nopolist has an incentive to rent, proving to customers that it will

not reduce prices too much in the future. By contrast, a firm that
either faces or will shortly face some competitors in its current
product line or some closely related new products will also have
a contrasting incentive to sell. The sales, like greater durability,
free the firm to place more new units in the future and may thus
improve the firms competitive position. Additionally, the smaller
a firms market share the more elastic its demand for any given
market elasticity. With price less sensitive to any given percentage change in the firms quantity, there is less incentive to rent
rather than sell to commit to low quantity.
To check the prediction that as a dominant firm encounters
more competition it will increase its sales-rental ratio, I gathered
all publicly available data on the breakdown ofrevenues between
sales and rentals for both IBM and Xerox. As Table I shows, both
firms have substantially increased their sales-rental ratios since
1970.
Unfortunately, there are at least five caveats that reduce the
TABLE I
BREAKDOWN OF REVENUES,

IBM
Percentage
Sales Rentals
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
1969
1968
1967
1966

57.9
48.9
44.4
41.7
41.4
41.5
39.1
36.6
31.5
33.8
30.7
30.2
26.4
27.0
35.8
41.8
35.0
31.6

Services

23.0
32.4
37.3
41.5
44.0
46.4

19.1
18.7
18.3
16.9
14.5
12.1
60.9
63.4
68.5
66.2
69.3
69.8
73.6
73.0
64.2
58.2
65.0
68.4

19661983
Xerox
Percentage
Sales Rentals
45.9
41.2
40.1
37.9
34.2
31.3
26.9
22.8
20.0
20.1
19.1
20.1
20.3
21.8
26.2
21.8
N.A.
N.A.

Services

39.1
47.0
51.0

15.1
11.8
8.9
62.1
65.8
68.7
73.1
77.2
80.0
79.9
80.9
79.9
79.7
78.2
73.8
78.2
N.A.
N.A.

N.A. = not available.

Source. 10-K Reports filed by IBM and Xerox with Securities and Exchange Commission, various years.

A THEORY OF PLANNED OBSOLESCENCE

745

value ofthis evidence in confirming the theory. First, and perhaps

most important, conventional price discrimination theory has ar-

746

QUARTERLY JOURNAL OF ECONOMICS

ductions has been dramatically increased to the point where the

tax advantage to manufacturer leasing has been virtually eliminated. This change in the tax laws has probably affected firms
sales-rental policies.28
Nevertheless, analysis in the business press has argued that
IBMs recent policy of increasing sales versus rentals has been
implemented because ofcompetitive considerations. The IBM Credit
Corp. was established two years ago to facilitate its customers
outright purchase ofmachinery, altering the companys historical
policy of encouraging customers to lease rather than buy its machines. As a result the fraction ofcompany revenues derivedfrom
rentals has dropped by over a third in two years.29 ~More important, the trend to customers purchasing more oftheir equipment
frees the company to introduce machines more frequentlya lethal competitive edge.30
VI. CONCLUSION
Monopolists generally have an incentive to produce goods
with inefficiently short useful lives. There are only minor exceptions to this rule. The reason for this rule derives from the perfection constraint of the durable goods monopolist, which forces
him in the long run to charge lower prices than would an unconstrained firm. By reducing the durability of its output, the monopolist can reduce the cost of its perfection constraint.
An oligopolist, or a monopolist facing future entry, has the
same considerations as the monopolist plus the extra consideration ofhow its durability will effect competitors future strategies.
If the firm faces Cournot-Nash competition, it will usually find
that increased durability will reduce competitors future output.
Since a reduction in competitors output will, all else equal, raise
a firms profits, such oligopolists have a countervailing incentive
to increase durability and may choose either excessively long or
short lives for their products.
28. For a lucid summary of these issues see Miller and Upton [19761.Of
course, the 1981 tax law also created incentives for some firms, most notably
Boeing, to lease rather than sell. The reason is that when equipment such as an
airliner was transferred abroad but used partially in the United States, the 1981
law allowed the lessor a 10 percent investment tax credit and an accelerated
depreciation writeoff. See Merry ~1983].This tax advantage has since been eliminated.
29. See Table I.
30. See Andrew Popper, How the IBM Juggernaut Will Keep Rolling, Business Week, July 16, 1984, p. 106.

A THEORY OF PLANNED OBSOLESCENCE

747

If a firm wishes to deter entry into its markets, then it will

prefer longer durabilities, orequivalently it will have apreference
to sell rather than rent more of its output. On the other hand, if
oligopolists can collude to set the durability ofthe industrys products, they will opt for some planned obsolescence. Antitrust policy
that requires durable goods monopolists to sell rather than rent
their products will reduce their profitability but may also reduce
welfare. Thus, while such policies reduce monopoly power, they
will not necessarily add to efficiency.
The movements of IBM and Xerox to a greater reliance on
sales rather than rentals as they have encountered greater competition are loosely consistent with some ofthe predictions ofthis
theory.
Perhaps the greatest weakness ofthis paper is that it follows
in the tradition of using durability as a proxy for obsolescence.
This assumption, combined with the perfect second-hand market
assumption, permits the model to regard goods produced at different times as homogeneous, and greatly simplifies the analysis.
But planned obsolescence is much more than a matter of durability; it is also and perhaps primarily about how often a firm will
introduce a new product, and how compatible the new product
will be with older versions. In a market such as textbooks, the
publishers inability to internalize the capital losses suffered by
holders of used books seems to lead to excessive efforts to make
succeeding editions of a book incompatible with one another; this
result seems consistent with our theory.3 But in other markets,
such as that for personal computers, customers may value their
purchases more highly the greater their expectations of future
sales. Such markets are not well accommodated by the model
presented here. However, modeling such markets will necessarily
require abandonment ofthe perfect second-hand market assumption, because with perfect second-hand markets and downward
sloping demand curves first-period purchasers must necessarily
lose by increased second-period sales. The inability ofcustomers
to resell will introduce considerations akin to those of conventional price discrimination into the analysis. This complication,
combined with the difficulty in allowing for product differentia-

31. John Kaplan tells the story of one of his Stanford law students asking if
he could get by with the previous edition of Kaplans textbook in his course. The
author responded to the student, If an intelligent person is revising his textbook,
do you think hes going to redo it in such a way that you can use the old version?

748

QUARTERLY JOURNAL OF ECONOMICS

tion in an enlightening way, means that further advances in this

field will necessarily require models that are more complex than
this one.
These issues mean that planned obsolescence is still adifficult
and poorly understood topic. It is my hope, however, that the

A THEORY OF PLANNED OBSOLESCENCE

749

Miller, Merton H. and Charles W. Upton, Leasing, Buying and the Cost of Capital
Services, Journal ofFinance, XXXI (June 1976) 78798.
Salinger, Michael A., Tobins qs q, Unionization, and the Concentration-Profits
Relationship, Rand Journal of Economics, XV (Summer 1984), 15970.
Schmalensee, Richard, Regulation and the Durability of Goods, Bell Journal of
Economic Goods and Management Science, I (Spring 1970), 5464.
,Market Structure, Durability, and Quality: A Selective Survey, Economic
Inquiry, XVII (April 1979), 17796.
Sieper, E., and Peter L. Swan, Monopoly and Competition in the Market for
Durable Goods, Review ofEconomic Studies, XL (July 1973), 33351
Sobel, Joel, The Timing of Sales, Review ofEconomic Studies, LI (April 1984),
35368.
Stokey, N. L., Rational Expectations and Durable Goods Pricing, Bell Journal
of Economics, XII (Spring 1981), 11228.
Swan, Peter L., Optimum Durability, Second-Hand Markets, and Planned Obsolescence, Journal of PoliticalEconomy, LXXX (May/June 1972), 57585.
Product Durability Under Monopoly and Competition: Comment, Econometrica, XLV (Jan. 1977), 22935.
Wolfson, Mark A., Tax, Incentive, and Risk-Sharing Issues in the Allocation of
Property Rights: The Generalized Lease-or-Buy Problem, Journal ofBusiness, LVIII A ril 1985) 15972.