Durable Good Theory

American Economic Association
Durable Goods Theory for Real World Markets

Author(s): Michael Waldman
Source: The Journal of Economic Perspectives, Vol. 17, No. 1 (Winter, 2003), pp. 131-154
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/3216843
Accessed: 27/07/2010 07:12
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=aea.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
Journal of Economic Perspectives.
http://www.jstor.org
Journal of Economic Perspectives?Volume 17, Number 1?Winter 2003?Pages 131-154
Durable Goods for Real World

Theory
Markets
Michael Waldman
goods constitute an important part of economic production.

In 2000, personal consumption expenditures on durables exceeded
Durable $800 billion. In the manufacturing sector in the United States in the year
2000, durable goods production constituted roughly 60 percent of aggregate
production.
Durable goods pose a number of questions for microeconomic analysis. One
set of questions involves durability choice and the related issue of "planned obso-
lescence." For example, do firms have an incentive to reduce durability below the
efficient level so that units break down quickly? Also, to what extent do firms have
an incentive to introduce
new products that make old units obsolete? A second set
of questions revolves around timing issues. How are current prices and marketing
strategies affected by a producer's actions tomorrow that affect the future value of
units the producer sells today? A third set of issues revolves around information
asymmetry. In many durable goods markets, buyers are unable to judge the quality
of durable units offered for sale. As a result, the problem of adverse selection can
arise, where sellers withdraw high-quality units from the market because consumers
are unable to perceive high quality and are thus unwilling to pay a high price
for it.
The early 1970s witnessed three
major advances in the microeconomic theory
of durable goods. Oddly enough, one of these was mainly followed up in the 1970s,
one was largely pursued in the 1980s, and the other was investigated only in the
1990s. In a series of papers in the early 1970s, Peter Swan considered the question
of optimal durability (Swan, 1970, 1971; Sieper and Swan, 1973). Most of the work
? Michael Waldmanis the Charles H. Dyson Professor in Management and Professor of

Economics, Johnson Graduate School of Management, Cornell University, Ithaca, New York.
His e-mail address is (mw46@cornell.edu).
132 Journal of Economic Perspectives
in the 1970s on durable

goods theory focused on the generality of Swan's results.
A second major advance in the early 1970s, by Ronald Coase (1972), involved time
inconsistency. The focus here is on difficulties that arise because durable units sold
tomorrow affect tomorrow's value of units sold today, with the result that the
producer's sequence of outputs may not maximize the firm's overall profitability.
Much of the work on durable goods theory in the 1980s focused on Coase's
insights. The third contribution was George Akerlof's (1970) analysis of asymmetric
information and adverse selection. This work was not originally perceived as being
especially relevant to durable goods theory, but after all, Akerlof's primary appli?
cation was the used car market. By the late 1990s, these insights were being
incorporated into models of durable goods.
Each of these advances identified an important issue concerning durable
goods markets. But each also considered simplified models that gave very incom?
plete pictures of most actual durable goods markets. For example, Swan (1970,
1971) and Sieper and Swan (1973) made unrealistic assumptions concerning
substitutability between new and used units. Coase (1972) assumed both output
and price are not contractible. Akerlof (1970) focused on secondhand markets and
ignored the market for new units. Since many of the extensions of these analyses
maintained thesestrong assumptions, our
understanding of certain classes of
theoretical modelswas significantly advanced, but our understanding of real world
markets for durable goods was not advanced as far.
The situation has changed over the last ten years. A number of authors,
including Igal Hendel, Alessandro Lizzeri and me, have moved beyond those initial
fundamental analyses and considered more realistic models of durable goods
markets. As a result, we now have a clearer understanding of a variety of phenom?
ena concerning durable goods markets. For example, in contrast to Swan's (1980)
initial thinking on the subject?but consistent with what publishers actually say (in
private)?one role played by the introduction of a new edition of a textbook is
indeed to "kill off" the market for used textbooks. Another example is a better
understanding of the role (and recent
growth) of leasing in the market for new
cars, which a number of authors have argued is a response to adverse selection in
the market for used cars. A third example is a better understanding of why firms
sometimes monopolize aftermarkets for their own products, as has been alleged in
court cases involving firms such as Kodak, Data General, Unisys and Xerox. Finally,
we now better understand the upgrade process that is common in the software
industry and, in particular, a tool used by Microsoft. In this paper, I discuss the
intellectual journey of the microeconomic theory of durable goods over the last
30 years, with an emphasis on recent contributions.
Before proceeding, it is worth
noting that much of the following discussion
concerns monopoly producers of durable goods selling to consumers. This should
not be interpreted to mean that the insights from the analyses discussed only apply
to such settings. Even though most durable goods producers are not monopolists,
most do have market power, and monopoly analyses should provide useful insights.
Similarly, the problems faced by durable goods producers selling to firms should be
Michael Waldman 133
similar to problems faced by durable goods producers selling to consumers, and

thus the analyses that follow should help us better understand the issues faced by
the producers of durable intermediate inputs.
Three Classic Contributions to Durable Goods Theory
The
early 1970s saw three classic contributions to durable goods theory: an
analysis of optimal durability by Swan (1970, 1971) and Sieper and Swan (1973); an
analysis of time inconsistency by Coase (1972); and the analysis of adverse selection
by Akerlof (1970).
Optimal Durability
In the late 1960s, a number of authors considered whether a durable goods
monopolist would choose the same
durability as competitive producers or, simi?
larly, whether such a firm would choose the socially optimal level of durability. The
general theme of this literature was that a monopolist would choose less durability
than would competitive producers or, similarly, an inefficiently low level of dura?
bility (for example, Kleiman and Ophir, 1966; Levhari and Srinivasan, 1969;
Schmalensee, 1970). In a series of papers in the early 1970s, however, Swan showed
this conclusion was incorrect (Swan, 1970, 1971; Sieper and Swan, 1973).
To understand Swan's insight, think of the case where a unit provides the same
stream of services for the first T periods and then becomes useless. The choice of
durability is then simply the choice of T. A real world product with roughly this
feature is light bulbs. Thus, think of a monopolist that sells light bulbs to consum?
ers, and consider the monopolist's choices of output and durability from a steady-
state perspective. If consumers only care about the amount of light consumed and
not directly about how long a bulb lasts, then the monopolist will choose output
and durability such that the light provided is produced at minimum cost. In other
words, although the monopolist produces less than the socially optimal level of
light because of standard monopoly incentives to produce less than the efficient
amount, there is no distortion in terms of durability since the monopolist will
produce at minimum cost. But this is the same as saving that (taking as fixed the
amount of light produced) the monopolist chooses the socially efficient durability
level. Further, in many cases, this also means the monopolist chooses the same
durability as a competitive industry would choose.
Swan did not just show this result for situations in which a product provides the
same stream of services until it wears out, but rather showed it holds given the
common assumption dating back to Wicksell
(1934) that "service flow is propor?
tional to the stock on hand." To understand
what this means, consider a monop?
olist who has two choices. The monopolist can produce units that depreciate so that
after one period they are equivalent of half of a new unit and after two periods are
worthless. Alternatively, the firm can produce units that after one period are
equivalent of a quarter of a new unit and after two periods are worthless. From the
standpoint of the steady state, the service stream provided per period is the same
whether the monopolist produces ten of the former type of unit per period or
twelve of the latter. Swan showed that, given this assumption that some number of
used units is a perfect substitute for a new unit, the same logic as above holds, and
thus, the monopolist chooses the durability level that produces the desired service
stream at minimum cost.
The 1970s saw numerous investigations of the robustness of Swan's conclusions
to relaxation of the specific assumptions he employed. For example, Barro (1972)
allowed consumers and the firm to have different discount rates, Schmalensee
(1974) and Su (1975) introduced a maintenance decision, and Auernheimer and
Saving (1977) introduced short-run nonconstant returns to scale. Schmalensee
(1979) surveys Swan's
papers and the literature that followed. The conclusion of
this literature, not surprisingly, is that Swan's findings are robust to relaxing some
of his assumptions, but not others.
However, one of the assumptions that received limited attention in this liter?
ature is that some number of used units is a perfect substitute for a new unit. Later,
I argue that this assumption is unrealistic for most durable products, and a more
realistic assumption that captures imperfect substitutability between new and used
units yields significantly different conclusions.
Time Inconsistency and Durable Goods

Coase (1972) argued that a durable goods monopolist faces a problem of time
inconsistency. The problem arises because durable goods sold in the future affect
the future value of units sold today, and in the absence of the ability to commit, the
monopolist does not internalize this externality.
To understand Coase's
argument, it is easier to start with the later analysis of
Bulow (1982). considers
Bulow a monopolist of a perfectly durable good who sells
output in each of two periods. In this model, if the firm in the first period can
commit to a production level for the second period, then the firm's profit-
maximizing first-period choices are to commit first to selling zero in the second
period and then to produce the monopoly output level and sell it at the monopoly
price. But what if commitment is not possible? Then, if the firm tries to sell the
monopoly output in the first period, consumers will be unwilling to pay the
monopoly price. The reason is that consumers fear thefirm will collect the
monopoly price in the first period and then produce additional units in the second
period, thus driving down the second-period value of the used units that the
consumers own. The result is each consumer's willingness to pay in the first period
is reduced, and overall
monopoly profitability falls. Bulow further shows that the
monopolist can avoid this problem by leasing as opposed to selling. The logic here
is that leasing means the monopolist owns the used units at the beginning of the
second period, and thus the monopolist internalizes how its second-period actions
affect the value of those units.
Looking back, it seems a bit surprising (at least to me) that Coase in his 1972
paper did not consider a relatively simple case like the one considered by Bulow
Durable Goods Theory for Real World Markets 135
(1982). Rather, Coase considered the more

technically challenging case of a
durable goods monopolist who sells output in an infinite-period setting and cannot
commit to future production levels. Coase does not analyze the problem formally
but rather conjectures that, given rational consumer expectations, the monopolist's
price will fall immediately to marginal cost. (Coase also argued, as affirmed by
Bulow, that leasing would avoid the problem.) Much of durable goods theory in the
1980s focused on whether this conjecture was correct (for example, Stokey, 1981;
Gul, Sonnenschein and Wilson, 1986; Ausubel and Deneckere, 1989).1 The main
result is that Coase is correct if buyers' strategies do not depend on past behavior,
so that reputation formation by the seller is not possible. However, as Ausubel and
Deneckere show, if reputation formation is possible and marginal cost is equal to
or above the value the lowest valuation consumer places on the product, then
equilibria exist in which the monopolist regains someor all of its monopoly power.
Much of the rest of durable goods theory in the 1980s and early 1990s focused
either on the robustness of Coase's
conjecture to changes in the assumptions or on
tactics other than
leasing that a durable goods monopolist can use to avoid the
problem. On the former topic, Bond and Samuelson (1984) show that depreciation
and replacement sales reduce the monopolist's tendency to cut price, Kahn (1986)
shows a similar result when an upward-sloping marginal cost schedule is substituted
for the assumption of constant marginal costs, while Bagnoli, Salant and Swierzbin-
ski (1989) and Levine and Pesendorfer (1995) consider what happens when output
comes in discrete units. On the latter topic, Bulow (1986) extends his earlier
analysis by showing that a durable goods monopolist can reduce its time inconsis?
tency problem by reducing the durability of its output (this analysis is thus related
to the discussion of the previous subsection), Butz (1990) shows that contractual
provisions such as best-price provisions or most-favored-customer clauses can ame-
liorate the problem, while Karp and Perloff (1996) and Kutsoati and Zabojnik
(2001) consider whether a monopolist can reduce the problem by initially using an
inferior high-cost technology.
As I look back over the literatureon Coase's time inconsistency problem, I feel
its focus was misplaced to some extent. Most of this literature assumes commitment
about future prices and quantities is not possible, but in fact, consistent with Butz's
analysis, commitments on future prices and quantities are common. Think of a firm
that sells commemorative coins. It sells a good that is basically perfectly durable,
and it can sell the good today or at any later date. But in contrast to Coase's
conjecture, the price does not fall immediately to cost. One important reason is
that the firm can contractually commit to an upper bound on future quantities
through what is called a limited edition. In other words, Coase's argument is
important for understanding a variety of contractual provisions that allow firms
1A
closely related literature is the literature on bargaining with one-sided uncertainty where only the
uninformed party makes offers. Important papers in this literature are Sobel and Takahashi (1983),
Fudenberg, Levine and Tirole (1985) and Ausubel and Deneckere (1989). Ausubel and Deneckere's
paper includes a discussion of the relationship between the two literatures.
to avoid the time

inconsistency problem that he identified, but analyses that
assume no commitment at all is possible miss many of the important real world
implications.
In addition, Coase's argument is also important because the implications of
time inconsistency are much broader than the specific problem he identified.
Rather than time inconsistency only applying to future output, it applies to any
future action by the firm?such as an R&D decision or the choice of a repurchase
price?that affects the future value of used units. Hence, as discussed in later
sections, Coase's insight has implications for durable goods markets that are far
from his original discussion.
Adverse Selection and Durable Goods

Akerlof's (1970) analysis of adverse selection is well known as one of the
papers that launched the vast literature on the role of asymmetric information in
markets. Until recently, however, the implications of Akerlof's argument for dura?
ble goods markets were pretty much ignored. But Akerlof's main example was the
used car market, so clearly the argument is potentially important for durable goods
markets.2
Akerlof's (1970) model can be thought of as a supply-demand problem. The
suppliers are individuals, each of whom initially owns one used car and each knows
the quality of that used car. The demanders are individuals, each of whom initially
owns no used car and does not know the quality of any specific used car, although
these individuals do know the average quality of used cars offered for sale. Akerlof
assumes that demanders have a higher value for used cars than
do suppliers, so
under full information, all used cars are traded. Akerlof shows that, given the
asymmetry of information assumed, not all used cars are traded, so trade is less than
the efficient amount.3 The reason is that the used car price reflects the average
quality of used cars offered for sale, so a supplier with a high-quality used car keeps
it rather than sell it at a price that does not reflect its quality.
In addition to predicting less than the efficient level of trade in used car
markets, this perspective explains an interesting aspect of those markets. Although
many used cars are traded through a middleman, there are also many trades
between friends, relatives and acquaintances. In a world of full information, such
trades would be puzzling, because used car trades should move the cars to their
2 There is a small
empirical literature that looks for evidence of adverse selection in durable goods
markets. Bond (1982,1984) investigates the market for used pickup trucks and finds evidence of adverse
selection in the market for older trucks only, while Genesove (1993) finds some evidence for adverse
selection in dealer auction markets for used cars. More recently, Porter and Sattler (1999) find evidence
inconsistent with adverse selection in a study of the automobile market, while Gilligan (2002) and
Emons and Sheldon (2002) find evidence in favor of adverse selection?the first in a study of business
aircraft and the second in a study of automobiles.
3 Akerlof
(1970) considers a specification in which under asymmetric information there is no trade at
all. But this is neither required for the argument nor a general feature of the type of model he considers
(Wilson, 1980).
Michael Waldman 137
highest-value use, and there is little reason to believe that when someone wants to
sell a used car, the new highest-value user would be a friend, relative or acquain-
tance. Given asymmetric information and adverse selection, however, this feature of
used car marketsis not at all surprising. As discussed above, in the absence of
information concerning used car quality on the part of buyers, only low-quality used
cars are traded, and prices reflect the average quality of used cars offered for sale.
But this means owners of high-quality used cars prefer to trade (only trade) with
buyers who know the quality of the seller's used car, which is most likely when the
buyer is a friend, relative or acquaintance.
Although Akerlof's (1970) analysis has had a huge impact on economics as a
whole, the analysis initially had little impact on durable goods theory. However, this
has changed recently as a number of authors have started to incorporate this
argument into rich models of new and used durable goods.
Three Recent Contributions
Recent contributions to durable goods theory have served to extend and

refocus the theory so that it is more applicable to real world markets. I focus here
on recent contributions in three areas: durability choice, introductions of new
products and applying adverse section to a full model of new and used durable
goods.
Durability Choice in Real World Markets

Most of the literature concerning durability choice has followed Swan's lead
and assumed that service flow is proportional to stock on hand. Sieper and Swan
(1973, p. 334) describe the assumption this way:
In common with Wicksell (1934) and most subsequent treatments of product

durability, we shall assume that there exists a definition of product services
such that units of service are perfect substitutes in consumption (or produc?
tion) irrespective of the age or the durability of the product from which they
are derived. The power of this undoubtedly heroic abstraction is that it allows
a multiplicity of distinct products, classified by age and durability, to be
analyzed in terms of a single market for product services.
This approach has two problems. First, for most durable goods, it is not at all
realistic. For products such as automobiles, televisions, refrigerators and toasters, a
consumer cannot combine "service units" derived from a number of used units to
create a perfect substitute for a new unit. Second, since with this assumption all
consumers place the same relative value on new and used units, it allows no
important role for secondhand markets.
An alternative approach employed in Waldman (1996a) and Hendel and
Lizzeri (1999a) is to assume new and used units are imperfect substitutes that vary
in terms of quality, where durability is modeled as the speed with which the quality
of a unit deteriorates (see also Kim, 1989; Anderson and Ginsburgh, 1994). These
models also assume consumers who vary in their valuations for quality and a
secondhand market where used units are traded. that, both here and in
Note
papers discussed later, one reason for assuming that
consumers vary in their
valuations for quality is that it creates a role for secondhand markets. As units age
and fall in quality, the units are traded from high-valuation to low-valuation
consumers.
The main conclusion of these analyses is that a durable goods monopolist
typically underinvests in durability, with the result that the quality of used units is
less than the efficient level.4 The logic is that because new and used units are
substitutes?albeit imperfect ones?the price of a used unit on the secondhand
market the monopolist
constrains in terms of the price it charges for new units. By
reducing durability below the efficient level and thus the quality of used units below
that level, the monopolist reduces the substitutability between new and used units,
which, in turn, allows the firm to increase the price of new units.5
As explained in detail in Waldman (1996a), these analyses are similar to the
classic analyses of Mussa and (1978) and Maskin and Riley (1984) of a
Rosen
monopolist who sells a product line of different qualities to consumers who vary in
their valuations for quality. In those papers, the monopolist reduces the quality sold
to low-valuation consumers below the socially optimal level, because this allows the
firm to charge more for high-quality units sold to the high-valuation consumers. In
the durable goods problem, one can think of the monopolist as producing a
product line time, where new units are the high-quality
over units while the
different vintages of used units are the lower qualities in the product line. Thus,
reducing durability and in turn the quality of used units in my analysis and Hendel
and Lizzeri's is similar to the finding in the product line pricing literature that a
monopolist reduces the quality sold to low-valuation consumers.
The above logic has implications beyond durability choice. The general point
is that a durable goods producer with market power wants to lower the quality of
used units because this translates into a higher price for new units. One way to
achieve this is by reducing the physical quality of the product by reducing the
durability built into new units. Another way is to introduce frequent style changes
that reduce the "perceived" quality of used units. To see this, suppose consumers
4 In Waldman
(1996a), durability chosen is always below the efficient level, while Hendel and Lizzeri
(1999a) provide two different comparisons. They show that durability can be either above, below or
equal to the first-best level. My feeling, however, is that a more important comparison is to the
second-best level defined by the firm's actual choice of outputs. For this comparison, they find that
durability chosen is always below the efficient level.
5 In Hendel and Lizzeri's
(1999a) analysis, there is a second reason that the monopolist distorts
durability. This second reason is that socially optimal durability is a function of the average valuation for
quality of the used good consumers, while the secondhand market price is determined by the valuation
for quality of the marginal consumer of used goods. This second reason is related to the analysis of
Spence (1975) concerning monopoly quality choice of a firm that sells a single quality level to
consumers who vary in their valuations for quality.
prefer goods with the latest style. In such a world, style changes might be common
not because consumers of new goods demand them, as in Pesendorfer (1995), but
because style changes make used units less substitutable for new units and thus
allow producers to increase the price of new units. Later in this paper, I discuss
another implication that concerns whether durable goods producers ever have an
incentive to reduce or even to eliminate the availability of used units on the
secondhand market.
New Product Introductions

Most of the durable goods literature ignores introductions of new products.
This omission is significant, in that almost every durable goods market is charac?
terized by products that improve and/or change over time. Indeed, new product
introductions have a variety of implications for durable goods markets. A number
of authors have recently started to consider this issue.
There are a number of papers that focus
on pricing given new product
introductions (for example, Levinthal
and Purohit, 1989; Fudenberg and Tirole,
1998; Lee and Lee, 1998). The most thorough of these analyses appears in Fuden?
berg and Tirole's paper. They consider a two-period monopoly setting in which
output is perfectly durable, but second-period new units are higher quality than
first-period new units. Further, consumers in their model vary in their valuations for
quality. Focusing on what happens given a secondhand market, the fundamental
result in their analysis is closely related to the durability analyses discussed above.
In the durability analyses, new and used units are imperfect substitutes because
quality deteriorates as units age. This creates a linkage between prices of new and
used units, which leads to various results, including the one concerning durability
described above. In Fudenberg and Tirole's analysis, new and used units are
imperfect substitutes because new units improve over time. They show a similar
linkage between prices of new and used units and derive various implications
concerning price and output decisions.
Another important idea is that of planned obsolescence, which was initially
developed in Waldman (1993, 1996b; see also Choi, 1994; Fishman and Rob, 2000;
Kumar, 2002).6 The key insight here is that introductions of new products are
subject to a time inconsistency problem similar to that discussed by Coase (1972).
The Coase insight was that, because in later periods the firm will not internalize
6 The term
planned obsolescence
is sometimes used to refer to the behavior of a firm that underinvests in
durability from a social welfare standpoint (for example, Swan, 1972; Bulow, 1986; Carlton and Perloff,
1994). But Bulow (1986, p. 747) also states: "Perhaps the greatest weakness of this paper is that it follows
in the tradition of using durability as a proxy for obsolescence. . . . 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. . . ." The papers I
discuss above follow this suggestion of Bulow and treat planned obsolescence in terms of new product
introductions, rather than the durability decision. See Grout and Park (2001) for a recent paper that
uses the term planned obsolescence to refer to durability choice.
how its output choice affects the value of units previously sold, a monopoly seller of
a durable good who cannot commit has an incentive to sell "too many" units in later
periods, and this reduces

overall profitability. The recent papers concerning
planned obsolescence show a similar logic operates for new product introductions.
The introduction of a new product can lower the value of used units by making
them obsolete. just as a durable goods monopolist
Hence, has an incentive to sell
too many units in later periods with the result that its own profitability is reduced,
the firm also has too high an incentive to introduce new products in later periods
that make used units obsolete, with the result again that overall profitability is
reduced.
To see this more concretely, consider Waldman (1996b), which analyzes a
model to Fudenberg
similar and Tirole's (1998) model, but endogenizes the
quality of new units in the second period (related analyses appear in Nahm, 2001;
Ambj0rnsen, 2002). In particular, there is a research and development decision at
the beginning of the second period, where the probability the monopolist sells new
higher quality units in the second period is increasing in the R&D investment.
When the firm introduces higher-quality new units in the second period, the
second-period market value of used units is reduced because the units are
traded from high-valuation to low-valuation consumers (such trades do not
occur if higher-quality units are not introduced). In this setting, if the monop?
olist cannot commit in the first period to its second-period R&D level, then it
invests more in R&D than if it could commit. The result is that, because first-period
consumers anticipate this and are thus willing to pay less for new units, this
overinvestment in R&D lowers the firm's overall profitability. Again, as was true
for Coase's time inconsistency problem, the firm can avoid the problem by
leasing its output.
The time inconsistency problemconcerning new product introductions shows
one way in which Coase's (1972) insight is more than an insight about output
Any action
choice. that affects the value of used units previously sold?such as an
R&D decision that affects whether used units become obsolete?can be subject to
time inconsistency. Indeed, the problem of time inconsistency is potentially more
important for other choices than for output. As argued earlier, contractual provi?
sions allow firms to commit at least to some extent to future outputs and prices. In
contrast, the size and nature of R&D investments?along with a range of other
future actions?would seem to be much more difficult to specify in contracts in an
enforceable manner.
Adverse Selection in a Model of New and Used Goods

Akerlof
(1970) considered in an analysis of used durable
adverse selection
goods, but did not incorporate new durables into his analysis. Hendel and Lizzeri
(1999b) consider a model of new and used durable goods where the market for
used units is characterized by adverse selection. Their results confirm Akerlof's
Michael Waldman 141
analysis that adverse selection can result in too little trade on the secondhand
market, and they also find other interesting results.7
One of the more interesting features of their analysis is the reason for trade in
the secondhand market. In Akerlof's (1970) analysis, all individuals have a valua?
tion for a single used unit that depends on its quality. If trade occurs, it consists of
used units moving from individuals with low valuations who initially own the units
to those with high valuations who initially do not. In Hendel and Lizzeri's (1999b)
analysis, each individual has a valuation for a single unit?new or used?that
depends on its quality. As a result, high-valuation consumers buy new units and sell
used units to low-valuation consumers. In other words, although the sellers of used
units have higher valuations for quality, they find it optimal to sell their used units
because they place a higher value on purchasing a new unit once the used unit is
sold.
Modeling the new and used markets this way allows the analysis to capture
important links between these markets. For example, similar to earlier discussions,
the price at which used units trade on the secondhand market affects the willing?
ness of consumers to pay for new units. That is, an increase in the price at which
a consumer can sell a used unit increases the amount consumers are willing to pay
for a new unit, since buying new and selling used are complementary activities.
Another important link concerns the effect of adverse selection in the used
unit market on the operation of the new unit market. Inefficiency in the market for
used units limits the price that buyers of new units will pay. As a result, profit-
maximizing sellers of new units will want to limit the extent to which their units are
subject to adverse selection when
they become used. This link is not emphasized in
Hendel and Lizzeri (1999b) because they do not consider the profit-maximization
problem of sellers of new units (in that analysis, they ignore market structure and
just assume a constant flow of new units come onto the new unit market each
period). However, as I will discuss later, this perspective has been emphasized in
recent papers that argue that the role of leasing in the new car market is that it
limits adverse selection in the used car market.
Specific Firm and Industry Practices
In this section,
I discuss theoretical models used to explain specific firm and
industry practices. I focus mostly on models that build on analyses discussed earlier.
The specific topics covered are the following: the role of sales in durable goods
Kim (1985) is an earlier analysis that considers adverse selection in a model with both new and used
durable goods. That analysis incorporates a maintenance decision with the main result that, in contrast
to what was argued by Akerlof (1970), the used units traded on the secondhand market can be either
higher or lower in quality than the used units that are kept for a second period. Kim does not address
whether, as in Akerlof's (1970) analysis, the volume of trade given asymmetric information is above or
below the full information level.
markets; practices that limit availability of used goods; the role of leasing in the new
car market; models of Microsoft's behavior; and aftermarket monopolization.
Sales in Durable Goods Markets

Most markets
including durable goods markets are characterized by sales; that
is, temporary price decreases employed to increase quantity sold. Why might firms
employ sales in durable goods markets?8
Conlisk, Gerstner and Sobel (1984) address this question with an infinite
period model of a durable goods monopoly in which new consumers enter the
market each period. These new consumers to have the same distribu?
are assumed
tion of high and low valuations for the monopolist's product as consumers already
in the market. Further, as in many of the models discussed earlier, the good is
infinitely durable, so a consumer who purchases the good leaves the market
forever. They analyze this model assuming no commitment; that is, each period's
output is chosen at the beginning of the period. For most periods, the monopolist
sets price equal to the willingness-to-pay of high-valuation consumers. However,
after a number of periods, a sufficiently large number of low-valuation consumers
are in the market that the firm reduces the price and sells to low-valuation
consumers. Thus, in most periods, the price is high and only high-valuation
consumers purchase, but there are also periodic temporary price reductions tar?
geted at low-valuation customers.
This
equilibrium is an intuitively plausible description for why we observe
periodic sales in durable goods markets. However, what happens if firms can
commit to future prices and/or quantities? Sobel (1991) analyzes this issue using
the same model investigated in Conlisk, Gerstner and Sobel (1984) (in this paper,
Sobel also further analyzes the no-commitment case). He finds that with commit?
ment, there are no sales. Rather, when the proportion of high-valuation consumers
is high, the firm sets a high price each period and sells only to these consumers,
while when the proportion is low, it sets a low price each period and sells to both
groups. One might wonder why in the former case the firm does not periodically
reduce the price and sell to the low group. The reason is that offering a low price,
even just periodically, lowers high-valuation consumers' willingness to pay enough
that it is more profitable never to sell to the low group.
Sobel (1984) also extends Conlisk, Gerstner and Sobel's (1984) analysis, where
the focus is oligopoly. In this paper, Sobel shows the existence of equilibria similar
to that described for the monopoly
above and no-commitment cases. That is, in
most periods, a firm prices high and sells only to high-valuation consumers, but
periodically the firm reduces its price and sells to low-valuation consumers. The
logic here is similar to that given above. As the number of low-valuation consumers
in the market rises, the profitability of selling to the low group rises, and each firm
thus eventually decreases its price.
*
Salop and Stiglitz (1982) offer a model of sales in nondurable goods markets.
Reducing the Availability of Used Units

Does a durable goods producer ever have an incentive to reduce the availabil?
ity of used units? The standard argument concerning this issue is due to Swan
(1980; see also Rust, 1986). Swan argued that since a good's initial price reflects the
price at which it will change hands in the secondhand market in the future, the
availability of used units does not reduce the overall profitability of a monopoly
seller.For example, Swan (p. 77) wrote: "[T]he pure monopolist selling such a
durable item as an automobile is paid an amount which reflects the net present
value of the stream of automobile services to possibly a whole host of future owners.
Competitive secondhand auto dealers (or scrap merchants and recyclers in the case
of aluminum) can then buy and sell the item indefinitely without in any way
restricting the power of the monopolist as the original seller."
Several recent
papers show this argument misses an important aspect of the
problem (Waldman, 1996a, 1997; Fudenberg and Tirole, 1998; Hendel and Lizzeri,
1999a). The argument of these papers is closely related to the earlier discussion of
durability choice.Because of potential substitutability across different vintages, the
availability of used units lowers the monopolist's new unit price. One response is for
the monopolist to reduce the durability of new units. This reduces the substitut?
ability between new and used units and allows the firm to increase the price for new
units. The point here is that another possibility is that the firm reduces or even
eliminates the availability of used units, where the return is again a higher new unit
price.9
One real world practice along these lines is the lease-only policy; that is, the
policy of leasing and refusing to sell output. This practice is not widely prevalent
among durable goods producers, but there are a number of prominent instances
each associated with a manufacturer that had significant market power: United
Shoe in the market for shoe machinery; IBM in the computer market; and Xerox
in the copier market.10 Waldman (1996a, 1997) and Hendel and Lizzeri (1999a)
9 A similar
argument appears in the earlier analyses of Benjamin and Kormendi (1974), Miller (1974),
Liebowitz (1982) and Levinthal and Purohit (1989). Those papers start by specifying demand functions
and show that under certain conditions reducing the availability of used units increases the firm's
profitability. However, since these analyses start with demand functions, one is left wondering whether
the required conditions are plausible or even possible. The more recent papers listed above start with
consumer utility and show that the conditions on demand functions necessary for the reduced avail?
ability of used units to increase profitability are quite easy to satisfy. Finally, Anderson and Ginsburgh
(1994) in a setting similar to the papers mentioned above consider the related issue of whether a
monopolist has an incentive to increase the transaction costs of trading in the secondhand market and
shows that it does sometimes. Hendel and Lizzeri (1999a) also consider this issue in their analysis.
10Prior to
1953, United Shoe employed a lease-only policy in marketing its shoe manufacturing
machines. In 1953, the courts ruled a variety of the firm's practices to be illegal, and, in particular, it was
prohibited from using the lease-only practice (Kaysen, 1956). Prior to 1956, IBM employed a lease-only
policy in marketing its computers, while in 1956, it agreed to a Justice Department consent decree
requiring it to sell as well as lease its machines. However, even after the consent decree, IBM offered
prices for the two options such that most consumers chose to lease rather than buy (Soma, 1976; Fisher,
McGowan and Greenwood, 1983). In 1975, Xerox and the Federal Trade Commission entered into a
consent decree that stated that one of Xerox's violations of antitrust law was its use of a lease-only policy.
argue that these firms

employed the lease-only policy as a way of reducing the
availability of used goods. That is, employing this policy allows firms to scrap some
or all of the used units that are returned and in this way reduce used unit
availability.11
One interesting question is why would a firm use the lease-only policy to
reduce the availability of used units, rather than repurchase and scrap the used
units it wants to remove from the market. I analyze this question in my 1997 paper
and show that if the firm cannot commit to the subsequent repurchase price when
it sells new units, then the repurchase-and-scrap strategy does not work because of
time inconsistency. At the date that repurchases need to take place, the firm has an
incentive not to repurchase and instead to allow the units to be available for
consumption. Further, even if the firm can contractually commit to a high repur?
chase price, the lease-only policy is still preferred as long as there is a positive
probability of bankruptcy. The logic is that any positive probability of bankruptcy
means "full" commitment is impossible, so lease-only is preferred.
Another setting in which the argument of this subsection applies is the
textbook market. The argument there is that publishers of popular textbooks
introduce frequent new editions where the goal is reducing the value of used copies
to zero; that is, firms eliminate used unit availability by introducing new editions
and making used copies worthless. At first, this setting may seem inconsistent with
the thread of the discussion. After all, the textbook market is typically quite
competitive, while the above argument depends on significant market power.
However, the textbook case is in fact consistent with the argument because of an
unusual feature. The person who decides which textbook to use, the professor, is
different than the person who decides whether to use the book, the student. From
the standpoint of the professor, there are typically alternative texts, but once a
professor has chosen a text, the student typically has little choice. As a result, in
selling to students (as opposed to marketing to professors) a textbook publisher has
substantial market power, and thus, it is quite plausible that publishers use new
editions to kill off the market for used copies.12
11A number of other

explanations for the lease-only policy have been put forth. Two of these are worth
mentioning. First, Posner (1976) suggests that the lease-only policy may be a response to the Coase time
inconsistency problem. However, avoiding the Coase problem requires short-term leases (DeGraba,
1994), while United Shoe, for example, employed long-term leases lasting 17 years. Second, Wiley,
Rasmusen and Ramseyer (1990) and Masten and Snyder (1993) ignore the monopoly aspect of the cases
and put forth efficiency rationales more consistent with a competitive market. However, given the
practice is not widely prevalent among durable goods producers, but instead has been observed in a
number of instances of firms with significant market power, this perspective seems suspect.
12The difference between the
argument here and the earlier argument involving obsolescence and new
product introductions concerns whether the firm is choosing the privately optimal frequency of new
editions. Here I am assuming that the firm chooses the privately optimal frequency so when it introduces
a new edition overall profits rise. In the earlier argument the firm faced a time-inconsistency problem
with the result that the firm introduced new products too frequently and, as a consequence, overall
profits fell. Since textbook publishers are in the market for many periods and can thus establish a
reputation for how often new editions are introduced, I feel that assuming that they choose the privately
optimal frequency of new editions is likely to be the more accurate approach.
Michael Waldman 145
Leasing in the New Car Market

Over the last 20 years, there has been dramatic growth in new car leasing. For
example, between 1990 and 1998, the percentage of new cars leased in the United
States grew from 7.3 percent of the market to 29.2 percent.13 What is the role of
new car leasing?
There are two standard
explanations, but both have problems. First, the
business press argues that leasing is a way for consumers to lower their monthly
payments (for example, Woodruff, 1994). That is, either because of consumer
myopia or imperfect capital markets, leasing is perceived as a way of making new
cars more affordable. But this argument suggests that, since high-income con?
sumers are likely both to be less myopic and have better access to capital
markets, leasing should be negatively related to consumer income. This predic?
tion, in turn, is problematic, since evidence indicates that leasing is positively
related to consumer income (for example, Aizcorbe and Starr-McCluer, 1997).
Second, another explanation is that the Tax Reform Act of 1986 included a
number of changes that made automobile leasing more attractive (for example,
Crocetti, 1988; Auster, 1990). The problem here is that this explanation pre?
dicts growth in leasing after 1986, but only until the new car market achieved a
new equilibrium, while the evidence indicates substantial growth long after
1986.
An alternative explanation is that leasing in the new car market is a response
to adverse selection in the used car market. Hendel and Lizzeri (2002) approach
the problem mainly from the perspective of a monopolist, while Johnson and
Waldman (2002) focus on competition. However, the basic logic of these analyses
is similar. Adverse selection in the used car market arises from the private infor?
mation of used car sellers, but this means leasing should avoid the problem,
because with leasing, used cars are returned to dealers and dealers have no private
information.14 In the monopoly case, this results in leasing, because avoiding
inefficiency in the used car market increases consumers' willingness to pay and thus
also increases firm profits. In the competitive case, this results in leasing because
avoiding inefficiency improves consumer welfare, so consumers prefer it. (In the
competitive case, consumers receive any increase in social welfare because firms
earn zero profits in equilibrium.)
This argument matches well with empirical evidence concerning the operation
of markets for new and used cars. For example, since leasing but not purchasing
avoids adverse selection, this argument predicts used cars that were leased
when new should sell for more than used cars that were purchased when new.
13The source for these

figures is CNW Marketing Research.
14To be
precise, the assumption that dealers have no private information concerning the used cars that
are returned is sufficient but not necessary for leasing to avoid the adverse selection problem. The
reason is that adverse selection arises when new cars are sold because of asymmetric information and
because potential sellers of used cars have the option of driving their used cars. Hence, even if leasing
does not eliminate the private information, it avoids the adverse selection problem as long as dealers
have no alternative use for the used cars that are returned.
Remember, adverse selection only low-quality used cars are traded, but if
means
adverse selection is avoided, then high-quality used cars are traded, too. This
prediction is consistent with evidence found in Desai and Purohit (1998). Similarly,
since leasing avoids adverse selection and should thus increase the volume of trade,
this explanation predicts the average length of time a new car is held before
changing hands should be shorter for leased cars. This prediction is consistent with
evidence found in Sattler
(1995).
This perspective on new car leasing suggests two interesting questions. First,
what is the role of the buy-back price in lease contracts for new cars, where the
buy-back price refers to the price typically included in lease contracts at which the
lessee can purchase the car at the end of the lease period? Hendel and Lizzeri
(2002) explain this in terms
of efficient matches in the used car market. That is,
having buy-backs means
used cars do not look completely homogeneous from the
standpoint of used car buyers, and as a consequence, in their analysis the monop?
olist increases its profits by employing buy-backs because it results in better match?
ing between used car buyers and the used cars purchased.
In contrast, Johnson and Waldman (2002) provide an explanation based on
efficient buy-backs. They consider a competitive model where, in contrast to
Akerlof's (1970) analysis, the deterioration in quality of the highest quality used
cars is so small that it is efficient for these cars not to be traded. In turn, buy-backs
are employed because they move the equilibrium toward this outcome, since they
allow the highest-quality used cars to be bought back at the end of the lease
contract. Note, since only the highest-quality used cars should be bought back, this
argument predicts very high buy-back prices, which is consistent with evidence
Johnson and Waldman collected from advertisements that appeared in the Sunday
New York Times during 1998.
The second question is this: why has new car leasing grown so dramatically
during the last 20 years? Hendel and Lizzeri (2002) argue that this growth may be
the result of improved durability of new cars over this period. The logic here is that
because a driver of a new car is more
likely to drive a car when it becomes old if its
quality has not fallen
too much, improved durability will aggravate the adverse
selection problem. Thus, since leasing reduces the adverse selection problem, it is
used more often as durability rises.
In contrast, Johnson and Waldman (2002) discuss an extension of their model
that incorporates a moral hazard problem concerning inadequate consumer main?
tenance associated with leasing. In this extension, leasing rises when new cars
become more reliable or trouble free because more reliable new cars means this
moral hazardproblem is small. The logic is that some aspects of the moral hazard
problem are avoided by including scheduled maintenance activities in lease con?
tracts, and when new cars become more reliable or trouble free there is less need
for maintenance outside of these scheduled activities. Johnson and Waldman then
argue that the growth in leasing may be the result of the improved reliability of new
cars over time.
Microsoft's Behavior
Since computer software is durable, some of the literature concerning Mi?
crosoft's behavior is relevant for understanding durable goods markets. In this
section, I discuss two issues: Microsoft's practice of frequent upgrades and Mi?
crosoft's low-price strategy in marketing Windows.15
In the earlier discussion of planned obsolescence, I argued that from the
standpoint of its own profitability, a durable goods monopolist that cannot commit
will introduce new products that make used units obsolete
too frequently. Waldman
(1993) and Choi (1994) show, in particular, that a monopolist selling a durable
good characterized by network externalities may practice planned obsolescence by
introducing new products that are incompatible with old ones. Recently, Ellison
and Fudenberg (2000) extended those analyses to investigate excessive upgrades in
software markets.
They consider a two-period model characterized by network
externalities in which a monopolist can introduce a higher quality new product in
the second period that is "backwards compatible" only. Backwards compatibility
means consumers who purchase the new good in the second period derive a
network benefit from consumers
who purchased the old good in the first, but a
first-period purchaser only benefits from second-period purchasers if the first-
period purchaser buys the new product in the second. Ellison and Fudenberg show
that if the monopolist cannot commit in the first period to whether it will upgrade
in the second, then both from the standpoint of its own profitability and from that
of social welfare, it upgrades too frequently. The reason is that in the second
period, the monopolist does not internalize how its product choice affects the
second-period value of units it sold as new in the first period.

This analysis suggests Microsoft lowers its own profitability with its frequent
upgrades, but if true, then Microsoft would have an incentive to choose actions that
constrain its own ability to introduce upgrades. Since Microsoft does not seem to be
taking any such actions, it is possible that a different perspective would be more
accurate.
Ellison
and Fudenberg (2000) also provide a second analysis. This analysis also
assumes an upgrade that is backwards compatible only, but, importantly here, they
assume commitment is possible so no time inconsistency problem exists. Also, this
analysis (but not the earlier one) has heterogeneous consumers. Here they show
that from a social welfare standpoint there may again be excessive upgrades. The
logic, related to Spence (1975), focuses on the distinction between marginal and
average consumers. When themonopolist thinks

upgrading about
given it can
commit in advance, it takes into account the negative impact the upgrade has on
the marginal consumer of the original product. But social welfare depends on the
15Because of the
ongoing court case, another behavior that has generated significant attention is
Microsoft's tying of Windows and Internet Explorer. Because most analyses of that issue do not focus on
the durable nature of the products, I do not discuss that literature here. For papers related to the tying
of Windows and Internet Explorer, see Whinston (1990, 2001), Farrell and Katz (2000), Choi and
Stefanidis (2001) and Carlton and Waldman (2002).
negative impact on the average consumer, so the firm will upgrade too often when
that negative impact is higher for the average consumer than for the marginal
consumer.
The secondaspect of Microsoft's behavior I consider is its pricing strategy for
Windows. During the ongoing court case, various observers noted that Microsoft's
price for Windows is below what a pure monopolist would charge; for example,
Richard Schmalensee (1999) stated that "a real monopolist?one who extracted
the last dollarof profit from consumers?would charge hundreds of dollars more
for the software that runs modern PCs. . ." Building on the earlier work of Buco-
vetsky and Chilton (1986), Hoppe and Lee (2002) provide an explanation for this
behavior based on the durability of the product. They argue that Microsoft charges
a low price to deter entry, or in other words, it limit prices. That is, because of the
durable nature of the product, the firm prices low and sells a large quantity. The
reason is that an individual who already owns an operating system has a lower
willingness to pay for a competing operating system than an individual who does
not. Hence, charging a low price can deter future entry because it lowers the
potential profitability of a rival investing in research and development and subse-
quently introducing a competing operating system.
Aftermarket Monopolization
A series of court cases involving firms such as Kodak, Data General, Unisys and
Xerox concern the issue of aftermarket monopolization. Aftermarkets refers to
markets for complementary goods and services such as maintenance and replace?
ment parts that may be needed after a consumer has purchased a durable good.
Aftermarket monopolization means the original durable goods producer stops
alternative producers from offering the complementary good or service, with the
result that the original producer monopolizes the aftermarket. For example, in
Eastman Kodak Co. v. Image Technical Services (504 U.S. 451 [1992]), the U.S.
Supreme Court heard a case in which Kodak refused to sell spare parts to alterna?
tive maintenance suppliers with the result that consumers of Kodak's products had
no option but to purchase maintenance from Kodak.
In-depth discussions of
aftermarket monopolization appear in Shapiro (1995), Chen, Ross and Stanbury
(1998) and Carlton (2001). Here, I lay out the major theories.
Because the Supreme Court ruling in the Kodak case was that a durable goods
producer with no market power that monopolizes an aftermarket for its own
product can be guilty of an antitrust violation, much of the literature focuses on
competition in the market for new units. I begin by discussing two closely related
theories in which maintenance market monopolization allows competitive produc?
ers to exploit market power after consumers are locked in.
The first is referred to as the costly information theory. It holds that consumers
are locked in once they purchase a new unit from a durable goods producer and
that consumers ignore the cost of maintenance in their original purchase decisions.
In this situation, producers exploit consumers' locked-in positions by first stopping
other firms from selling maintenance and then raising the maintenance price. The
Michael Waldman 149
result is a standard deadweight loss due to the monopoly pricing of maintenance,

but no transfer between consumers and firms since competition in the market for
new units causes the price of new units to fall so that a firm receives zero profits
across the two markets in equilibrium.
A related theory is the lack-of-commitment theory developed in Borenstein,
Mackie-Mason and Netz (1995). The difference here is that consumers anticipate
whether the maintenance market will be monopolized and pay less for a new unit
when they expect monopolization. In such circumstances, a durable goods pro?
ducer would want to commit to allowing competition in the maintenance market,
but monopolization occurs in this model because of an inability to commit. In this
theory, as in the costly information theory, the cost of the practice is the deadweight
loss due to the monopoly pricing of maintenance.16
Although these theories were put forth in response to the recent court cases,
each has problems concerning applicability to those cases. The costly information
theory assumes uninformed consumers, which seems unlikely in some of the cases
in which consumers were sophisticated businesses and the cost of maintenance
was a significant proportion of the total cost of using the product. The lack-of-
commitment theory assumes commitment is impossible, but this assumption seems
suspect because long-term maintenance contracts are quite common in many
industries where the practice has been observed.
Because of these concerns, I find two other theories more appealing. First, the
practice of aftermarket monopolization may be used to price discriminate more
effectively, as argued in Chen and Ross (1993) and Klein (1993). This argument is
similar to the explanation for tie-ins used, for example, to explain IBM's one-time
practice of requiring purchasers of its tabulating machines to also purchase its
punch cards. In this theory, consumers with higher valuations for the durable
goods producer's product are heavy users of maintenance, with the result that the
seller more effectively price discriminates by monopolizing the maintenance mar?
ket and raising its price. This theory provides a clear rationale for maintenance
market monopolization by firms with significant market power, but has difficulty
explaining maintenance market monopolization by firms with little or no market
power. Although Klein (1993) argues that since we observe significant price dis?
crimination even in industriesthat are quite competitive, the price discrimination
argument should not be ruled out as a possible explanation for why a firm would
monopolize the maintenance market in such an industry.17
16A third
theory along these lines is the surprise theory. In this theory, consumers expect the
maintenance market to be competitive and are surprised when monopolization occurs. Most discussions
of this theory claim that, in addition to the deadweight loss associated with the other two theories, there
is a transfer between the consumers and the firm. However, if competitive firms lower the price for new
units until they receive zero profits in equilibrium, as occurs in the costly information theory, then there
is no transfer and the costly information and surprise theories are equivalent.
17Two other
explanations for why a firm with significant market power would monopolize the mainte?
nance market for its own product are put forth in Hendel and Lizzeri (1999a) and Morita and Waldman
(2002). The former paper argues that maintenance market monopolization arises because it allows a
Second, an efficiency argument has recently been put forth that builds on the
earlier work of Schmalensee (1974), Su (1975) and Rust (1986). Those papers show
that a durable goods monopoly paired with competitive maintenance leads to
inefficiency. That is, because the monopolist sets the price for new units above
marginal cost while maintenance is priced competitively, consumers sometimes
maintain used units when purchasing new units would be efficient. Carlton and
Waldman (2001) and Morita and Waldman (2001) show that this translates into a
reason for maintenance market monopolization both when the market for new
units is monopolized and when it is competitive. The monopoly case is straightfor?
ward, since a monopolist of both markets can increase profits by setting prices that
reduce the distortion. But the result also holds under competition when consumers
bear switching costs from changing brands. Switching costs create market power in
the market for new units at the time a consumer chooses whether to replace or to
maintain a used unit, and thus, monopolizing the maintenance market again allows
a firm to reduce the distortion. This latter argument is potentially important for a
number of the court cases, including the Kodak case, because of the presence of
significant consumer switching costs.
Conclusion
Early work on durable goods theory during the 1970s and 1980s established
important fundamental principles. But because of some unrealistic assumptions,
this literature only limited progress in advancing
made our understanding of real
world marketsfor durable goods. In the last ten years, however, the literature has
started to consider more realistic models, with the result being significant advances
in our understanding of these markets. For example, we now better understand
durability choice and the issues associated
with new products introductions. We
also have a better understanding of a variety of specific practices employed in
durable goods markets, such as those used to make goods produced in earlier
periods obsolete, leasing in the new car market and aftermarket monopolization.
Although durable goods theory has made large advances in the last ten years,
there are a number of areas where I believe further research could be fruitful. First,
most of the literature assumes either monopoly or perfect competition, while
clearly most real world markets are either oligopolistic or monopolistically com?
petitive. It might be useful, therefore, to incorporate some of the recent advances
into oligopoly and monopolistically competitive models.18 Second, outside of the
literature on aftermarkets and the lease-only analyses of Waldman (1997) and
durable goods monopolist to raise the price for new units by more effectively controlling the speed with
which the quality of its product deteriorates. The latter paper argues that, much like leasing, monop-
olizing the maintenance market can be used to avoid problems due to time inconsistency.
18There is a small
existing literature on durable goods oligopoly. See, for example, Sobel (1984), Gul
(1987) and Esteban and Shum (2002). Bulow (1986) also contains some analysis of oligopoly.
Hendel and Lizzeri (1999a), there has been little research concerning the antitrust
implications of the recent advances, and more such analyses would be useful. For
example, Carlton and Gertner (1989) show, using a Swan-type analysis, that merg?
ers lead to substantially smaller
deadweight losses in durable good industries as
opposed to nondurable good industries, and it would be useful to explore the
robustness of this conclusion to the introduction of a more modern approach to
modeling durability. Third, most of the recent literature focuses on consumer
durables. But many durables are intermediate inputs used to produce consumer
goods. Some of the work on the theory of the firm, such as Williamson (1975,
1985), considers this type of intermediate input, either implicitly or explicitly. But
given the recent advances in the theory of durable consumer goods, it would be
interesting to explore whether our understanding of markets for durable consumer
goods has implications for durable intermediate inputs not captured by existing
theory.
? / would like to thank Brad De Long, Igal Hendel, Justin Johnson, Alan Krueger,
Alessandro Lizzeri, Hodaka Morita and Timothy Taylor for comments on earlier drafts,
Dennis Carlton for conversations concerning topics discussed in this paper and Ari Gerstlefor
research assistance.
References
Aizcorbe, Ana and Martha Starr-McCluer. Bought or Leased." Taxation for Accountants.
1997. "Vehicle Ownership, Purchases, and Leas? June, 44, pp. 352-55.
ing: Consumer Survey Data." MonthlyLabor Re? Ausubel, Lawrence and Raymond Deneckere.
view.June, 120:6, pp. 34-40. 1989. "Reputation in Bargaining and Durable
Akerlof, George. 1970. "The Market for 'Lem- Goods Monopoly." Econometrica.May, 57:3, pp.
ons': Quality Uncertainty and the Market Mech? 511-31.
anism." QuarterlyJournal of Economics.August, Bagnoli, Mark, Stephen Salant and Joseph
84:3, pp. 488-500. Swierzbinski. 1989. "Durable-Goods Monopoly
with Discrete Demand." Journal of Political Econ?
Ambj0rnsen, Terje. 2002. "Improving Durable
Goods and the Osborne Effect." Mimeo, Univer? omy.December, 97:6, pp. 1459-478.
Barro, Robert. 1972. "Monopoly and Con-
sity of Oslo. trived Depreciation." Journal of PoliticalEconomy.
Anderson, Simon and Victor Ginsburgh. 1994.
"Price Discrimination via Second-Hand Mar? May/June, 80:3. pp. 598-607.
kets." European EconomicReview.January, 38:1, Benjamin, Daniel and Roger Kormendi. 1974.
"The Interrelationship Between the Markets for
pp. 23-44. New and Used Durable Goods." Journal of Law
Auernheimer, Leonardo and Thomas Saving. and Economics.October, 17:2, pp. 381-401.
1977. "Market Organization and the Durability Bond, Eric. 1982. "A Direct Test of the 'Lem-
of Durable Goods." Econometrica. January, 45:1, ons' Model: The Market for Used Pickup
pp. 219-28. Trucks." AmericanEconomicReview. September,
Auster, Rolf. 1990. "Extent of Business Use 72:4, pp. 836-40.
May Determine Whether an Auto Should be Bond, Eric. 1984. "Test of the Lemons Model:
Reply." AmericanEconomicReview.74:4, pp. 801- verage Theory." Rand Journal of Economics.

04. Spring, 32:1, pp. 52-71.
Bond, Eric and Larry Samuelson. 1984. "Du? Coase, Ronald. 1972. "Durabilityand Monop?
n
rable Good Monopolies with Rational Expecta? oly. Journal of Law and Economics.April, 15:1, pp.
tions and Replacement Sales." Rand Journal of 143-49.
Economics.15:3, pp. 336-45. Conlisk, John, Eitan Gerstner and Joel Sobel.
Borenstein, Severin, Jeff Mackie-Mason and 1984. "Cyclic Pricing by a Durable Goods Mo?
Janet Netz. 1995. "Antitrust Policy in Aftermar- nopolist." QuarterlyJournal of Economics.August,
kets." Antitrust Law Journal. Winter, 63:2, pp. 99:3, pp. 489-505.
455-82. Crocetti, Nicholas. 1988. "Automobile and
Other Leasing Arrangements Remain Attractive
Bucovetsky, Sam and John Chilton. 1986.
"Concurrent Renting and Selling in a Durable Despite Loss of Some Benefits." Taxationfor Ac-
Goods Monopoly Under Threat of Entry."Rand countants.April, 40:4, pp. 198-201.
Journal of Economics.Summer, 17:2, pp. 261-75. DeGraba, Patrick. 1994. "No Lease is Short
Bulow, Jeremy. 1982. "Durable Goods Monop? Enough to Solve the Time Inconsistency Prob?
olists."Journal of PoliticalEconomy.April, 90:2, pp. lem." Journal of Industrial Economics.December,
314-32. 42:4, pp. 361-74.
Bulow, Jeremy. 1986. "An Economic Theory Desai, Preyas and Devavrat Purohit. 1998.
of Planned Obsolescence." QuarterlyJournal of "Leasing and Selling: Optimal Marketing Strate?
Economics.November, 101:4, pp. 729-49. gies for a Durable Goods Firm." Management
Science.November, 44:11, pp. S19-S34.
Butz, David. 1990. "Durable-Good Monopoly
and Best-Price Provisions." AmericanEconomicRe? Ellison, Glenn and Drew Fudenberg. 2000.
"The Neo-Luddite's Lament: Excessive Up?
view. December, 80:5, pp. 1062-076.
Carlton, Dennis. 2001. "A General Analysis of grades in the Software Industry." RandJournal of
Economics.Summer, 31:2, pp. 253-77.
Exclusionary Conduct and Refusal to Deal: Why
Emons, Winand and George Sheldon. 2002.
Aspen and Kodak are Misguided." AntitrustLaw "The Market for Used Cars: A New Test of the
Journal. 68:3, pp. 659-83. Lemons Model." Mimeo, University of Bern.
Carlton, Dennis and Robert Gertner. 1989.
Esteban, Susanna and Matthew Shum. 2002.
"Market Power and Mergers in Durable-Good
"Durable-Goods Oligopoly with Secondary Mar?
Industries." Journal of Law and Economics.Octo?
kets: The Case of Automobiles." Mimeo, Johns
ber, 32:2, pp. 203-26.
Hopkins University.
Carlton, Dennis and Jeffrey Perloff. 1994. Farrell, Joseph and Michael Katz. 2000. "Inno?
Modern Industrial Organization, Second Edition.
vation, Rent Extraction, and Integration in Sys?
New York: HarperCollins. tems Markets."Journal of IndustrialEconomics.De?
Carlton, Dennis and Michael Waldman. 2001. cember, 48:4, pp. 413-32.
"Competition, Monopoly, and Aftermarkets." Fisher, Franklin, John McGowan and Joen
Mimeo, Cornell University. Greenwood. 1983. Folded,Spindledand Mutilated:
Carlton, Dennis and Michael Waldman. 2002. EconomicAnalysis and U.S. v. IBM. Cambridge,
"The Strategic Use of Tying to Preserve and Mass.: MIT Press.
Create Market Power in Evolving Industries." Fishman, Arthur and Rafael Rob. 2000. "Prod?
Rand Journal of Economics.Summer, 33:2, pp. uct Innovation by a Durable-Good Monopoly."
194-220. Rand Journal of Economics.Summer, 31:2, pp.
Chen, Zhiqi and Thomas Ross. 1994. "Refusals 237-52.
to Deal, Price Discrimination and Independent Fudenberg, Drew and Jean Tirole. 1998. "Up?
Service Organizations." Journal of Economicsand grades, Trade-ins, and Buybacks."RandJournalof
ManagementStrategy.Winter, 2:4, pp. 593-614. Economics.Summer, 29:2, pp. 235-58.
Chen, Zhiqi, Thomas Ross and W. T. Stan- Fudenberg, Drew, David Levine and Jean Ti?
bury. 1998. "Refusalsto Deal and Aftermarkets." role. 1985. "Infinite Horizon Models of Bargain?
Reviewof IndustrialOrganization.April, 13:1-2, pp. ing with One-Sided Incomplete Information," in
131-51. Game TheoreticModelsof Bargaining.A. Roth, ed.
Choi, Jay Pil. 1994. "Network Externality, Cambridge: Cambridge University Press, pp. 73-
Compatibility Choice, and Planned Obsoles? 98.
cence." Journal of IndustrialEconomics.June, 42:2, Genesove, David. 1993. "Adverse Selection in
pp. 167-82. the Wholesale Used Car Market."Journal of Po?
Choi, Jay Pil and Christodoulos Stefanadis. litical Economy.August, 101:4, pp. 644-65.
2001. "Tying, Investment, and the Dynamic Le- Gilligan, Thomas. 2002. "Adverse Selection
Michael Waldman 153
and Trade in Used Durable Goods: Evidence crimination in Durable Goods Monopoly with
from the Market for Business Aircraft." Mimeo, Resale Trading." InternationalJournal of Industrial
University of Southern California. Organization.November, 20:9, pp. 1313-339.
Grout, Paul and In-Uck Park. 2001. "Compet? Kutsoati, Edward and Jan Zabojnik. 2001. "Du?
itive Planned Obsolescence." Mimeo, University rable Goods Monopoly, Learning-by-Doing and
of Pittsburgh. 'Sleeping Patents.'" Mimeo, University of South?
Gul, Faruk. 1987. "Noncooperative Collusion ern California.
in Durable Goods Oligopoly." Rand Journal of Lee, In Ho and Jonghwa Lee. 1998. "ATheory
Economics.Summer, 18:2, pp. 248-54. of Economic Obsolescence." Journal of Industrial
Gul, Faruk, Hugo Sonnenschein and Robert Economics.September, 46:3, pp. 383-401.
Wilson. 1986. "Foundations of Dynamic Monop? Levhari, David and T. Srinivasan. 1969. "Du?
oly and the Coase Conjecture." Journal of Eco? rability of Consumption Goods: Competition
nomic Theory.June, 39:1, pp. 155-90. versus Monopoly." American Economic Review.
Hendel, Igal and Alessandro Lizzeri. 1999a. March, 59:1, pp. 102-07.
"Interfering with Secondary Markets."RandJour? Levine, David and Wolfgang Pesendorfer.
nal of Economics.Spring, 30:1, pp. 1-21. 1995. "When are Agents Negligible?" American
Hendel, Igal and Alessandro Lizzeri. 1999b. EconomicReview.December, 85:5, pp. 1160-170.
"Adverse Selection in Durable Goods Markets." Levinthal, Daniel and Devavrat Purohit. 1989.
AmericanEconomicReview. December, 89:5, pp. "Durable Goods and Product Obsolescence."
1097-115. MarketingScience.Winter, 8:1, pp. 35-56.
Hendel, Igal and Alessandro Lizzeri. 2002. Liebowitz, Stan. 1982. "Durability, Market
"The Role of Leasing Under Adverse Selection." Structure, and New-Used Goods Models." Amer?
Journal of Political Economy.February, 110:1, pp. ican Economic Review. September, 72:4, pp.
113-43. 816-24.
Hoppe, Heidrun and In Ho Lee. 2002. "Entry Maskin, Eric and John Riley. 1984. "Monopoly
Deterrence and Innovation in Durable-Goods with Incomplete Information." Rand Journal of
Monopoly." Mimeo, University of Southampton. Economics.15:2, pp. 171-96.
EuropeanEconomicReview.Forthcoming. Masten, Scott and Edward Snyder. 1993.
Johnson, Justin and Michael Waldman. 2002. "United States versus United Shoe Machinery
"Leasing, Lemons, and Buy-Backs."Mimeo, Cor? Corporation: On the Merits."Journal of Law and
nell University. RandJournal of Economics.Forth? Economics.April, 36:1, pp. 33-70.
coming. Miller, H. Lawrence Jr. 1974. "On Killing Off
Kahn, Charles. 1986. "The Durable Goods Mo? the Market for Used Textbooks and the Rela?
nopolist and Consistency with Increasing Costs." tionship Between Markets for New and Used
Econometrica.March, 54:2, pp. 275-94. Secondhand Goods." Journal of PoliticalEconomy.
Karp, Larry and Jeffrey Perloff. 1996. "The May/June, 82:3, pp. 612-19.
Optimal Suppression of a Low-Cost Technology Morita, Hodaka and Michael Waldman. 2001.
by a Durable-Good Monopoly." Rand Journal of "Competition, Monopoly Maintenance, and
Economics.Summer, 27:2, pp. 346-64. Consumer Switching Costs." Mimeo, Cornell
Kaysen, Carl. 1956. UnitedStatesv. UnitedShoe University.
MachineryCorporation: An EconomicAnalysis of an Morita, Hodaka and Michael Waldman. 2002.
Antitrust Case. Cambridge, Mass.: Harvard Uni? "Durable Goods, Monopoly Maintenance, and
versity Press. Time Inconsistency." Mimeo, Cornell University.
Kim, Jae-Cheol. 1985. "The Market for 'Lem? Mussa, Michael and Sherwin Rosen. 1978.
ons' Reconsidered: A Model of the Used-Car "Monopoly and Product Quality."Journal of Eco?
Market with Asymmetric Information." American nomic Theory.August, 18:2, pp. 301-17.
EconomicReview.September, 75:4, pp. 836-43. Nahm, Jac. 2001. "Durable Goods Monopoly
Kim, Jae-Cheol. 1989. "Trade in Used Goods with Endogenous Innovation." Mimeo, Hong
and Durability Choice." International Economic Kong University of Science and Technology.
Journal. Autumn, 3:3, pp. 53-63. Pesendorfer, Wolfgang. 1995. "Design Inno?
Kleiman, Ephraim and T. Ophir. 1966. "The vation and Fashion Cycles." AmericanEconomic
Durability of Durable Goods." Reviewof Economic Review.September, 85:4, pp. 771-92.
Studies.August, 33:2, pp. 165-78. Porter, Robert and Peter Sattler. 1999. "Pat?
Klein, Benjamin. 1993. "MarketPower in An? terns of Trade in the Market for Used Durables:
titrust: Economic Analysis After Kodak."Supreme Theory and Evidence." Mimeo, Northwestern
CourtEconomicReview.3, pp. 43-92. University.
Kumar, Praveen. 2002. "Price and Quality Dis- Posner, Richard. 1976. AntitrustLaw: An Eco-
nomicPerspective.Chicago: University of Chicago Su, Teddy T. 1975. "Durability of Consump?

Press. tion Goods Reconsidered." AmericanEconomicRe?
Rust, John. 1986. "When is it Optimal to Kill view. March, 65:1, pp. 148-57.
off the Market for Used Durable Goods?" Econo? Swan, Peter. 1970. "Durability of Consump?
metrica.January, 54:1, pp. 65-86. tion Goods." AmericanEconomicReview. Decem?
Salop, Stephen andjoseph Stiglitz. 1982. "The ber, 60:5, pp. 884-94.
Theory of Sales: A Simple Model of Equilibrium Swan, Peter. 1971. "The Durability of Goods
Price Dispersion with Identical Agents." Ameri? and Regulation of Monopoly." BellJournal ofEco?
can EconomicReview.December, 72:5, pp. 1121- nomics.Spring, 2:1, pp. 347-57.
130. Swan, Peter. 1972. "Optimum Durability, Sec?
Sattler, Peter. 1995. "New Cars, Used Cars and ond Hand Markets, and Planned Obsolescence."
Clunkers: The Imperfect Substitutability of Used Journal of Political Economy.May/June, 80:3, pp.
Goods and Patterns of Trade in the Market for 575-85.
Durables: Theory and Evidence." Mimeo, North- Swan, Peter. 1980. "Alcoa: The Influence of
western University. Recycling on Monopoly Power."Journal of Politi?
Schmalensee, Richard. 1970. "Regulation and cal Economy.February, 88:1, pp. 76-99.
the Durability of Goods." BellJournal of Econom? Waldman, Michael. 1993. "A New Perspective
ics. Spring, 1:1, pp. 54-64. on Planned Obsolescence." QuarterlyJournal of
Schmalensee, Richard. 1974. "Market Struc? Economics.February, 108:1, pp. 273-83.
ture, Durability, and Maintenance Effort." Re? Waldman, Michael. 1996a. "Durable Goods
view of EconomicStudies.April, 41:2, pp. 277-87. Pricing When Quality Matters."Journal of Busi?
Schmalensee, Richard. 1979. "Market Struc? ness. October, 69:4, pp. 489-510.
ture, Durability, and Quality:A Selective Survey." Waldman, Michael. 1996b. "Planned Obsoles?
EconomicInquiry.April, 17:2, pp. 177-96. cence and the R&D Decision." Rand Journal of
Schmalensee, Richard. 1999. "The Gover- Economics.Autumn, 27:3, pp. 583-95.
ment's Soft Case Against Microsoft." Boston Waldman, Michael. 1997. "Eliminating the
Globe.City Edition, July 10, p. A17. Market for Secondhand Goods." Journal ofLaw
Shapiro, Carl. 1995. "Aftermarkets and Con? and Economics.April, 40:1, pp. 61-92.
sumer Welfare: Making Sense of Kodak."Anti? Whinston, Michael. 1990. "Tying, Foreclosure,
trustLawJournal. Spring, 63, pp. 483-511. and Exclusion." AmericanEconomicReview. Sep?
Sieper, E. and Peter Swan. 1973. "Monopoly tember, 80:4, pp. 837-59.
and Competition in the Market for Durable Whinston, Michael. 2001. "Exclusivityand Ty?
Goods." Reviewof EconomicStudies.July, 40:3, pp. ing in US. v. Microsoft What We Know, and
333-51. Don't Know." Journal of Economic Perspectives.
Sobel, Joel. 1984. "The Timing of Sales." Re? Spring, 15:2, pp. 63-80.
view of EconomicStudies.July, 51:3, pp. 353-68. Wicksell, Knut. 1934. Lectureson PoliticalEcon?
Sobel, Joel. 1991. "Durable Goods Monopoly omy,Volume1. London: Routledge & Kegan Paul.
with Entry of New Consumers." Econometrica. Wiley, John, Eric Rasmusen and Mark Ram-
September, 59:5, pp. 1455-485. seyer. 1990. "The Leasing Monopolist." UCLA
Sobel, Joel and Ichiro Takahashi. 1983. "A Law Review.April, 37, pp. 693-731.
Multistage Model of Bargaining." Review of Eco? Williamson, Oliver E. 1975. Marketsand Hier?
nomicStudies.July, 50:3, pp. 411-26. archies:Analysis and Antitrust Implications.New
Soma, John T. 1976. The ComputerIndustry:An York: The Free Press.
Economic-LegalAnalysis of its Technology and Williamson, Oliver E. 1985. TheEconomicInsti?
Growth.Lexington, Mass.: Lexington Books. tutionsof Capitalism:Firms,Markets,RelationalCon?
Spence, A. Michael. 1975. "Monopoly, Quality tracting.New York: The Free Press.
and Regulation." Bell Journal of Economics.Au- Wilson, Charles. 1980. "The Nature of Equi?
tumn, 6:2, pp. 417-21. librium in Markets with Adverse Selection." Bell
Stokey, Nancy. 1981. "Rational Expectations Journal of Economics.Spring, 11:1, pp. 108-30.
and Durable Goods Pricing." BellJournal of Eco? Woodruff, D. 1994. "Leasing Fever." USA To?
nomics.Spring, 12:1, pp. 112-28. day. April 18, pp. 1B-4B.

Durable Good Theory

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Durable Good Theory

Enviado por

Direitos autorais:

Formatos disponíveis

American Economic Association

Durable Goods Theory for Real World Markets

Durable Goods for Real World

goods constitute an important part of economic production.

? Michael Waldmanis the Charles H. Dyson Professor in Management and Professor of

in the 1970s on durable

similar to problems faced by durable goods producers selling to consumers, and

Three Classic Contributions to Durable Goods Theory

Time Inconsistency and Durable Goods

(1982). Rather, Coase considered the more

important for understanding a variety of contractual provisions that allow firms

to avoid the time

Adverse Selection and Durable Goods

Three Recent Contributions

Recent contributions to durable goods theory have served to extend and

Durability Choice in Real World Markets

In common with Wicksell (1934) and most subsequent treatments of product

New Product Introductions

periods, and this reduces

Adverse Selection in a Model of New and Used Goods

Specific Firm and Industry Practices

Sales in Durable Goods Markets

Reducing the Availability of Used Units

argue that these firms

11A number of other

Leasing in the New Car Market

13The source for these

second-period value of units it sold as new in the first period.

average consumers. When themonopolist thinks

result is a standard deadweight loss due to the monopoly pricing of maintenance,

Reply." AmericanEconomicReview.74:4, pp. 801- verage Theory." Rand Journal of Economics.

nomicPerspective.Chicago: University of Chicago Su, Teddy T. 1975. "Durability of Consump?

Você também pode gostar