This article was downloaded by: [128.122.253.

228] On: 02 June 2015, At: 07:39

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA

Marketing Science
Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org

Competition in Durable Goods Markets: The Strategic

Consequences of Leasing and Selling
Preyas S. Desai, Devavrat Purohit,

To cite this article:

Preyas S. Desai, Devavrat Purohit, (1999) Competition in Durable Goods Markets: The Strategic Consequences of Leasing and
Selling. Marketing Science 18(1):42-58. http://dx.doi.org/10.1287/mksc.18.1.42

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact permissions@informs.org.

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.

© 1999 INFORMS

Please scroll down for article—it is on subsequent pages

INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
 Competition in Durable Goods Markets:
The Strategic Consequences of Leasing
and Selling
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
Preyas S. Desai • Devavrat Purohit
desai@mail.duke.edu • purohit@mail.duke.edu
Fuqua School of Business, Duke University, Durham, North Carolina 27708
Abstract
In marketing durable goods, manufacturers use varying de-
grees of leasing and selling to consumers, e.g., cars, photo-
copiers, personal computers, airplanes, etc. The question that
this raises is whether the distinction between leases and sales
is simply one of price, or whether the proportion of leases
and sales effects a firm’s ability to compete in the market. In
this paper we use two approaches to argue that leasing and
selling create strategic consequences that extend beyond
prices. First, we develop a stylized theoretical model that
shows that the optimal proportion of leases and sales de-
pends on the competitiveness of the market and on the in-
herent reliability of the firm’s product. And second, we find
support for the implications of our theoretical model with
data from the automobile industry.
The U.S. automobile industry has seen a large increase in
leasing over the last five years. However, the extent to which
leasing has been embraced varies widely across manufactur-
ers. For example, in 1993 the sport utility segment had the
following lease percentages: Ford Explorer, 29%; Jeep Grand
Cherokee, 24%; Toyota 4-Runner, 11%; and Chevrolet Blazer,
9%. In addition, manufacturers often vary lease percentages
across models. For example, in 1993 Ford leased 22% of its
Crown Victoria model, 35% of its Taurus model, and 42% of
its Probe model. A popular argument for why we see these
differences is that higher priced cars are leased more often
because leasing makes them more “affordable.” However,
this rationale is not compelling in the face of our data. For
example, the Ford Probe was priced significantly lower than
the Crown Victoria and yet it was leased almost twice as
often.
Marketing Science/Vol. 18, No. 1, 1999
pp. 42–58
To develop a better understanding of why we observe dif-
ferences in the proportion of leasing, we develop a two-
period model of a duopoly in which each manufacturer
chooses its optimal quantity and the fraction of units it wants
to lease. We find that in equilibrium neither firm leases all
its units—either they use a mix of leasing and selling or they
use only selling. Our analysis suggests that the fraction of
leased cars decreases as the manufacturers’ products become
more similar and the competition between them increases.
The intuition for this result is that a higher fraction of leases
puts the firm at a competitive disadvantage in the future.
This occurs because, unlike firms that sell their product, firms
that lease are at a price disadvantage.
Another important finding in this paper is that the extent
of leasing chosen by a manufacturer depends on the reli-
ability of its product. In particular, all else being equal, the
lower a product’s reliability, the lower its proportion of
leases. Within the context of the automobile industry, this
suggests that more expensive cars may be leased more often
because they are of higher quality and not necessarily be-
cause they are more expensive.
Finally, we test the implications of our theoretical model
with data from the U.S. automobile market. In particular, for
1993 model year cars, we develop a measure of reliability
using data from Consumer Reports. In addition, we develop a
measure of the extent of competition in each segment of the
automobile market. We support our hypotheses by finding
that the extent to which a car model is leased depends
strongly on its predicted reliability and on the competitive
intensity within the segment.
(Competition; Automobile; Lease; Durable Goods)
0732-2399/99/1801/0042$05.00
Copyright q 1999, Institute for Operations Research
and the Management Sciences
 COMPETITION IN DURABLE GOODS MARKETS:
THE STRATEGIC CONSEQUENCES OF LEASING AND SELLING
1. Introduction
A firm that markets durables also has to decide
whether to market its product through leases or sales.
This lease-sell decision has commonly been thought of
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
as a simple problem of determining an optimal price.
However, because leasing and selling create strategic
consequences that extend beyond prices, this paper ar-
gues that the lease-sell decision is more than a mere
pricing problem. In particular, for a marketer of a du-
rable good, leased and sold units pose different levels
of competition in the future. To highlight the strategic
nature of these decisions, consider the following
examples.
1. When the old Bell system had a monopoly, it re-
fused to sell telephones to consumers (Brenner 1992).
At present most manufacturers sell their telephones.
2. In 1993 in the automobile market, Nissan leased
25% of its Altima automobiles, while Acura leased 60%
of its Integras. Both the Altima and the Integra com-
pete in the same segment of the U.S. car market.
3. In the high-end photocopier market, over 80% of
the units are leased, whereas few copiers are leased in
the lower-end market (Menezes and Serbin 1993).
These examples suggest not only that the leasing and
selling strategies chosen by firms are quite different
from one another, but also that the strategy could be
influenced by the competitive environment in the in-
dustry. From a research perspective, the interesting
question is “Why?” To answer this question, we need
to understand the conditions that would lead manu-
facturers to lease most of their units (such as the old
AT&T), sell nearly all of them (such as low-end copi-
ers), or adopt a combination of both leasing and selling
(such as Acura and Nissan). In this paper, we examine
the idea of an “optimal” combination of leases and
sales, and how it can be affected by market and firm
characteristics. By offering a compelling rationale for
why different markets have varying proportions of
leases and sales, we show that the lease-sales ratio is
much more than a pricing problem. We take two im-
portant steps to achieve this goal. First, we develop a
stylized model to understand the strategic effects of
leasing and selling in the presence of competition; and
second, we test the implications of our model with data
from the automobile industry.
Marketing Science/Vol. 18, No. 1, 1999
The U.S. automobile market represents an interest-
ing arena in which to study firms’ leasing strategies.
Not only are leases widely available, but manufactur-
ers have embraced them to varying degrees. Some in-
dustry observers have attributed the recent surge in
leasing in the U.S. automobile market largely to the
“affordability” benefit of leasing.1 In particular, they
argue that a key factor favoring leasing is the benefit
of lower monthly payments which make cars more
affordable. Because lessees do not have to finance the
residual value of their cars, their monthly payments
are much lower than they would be were they to fi-
nance the purchase of the car. Thus, some industry ob-
servers argue that leasing allows consumers to get new
cars that they could not otherwise afford. This afford-
ability benefit appears especially important in recent
times as the growth in new car prices has far out-
stripped the growth in family incomes, and monthly
budget constraints have become tighter (Business Week
1994). A direct implication of this argument is that leas-
ing is beneficial in meeting the needs of consumers
with credit or cash-flow constraints whose limited ca-
pacity to borrow money might not allow them to pur-
chase a car.
Although the affordability argument has a certain
intuitive appeal—it may even hold for certain consum-
ers—it does not fully explain the institutional realities
of the automobile market. For example, General Mo-
tors has a “Smart Lease” program in which the lessee
pays the entire leasing payment in a lump-sum at the
beginning of the contract. Clearly, a “Smart Lease”
does not benefit consumers with cash-flow problems.2
Another interesting example is that of Mitsubishi Mo-
tors, which has developed a loan financing program
for buyers of its cars in which the low monthly pay-
ment benefit of leasing is replicated (Wall Street Journal
1995). As this example shows, if there is demand for
financing with low monthly payments, auto manufac-
turers or financial institutions can create loan products
1
This problem in the automobile industry is different from that an-
alyzed by Purohit and Staelin (1994) and Purohit (1997), in which
the focus is on the competition between rental agencies and car deal-
ers. The problem we address in this article focuses on the leasing
and selling decision from a single source.
2
Interestingly, the Wall Street Journal (1995) reports that about 31%
of Cadillac leases were single payment leases.
43
 DESAI AND PUROHIT
Competition in Durable Goods Markets
that meet this demand. Finally, the affordability ar-
gument mostly ignores the effects of competition and
possible strategic behavior by the manufacturers.
The relationship between leases (or rentals) and
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
sales has long been studied by academics. Beginning
with Coase’s conjecture (Coase 1972), the durable
goods literature has shown that leasing is more prof-
itable than selling for the manufacturer (see, for ex-
ample, Stokey 1981, Bulow 1982, Gul et al. 1986). These
papers show that leasing is more profitable mainly be-
cause it allows the manufacturer to charge higher
prices over time. Intuitively, when the manufacturer
sells its product to current consumers, it has an incen-
tive to reduce the price in future periods to skim the
market and attract any remaining consumers. There-
fore, strategic consumers who anticipate lower prices
in the future, may decide to wait until prices come
down. In the most extreme case, this waiting can force
the monopolist to lower its price immediately and sell
at the lowest possible price (see, for example, Tirole
1988).
On the other hand, if the firm is leasing, it has less
incentive to reduce price in the future because the
“cost” of lower prices will be borne by the owner of
the product, which in the case of leases is the manu-
facturer. Thus, there is no point in hurting itself and
the monopolist keeps prices at a high level. One can
see that if the monopolist could commit not to lower
its prices over time, then a sales strategy could work
just as well as a leasing strategy. However, this com-
mitment is not time consistent—that is, after getting
consumers to purchase its product at a high price, the
monopolist has an incentive to go back on its word and
lower prices to sell to the remaining customers. Con-
sumers recognize the monopolist’s problem with time-
consistency and hence do not believe any promises.3
While the theoretical rationale developed above is
appealing intuitively, it is not consistent with the fact
that for almost all consumer durables, selling is ob-
served greater proportions than leasing.4 Bucovestky
3
If the monopolist could make its commitment not to lower prices
credible, then a sales strategy can work as well as a lease strategy.
Stores often achieve this by promising to refund price differences if
they lower prices in the 30 days following a purchase.
4
It has also been argued that consumer moral hazard may reduce
44
and Chilton (1986) shows that a monopolist facing a
threat of entry would sell some units to deter entry. By
selling, rather than leasing, to consumers in the first
period, the monopolist incumbent locks some consum-
ers (first-period buyers) out of the market and the en-
trant does not have access to them in period 2. How-
ever, because there is no competition in the first period,
it is not clear whether selling dominates only because
it can deter entry. Our paper adds the next step by
allowing competition in both periods such that the
lease versus sell decision is based on considerations
other than locking out customers.
In a related paper, Bulow (1986) models durability
choice in an oligopoly and finds that the lease/sales
ratio is dependent on the number of firms in the mar-
ket. However, Bulow’s model is such that durability
and the sales/rental are equivalent choices. In our pa-
per, we treat durability as being determined exoge-
nously; this allows us to examine how the sales/rental
ratio varies not only with the degree of market com-
petitiveness but also with the inherent durability of the
product. In addition, Bulow (1986) models durability
as deterioration that eliminates a certain percentage of
units in future periods. Thus, used units in the market
are not differentiated from new ones and, hence, both
command the same price. On the other hand, we
model deterioration such that each product wears out
over time, with the result that used products are dif-
ferentiated from new ones. This characterization of de-
terioration is more appropriate for markets where new
and used goods are differentiated and have different
market prices, e.g., automobiles.
In order to study firms’ incentives to lease, we de-
velop a two-period setting where two manufacturers
of differentiated durable goods compete by choosing
their optimal leasing and selling strategies. We begin
our analysis by first considering the case of a monop-
olist. This case replicates earlier findings in this area
and serves as a useful benchmark for the subsequent
case that allows competition in the market. To model
competition, we consider a symmetric duopoly and
the attractiveness of leasing. However, as Bond and Samuelson
(1984) and Desai and Purohit (1998a) show, under certain conditions,
deterioration may actually benefit the firm.
Marketing Science/Vol. 18, No. 1, 1999
 DESAI AND PUROHIT
Competition in Durable Goods Markets
derive the equilibrium fraction of leasing. We find that
depending on the degree of competitive rivalry and
the deterioration of cars, either pure selling or a mix
of leasing and selling is the optimal strategy for the
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
firms. Importantly, we find that neither firm leases all
the units in equilibrium. Although leasing helps keep
prices at higher levels, it is not as competitive as sales.
If a firm leases all cars in a competitive market, its
prices will be too high and its market share too low.
Therefore, firms never lease all their cars in a compet-
itive equilibrium. In fact, we find that as the degree of
rivalry between the two firms increases, the equilib-
rium fraction of leased cars decreases. Eventually,
when the rivalry increases beyond a certain level, then
the firms do not lease any units.
We also find that the equilibrium fraction of leased
cars decreases as cars deteriorate more. The reason is
that as cars deteriorate more, the second period reve-
nue from used cars becomes less significant, and the
time consistency problem becomes less severe. This in-
duces the firm to reduce the fraction of leased cars. We
then relax the symmetry assumption and allow the de-
terioration of the two manufacturers’ cars to be differ-
ent. We find that all results of the symmetric duopoly
extend to the asymmetric case. Finally, we test the im-
plications of our model with data from the automobile
market. We are able to show that the extent of leasing
in the automobile industry is strongly dependent on
the degree of competitive intensity in various seg-
ments of the market and on the deterioration rate of
the car model.
The remainder of the paper is organized as follows.
In § 2, we describe our model. Sections 3 and 4 detail
the monopoly and symmetric duopoly analyses, re-
spectively. In § 5, we generalize our model to consider
different deterioration rates for the two manufacturers.
We empirically test our theoretical predictions in § 6
and conclude the paper in § 7.
2. Model
In this section, we describe the basic setup of our
model and lay out our assumptions regarding the
product, the manufacturers, and the consumers. To
motivate the marketing issues associated with leases
and sales, we choose the auto industry as our running
Marketing Science/Vol. 18, No. 1, 1999
example. However, the reader should note that our
analysis applies readily to any durable category where
products can be leased or sold.
Our interest is in investigating firms’ incentives to
lease or sell its products. Clearly, on the demand side,
if one segment of consumers prefers leasing and an-
other prefers buying, then firms have incentives to
make both options available to the market. However,
in keeping with the durable goods literature, our ob-
jective is to investigate a firm’s rationale in choosing
an optimal mix of leasing and selling and to under-
stand how it is affected by the nature of competition
in the market and the embedded quality in the prod-
uct. By developing a model in which consumers are
indifferent between buying or leasing, we are able to
control for differential consumer preferences and focus
solely on the effect of market competitiveness and
product characteristics.
2.1. Product
The product that we consider is a durable, which we
refer to as a car. To simplify the analysis, we assume
a constant marginal cost, c 4 0, of producing a car.
Because of the durable, or long-lived, nature of cars,
we incorporate multiple periods in our model. While
the amount of time is not crucial, it is important to
assume that cars last more than one period and that
they have finite lives. A two-period structure allows us
to study dynamic issues while retaining tractability.5
Thus, we model the useful life of a car into two periods:
The car is new when it is marketed in period 1, and
the same car is classified as used in period 2. To model
the differentiation between new and used cars, we as-
sume that cars produced in period 1 deteriorate by a
factor d, 0 # d # 1, at the end of period 1. If the cars
do not deteriorate, then d 4 0 and new and used cars
are identical; on the other hand, d 4 1 means that the
car has fully deteriorated in one period. Deterioration
can also be interpreted as the embedded quality of the
car. Although different cars may have different ds, it is
important to note that these ds are exogenous to the
system.
5
We have also analyzed the model under a three-period structure.
Although this structure makes parts of the model intractable, the
essential results are not affected. Details are provided in Desai and
Purohit (1998b).
45
 DESAI AND PUROHIT
Competition in Durable Goods Markets
Finally, it is important to distinguish between dete-
rioration and depreciation. While depreciation and de-
terioration are used interchangeably by many people,
within the context of our work, they have very specific
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
meanings. We define deterioration, d, as a measure of
the physical decay of the product over time, and de-
preciation as a measure of how the market price of the
good changes over time. However, because d influ-
ences the depreciation rate, these two variables are not
independent—holding all else constant, we would ex-
pect a car that deteriorates rapidly (i.e., has a higher d)
to also depreciate rapidly. Thus, although the extent of
depreciation of a used good is affected by the product’s
deterioration, it is also affected by other factors such
as manufacturers’ prices of new goods.
2.2. Manufacturer Strategies
We consider two competing manufacturers, each sell-
ing a single car model. We denote the focal manufac-
turer by small letters and its competitor by capital let-
ters. The problem of each manufacturer is to choose
the optimal number of units to market in periods 1 and
2, and the optimal mix of leases and sales in period 1.
Thus, in period 1, each manufacturer chooses its opti-
mal first period quantity (q1n or Q1n), and the fraction
of cars it wants to lease (f or F). That is, the focal man-
ufacturer leases fq1n units and sells (1 1 f)q1n units. A
lease contract in our model is for one period. At the
end of the lease contract, consumers return the cars to
the firm that then re-sells them in the open market.
Note that we are not precluding any consumers from
purchasing the car at the end of the lease period. In
fact, the same customer could conceivably lease the car
and then purchase it from the manufacturer.
Since our model ends in the second period, there is
no difference between selling and leasing in that pe-
riod. Therefore, new cars are “sold” at l2n, and old cars
at l2u. While we recognize the limitation of this struc-
ture, we also note that any finite period structure will
have this property.6
2.3. Consumer Demand
Consumer demand in our model has to include three
essential aspects of the market for cars. In particular,
6
We have derived the results in a three-period model as well as in a
model with an initial stock of used cars and an initial fraction of
leased cars. These alternate approaches complicate the analysis but
leave the basic results unchanged (Desai and Purohit 1998b).
46
the model should (1) capture the competition between
new and used cars; (2) capture the competition be-
tween the two manufacturers; and (3) have both lease
and sales prices. We follow the lead of earlier research
by Purohit and Staelin (1994) and incorporate these
three conditions through our postulated demand func-
tions. Let qij and Qij be the quantities of cars made by
two manufacturers, and lij and Lij be the one-period
prices where j 4 u, n denotes the type of car (used or
new, respectively) and i 4 1, 2 denotes period (one or
two). Then, each firm’s demand functions are given by:
Focal Firm:
l1n 4 a 1 q1n 1 eQ1n,
l2n 4 a 1 q2n 1 cq2u 1 e(Q2n ` cQ2u),
l2u 4 c(a 1 q2n 1 q2u 1 e(Q2n ` cQ2u)), (1)
Competitor:
L1n 4 a 1 Q1n 1 eq1n,
L2n 4 a 1 Q2n 1 cQ2n 1 e(q2n ` cq2u),
L2u 4 c(a 1 Q2n 1 Q2u 1 e(q2n ` cq2u)), (2)
where c 4 1 1 d represents the degree of substituta-
bility between new and used cars, and 0 , e , 1 rep-
resents the degree of competition between the two
manufacturers’ cars.
Recall that the problem requires us to include two
levels of competition: between new and used cars, and
between the two manufacturers. To capture the asym-
metric competition between new and used cars, as in
Purohit and Staelin (1994), we treat used cars as infe-
rior substitutes for new cars. In particular, c , 1 cap-
tures the basic notion that new cars are stronger com-
petitors for used cars than used cars are for new cars,
e.g., an additional unit of a used car decreases used car
prices more than it decreases new car prices. Note that
when c 4 1 then the cars do not deteriorate and the
distinction between used and new is moot.
The competition between the manufacturers is cap-
tured by the parameter 0 , e , 1, which captures the
manner in which quantity decisions by one manufac-
turer affect the demand for the other manufacturer’s
product. Our treatment is similar in spirit to demand
functions employed in McGuire and Staelin (1983),
Marketing Science/Vol. 18, No. 1, 1999
 DESAI AND PUROHIT
Competition in Durable Goods Markets
Coughlan and Wernerfelt (1989), Lal (1990), and sev-
eral other papers in this stream. As in these earlier pa-
pers, the parameter e represents the degree of substi-
tutability between the two manufacturers’ products.
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
The higher the value of e, the more substitutable are
the two manufacturers’ products and the more intense
is the competition between them.
The final aspect of the car market that we need to
capture is the presence of lease and sales prices. We do
this by noting that all car prices (lease or sale) should
reflect the fact that consumers value cars for the ser-
vices they provide. Because a car provides a stream of
services for two periods, we can think of a price in each
period to reflect the services provided during that pe-
riod. Thus, a car that is new in period 1 provides “new-
car” service in the first period and provides “used-car”
service in period 2. The two distinct services can be
priced differently, and we refer to them as one-period
prices in period 1, l1n, and in period 2, l2u.
Because lease contracts are for one period, then a
consumer who leases a car in period 1 only pays a one-
period price of l1n. On the other hand, a consumer who
buys the car in period 1 gets the ownership rights for
two periods. Therefore, the selling price of a car, p1n,
is the discounted sum of two one-period prices—the
new car price in period 1 and the expected used car
price in period 2. Because we assume that consumers
in our model are fully informed and have rational ex-
pectations, p1n 4 l1n ` ql2u, where l2u is the used car
price in the second period and 0 # q # 1 is the discount
factor. This also ensures that consumers are indifferent
between leasing and selling, which allows us to focus
on the firms’ incentives to lease and sell.7 Finally, note
that consumers who lease as well as those that pur-
chase both pay l1n for period 1 services of the car. This
assumption is consistent with institutional realities.
Consumers can always walk into an auto dealership
7
The assumption of fully informed consumers may be stronger than
necessary. In particular, note that consumers can always observe the
sales and lease price in period 1, from which they can infer the re-
sidual value of the used car in period 2. Further, because consumers
are indifferent between leasing and selling, the residual value pre-
dicted in period 1 has to be exactly the same as the used car price
that will prevail in period 2. Note that in the absence of this condi-
tion, there are significant arbitrage opportunities for the consumers
and the firm.
Marketing Science/Vol. 18, No. 1, 1999
and negotiate a price without revealing whether they
want to buy or lease the car. They can let the dealership
know their intent to buy or lease after the price is
agreed upon by both parties.
In summary, our postulated demand system reflects
the belief that as the quantity of a given type of car
increases, the prices of all types of cars should drop.
These reduced form demand functions exhibit prop-
erties that are intuitively appealing. First, price (lij or
Lij) decreases as one of the firms sells more units. Sec-
ond, price responds more to changes in its own quan-
tity than to the competitor’s quantity. Third, price in
the second period is affected by quantity decisions in
period 1; in particular, a firm’s period 2 price increases
as it (or its competitor) sells fewer units in the first
period. Fourth, the used car demand functions natu-
rally model depreciation as the loss in value of a car
as it deteriorates with time. Note that if the used cars
deteriorate completely, the above equations ensure
that their price falls to zero. As used cars get better
(deteriorate less), their price rises. Finally, from the in-
verse demand functions for new cars, we can also ob-
serve that as the rate of deterioration falls and used
cars become stronger competitors for new cars, used
car quantities have a greater impact on the prices of
new cars.
3. Monopoly Revisited
In this section, we briefly discuss a monopolist’s prob-
lem of determining the optimal balance between leas-
ing and selling. Several papers have analyzed the du-
rable goods monopolist’s problem and have found that
the monopolist’s optimal strategy is to lease its entire
production (see, for example, Bulow 1982). Here we
re-analyze the monopoly case for two reasons: (1) to
establish a benchmark case where the monopolist
leases the entire quantity, and (2) to highlight the im-
plications of leasing in the context of our model.
The basic model is solved by first determining the
firm’s optimal quantity choice in period 2, and then
solving the problem in period 1. This ensures that our
solutions are time-consistent and subgame perfect.
3.1. Period 2 Analysis
In the absence of competition, following Equations (1)
and (2), the second-period inverse demand functions
are as follows:
47
 DESAI AND PUROHIT
Competition in Durable Goods Markets
l2n 4 a 1 q2n 1 cq2u, (3)
l2u 4 c(a 1 q2n 1 q2u). (4)
Noting that q2u 4 q1n, we can write the firm’s second-
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
period profit, p2, as follows:
p2 4 q2nl2n ` fq2ul2u, (5)
where the first part reflects the profits from selling q2n
new cars, while the second part reflects the profits from
selling fq2u ex-leased cars. Note that the remaining (1
1 f)q2u used cars are available on the secondhand mar-
ket that is not under direct control of the firm. Solving
the firm’s first-order condition, we get its optimal
second-period quantity, q*2n, as
a 1 cq1n(1 ` f )
q*2n 4 , (6)
2
where the superscript * denotes the optimal value.
An important characteristic of the optimal second-
period quantity is that it decreases as the fraction of
cars leased in the first-period increases. The reason can
be understood from (5). The firm is selling two substi-
tute products in period 2 (used cars and new cars). An
increase in q2n decreases the new car price and also
decreases the used car price. As the fraction of leased
cars (f ) increases, the reduction in used car price hurts
the firm more. Therefore, the higher the fraction of
leased cars, smaller are the firm’s incentives to increase
the second-period quantity. Put differently, a higher
fraction of leased cars allows the firm to obtain higher
prices for both new and used cars in the second period.
3.2. First Period Analysis
Given the optimal solution for the period 2 problem,
we can now look at the firm’s first-period decisions.
The firm decides the first-period quantity and the frac-
tion of leased cars, keeping in mind the second-period
outcomes described above. The first-period prices are
given by:
l1n 4 a 1 q1n, (7)
p1n 4 l1n ` ql2u. (8)
The firm’s first-period profit is p1 4 fq1nl1n ` (1 1
f)q1np1n, where the first part reflects the profits from
leases and the second part the profit from sales. The
firm’s problem is to maximize the discounted sum of
48
two period profits, p 4 p1 ` qp*2 , by choosing the op-
timal quantity and leasing fraction. This yields:
f * 4 1, (9)
a
q*1n 4 . (10)
2(1 ` cq 1 c2q)
Thus, we see that the optimal strategy for the firm
is to lease the entire quantity of cars. While we do not
explicitly model consumers’ buying or leasing deci-
sions, the intuition for the above result is similar to the
time consistency problem associated with selling. Note
that holding used car prices constant, a car produced
in the first period gives the firm the same revenue
stream whether it is leased or sold. If the car is leased,
the firm gets l1n in the first period and l2u in the second-
period, bringing the net present value to l1n ` ql2u. If
the car is sold, the firm gets p1n 4 l1n ` ql2u in the first
period. Therefore, for a given value of l2u, the primary
factor driving the lease-sell balance is the profitability
of new cars sold in the second period. As noted earlier,
a higher value of f allows the firm to obtain a higher
price for the second-period new cars. Therefore, the
firm benefits by a higher value of f. In addition, a
higher value of f also helps the firm to support a higher
value of l2u, which in turn improves the profitability of
first-period cars.
It is well known that if the monopolist can commit
to the second period quantity in the first period itself,
pure selling (f 4 0) is as profitable as pure leasing (f
4 1). We also replicate this result in our model.8 An-
other interesting aspect of our monopoly results is the
role played by deterioration on the second period
quantity. Although we begin with reduced form de-
mand functions, our results are identical to the stan-
dard solution of the durable goods monopolist’s prob-
lem. This adds validity to our model and also
establishes that if we observe f , 1 in the duopoly case,
the decrease in the leasing fraction can be attributed
only to the existence of competition.
4. Competition Model
In this section, we consider the case where two firms
compete over two periods in a market for a differen-
tiated durable product. Comparing these results from
8
Details are in Desai and Purohit (1998b).
Marketing Science/Vol. 18, No. 1, 1999
 DESAI AND PUROHIT
Competition in Durable Goods Markets
those in the preceding section allows us to focus spe-
cifically on the effects of competition in durable goods
markets. As before, we first solve the period 2 problem
and then the period 1 problem.
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
4.1. Second-Period Analysis
In the second period, using the demand functions laid
out in Equations (1) and (2), both firms simultaneously
choose their optimal second period quantities. For the
focal firm, the problem is to maximize profits in period
2, p2 4 l2nq2n ` l2u fq1n, by choosing the optimal second-
period quantity, q2n. The competing firm has a similar
(symmetric) problem and is not presented here.
Solving both competitors problems simultaneously
yields:
2n 4
q**
a(2 1 e) 1 cq1n (2f ` 2 1 e2) 1 ecQ1n (1 1 F)
, quantity in a duopoly than in a monopoly.
4 1 e2
(11)
2n 4
Q**
a(2 1 e) 1 cQ1n (2F ` 2 1 e2) 1 ecq1n (1 1 f)
.
4 1 e2
(12)
There are several interesting characteristics of the
above solution. In particular,
Proposition 1. As a firm’s fraction of leased cars, f,
increases, the optimal quantity of new sales in period 2 de-
crease, ]q*2n/]f , 0. In addition, this effect increases with
the degree of competition between the competitors.
First, note that as in the monopoly case, we find an
intertemporal effect of leasing decisions, where the op-
timal second-period quantity decreases as the own
fraction of leased cars increases (see Equation (11)).
However, as Equations (11) and (6) show, the fraction
of leased cars has a stronger impact on the second pe-
riod quantity when there is competition. To see this
clearly, observe that
]q** (12cq1n)
2n
4 .
]f (4 1 e2)
As e increases, the magnitude of the partial deriva-
tive increases. In other words, the effect of f on the
second-period quantity gets amplified in the presence
Marketing Science/Vol. 18, No. 1, 1999
of competition. The reason is that a change in f affects
q*2n in two ways. First, there is a direct effect similar to
the effect discussed in the monopoly case: As the frac-
tion of leased cars increases, the firm needs to sell a
higher quantity of ex-leased cars in period 2. This
forces the firm to restrict the quantity of new cars in
the second period to obtain higher prices for both
products. The second, indirect effect, comes through
the competitor’s choice of its second-period quantity.
From Equation (11), note that an increase in f reduces
the focal firm’s optimal quantity q** 2n . This in turn in-
creases the competitor’s optimal quantity Q** 2n as de-
scribed in Equation (12), but an increase in Q2n further
reduces q2n. As the level of competition, e, increases,
this indirect effect becomes stronger. Therefore, the net
effect of these two effects is that the fraction of leased
cars has a stronger effect on the second-period optimal
Another interesting characteristic of the above so-
lution is that the focal firm’s optimal second period
quantity increases as the competitor’s fraction of leased
2n /]F . 0. The reason is that as the
cars increases, ]q**
competitor’s fraction of leased cars increases, the com-
petitor’s optimal second-period quantity decreases,
which in turn increases the focal firm’s optimal second-
period quantity (see Equations (11) and (12)).
Deterioration plays an interesting and intuitive role
in the firm’s choice of second-period quantity. In
particular,
]q**
2n 1eQ1n(1 1 F) 1 q1n(2 ` 2f 1 e2)
4 . (13)
]c (4 1 e2)
That is, as cars deteriorate less (recall that c 4 1 1 d),
the firm chooses a smaller second-period quantity. The
reason is that a lower deterioration rate implies that
new and used cars are less differentiated and more
substitutable. As used and new cars become closer sub-
stitutes, they affect each other’s prices more and the
firm needs to support higher new car prices by choos-
ing a smaller quantity.
4.2. First-Period Analysis
We now look at both firms’ first-period problems. In
the first period, each firm chooses its optimal first-
period quantity and the fraction of cars it wants lease.
49
 DESAI AND PUROHIT
Competition in Durable Goods Markets
In making these decisions, each firm takes into account
the effect of the first-period decisions on the second-
period behavior of both firms described above. If the
focal firm chooses a quantity q1n and a fraction of
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
leased cars f, its first-period profit is given by p1 4 (1
1 f)p1nq1n ` fl1nq1n and the total profit of the focal firm
is p 4 p1 ` qp*2 . Simultaneously solving both firms’
maximization problems, we get
e2(2 ` e 1 2cq ` 2c2q)
f** 4 Max 0, 1 1 5 c(4 ` 2e 1 e2)
, 6
1n 4
q**
5
a tition, each firm has two conflicting objectives. One is
2 ` e ` 2cq 1 2c2q
for f** . 0,
a(8 ` 4e 1 2e2 1 e3 ` e2cq)
(2 1 e)(2 ` e) ` 2cq(2 1 e)(2 ` e)2 1 c2q(12 ` 6e 1 3e2 1 2e3)
3
for f** 4 0.
In Equation (14), when
e2 (2 ` e 1 2cq ` 2c2q)
. 1,
(4 ` 2e 1 e2)c
the optimal value of the lease fraction is zero, and the
focal firm prefers to sell all its units. We discuss this
condition in more detail in the next section. Given the
symmetry between two firms, the competitor’s first-
period choices are also given by Equations (14) and
(15).
Proposition 2. In the presence of competition, each
firm’s optimal strategy is to choose either a combination of
selling and leasing or only selling.
Proof. In Equation (14), f ** , 1 for any e . 0. M
Proposition 2 shows that when there is competition
in the market, each firm is better off choosing a com-
bination of leasing and selling rather than only leasing.
The primary reason for the difference between the re-
sults in a monopoly and a duopoly is a result of the
difference in the second-period dynamics. As one
would expect intuitively, each firm would like to
charge a lower price in a duopoly than in a monopoly.
In our quantity choice model, this translates into firms
preferring a higher quantity in duopoly than in mo-
nopoly. Since the second-period quantity decreases
50
with the fraction of leasing, each firm prefers a lower
fraction of leasing in duopoly than in monopoly. Put
differently, an all leasing strategy results in too low a
second-period quantity and too high a price for the
firm’s new cars in the second period. Therefore, the
firm prefers a mix of leasing and selling and chooses a
fraction of leased cars that is strictly less than 1. In fact,
under certain conditions, the firms are better off selling
all their units.
(14) Proposition 2 also provides an explanation for why
we rarely observe manufacturers of durable goods
leasing their entire output. In the presence of compe-
to ensure time consistency (and managing demands
over time) and the other is to be sufficiently competi-
tive against its rival. While leasing helps the first ob-
jective, it hurts the second objective. Therefore, de-
pending on the degree of competition (e) and the
(15) goodness of used cars (c), each firm adopts a mix of
leasing and selling.
Our results are also interesting in that they theoret-
ically show that all three regimes—only selling, a mix
of leasing and selling, and only leasing—are possible
for different values of parameters.
4.3. The Effect of Competitive Intensity on
Leasing
We now discuss how the equilibrium fraction of leased
cars changes as the rivalry between the two firms in-
tensifies. In our model the competition between the
two firms is measured by the parameter e. As the two
firms’ products become more similar, the competition
between them intensifies and each firm’s choices affect
the other firm’s prices more. Therefore, a higher degree
of competition is captured by a higher value of e in our
model.
Proposition 3. Each firm’s optimal fraction of leasing
decreases as the competition between the firms intensifies,
]f**/]e , 0 and ]F**/]e , 0. If the competitive intensity
crosses a threshold of
e2(2 ` e 1 2cq ` 2c2q)
e* 4 ,
(4 ` 2e 1 e2)c
then each firm stops leasing and only sells its product.
Proof. See the appendix. M
Marketing Science/Vol. 18, No. 1, 1999
 DESAI AND PUROHIT
Competition in Durable Goods Markets
As we noted earlier, in order to maximize profits
over both periods, each firm adopts a mix of leases and
sales in the first period. In addition, the higher the frac-
tion of leases in period 1, then all else equal, the higher
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
is the market clearing price of new cars in period 2.
This implies that a higher fraction of leasing in one
period can put the firm at a competitive disadvantage
in the subsequent period. Therefore, the firm has to
balance the advantages of leasing a higher fraction
now and staying competitive in the future. Because it
can become more competitive (in terms of prices) if it
has a smaller fraction of leases in period 2, we find that
as the competitive intensity, e, increases, the firm re-
duces its fraction of leases.
Our results suggest a particular relationship be-
tween competitive intensity and the extent of leasing
that occurs in any market. While we give details of our
formal empirical test in the next section, below we dis-
cuss some anecdotal evidence that is consistent with
our results.
• In the automobile market, more expensive cars
typically are leased more than less expensive cars.
While leasing always involves a smaller cash flow than
buying, our results indicate that there may be more to
leasing than mere affordability. More expensive luxury
cars (e.g., BMW 7-Series or Lexus LS-400) are perhaps
sold more on nonprice attributes and have less intense
price competition among them than lower priced cars
such as Chevrolet Cavalier and Ford Escort (see, for
example, Bresnahan 1981). Our Proposition 3 suggests
that more expensive cars may be leased much more,
not only for affordability reasons but also because they
face less intense competition.
• Another interesting product category in this con-
text is telephone equipment (customer premises equip-
ment). Various government agencies and courts had
been chipping away at AT&T’s monopoly in the tele-
phone equipment market. In 1951, the famous Hush-
A-Phone case established customers’ right to own
equipment. While there was some competition in the
equipment market in the late 1970s, the biggest change
in the market occurred after the AT&T break-up in
1983 when both the RBOCs and AT&T were allowed
to market phones. At this stage, AT&T started selling
phones aggressively and the fraction of leased phones
Marketing Science/Vol. 18, No. 1, 1999
decreased drastically.9 Finally, note that while the local
phone companies use a mix of leasing and selling, al-
most all other firms rely entirely on selling. Our inter-
pretation of this phenomenon is that the introduction
of competition (and gradual increase in competition)
led to a decline in the fraction of leased phones.10
• Menezes and Serbin (1993) describe the U.S. pho-
tocopier market in 1989. They note that the fraction of
leased units was higher at the high end of the market.
They also note that in the high end, the competition
was primarily on product features and service, and the
top two competitors, Xerox and Kodak, together ac-
counted for about 76% of the market. If we take the
dominance of Xerox and Kodak as evidence of less
competition in the high-end market, then our result
explains why about 80% of all units were leased in this
market.
4.4. The Role of Deterioration
As Equation (14) shows, the level of deterioration also
influences the optimal fraction of leased cars. Propo-
sition 4 describes how the optimal fraction of leased
cars changes with changes in the deterioration level.
Proposition 4. As deterioration increases, each firm’s
optimal fraction of leasing decreases, ]f**/]c . 0 and ]F**/
]c . 0.
Proof. See the appendix. M
Proposition 4 states that firms progressively reduce
their leasing fraction as cars deteriorate more (i.e., c
increases), to the point where the firm chooses not to
lease any units. Note from Equation (14) that if the
value of c exceeds a critical level, then the fraction of
leasing goes to zero and the firm sells its entire output.
The optimal fraction of leasing is primarily determined
by trade-off between two factors: (1) inter-temporal
problem of ensuring time consistency, and (2) within-
period competition between the two firms. A higher
leasing fraction helps ensure time-consistency and
may result in a loss of market share to the competitor.
As used cars deteriorate more, their prices fall and
9
Humorist Russell Baker worries about the problem of buying
phones in his column in New York Times, January 3, 1983.
10
More details about the early competition in the CPE market are
available in Brenner 1992 (pp. 216–228). See also Shooshan (1984).
51
 DESAI AND PUROHIT
Competition in Durable Goods Markets
firms derive less revenue from them. As a result, the
second period becomes less significant and the time
consistency problem becomes less important. This
changes the firms’ incentives to favor selling more.
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
Proposition 4 also raises interesting issues about the
affordability argument for automobile leasing. It is be-
lieved that cars with higher resale values are leased
more because the difference between buying and leas-
ing cash flows is greater for such cars. That is, cars with
higher resale values give a higher affordability benefit
to consumers. However, the problem with this argu-
ment is that any loan can be stretched over a longer
horizon or it can be more back-ended. In such a case,
the difference between leasing and selling cash flows
can actually diminish as resale values increase. Our
results provide another explanation for why we may
observe higher resale values cars being leased more.
The reason is that as used cars become better, the firms
derive higher revenues from the second period and the
time-consistency issues become more important,
prompting firms to lease more than sell.
5. Differences in Deterioration
In the previous section, we considered a symmetric du-
opoly where each manufacturer’s cars deteriorated at
the same rate. We now relax the assumption of sym-
metry and allow differences in the rates at which the
manufacturers’ cars deteriorate. This issue is important
to consider because of the belief that deterioration af-
fects consumers’ willingness to pay for cars. For ex-
ample, publications such as Consumer Reports give each
car a rating on predicted reliability which directly ad-
dresses the deterioration of the car. Presumably such
information can help consumers make more informed
decisions about the true cost of purchasing a car. Our
model focuses on the firm’s side of the above problem.
In particular, is the firm with a higher rate of deterio-
ration necessarily worse off than the firm with a lower
rate of deterioration? A priori, the answer to this ques-
tion is not entirely clear. Note that a lower deteriora-
tion rate implies higher prices for used cars in the fu-
ture, and hence, a higher current selling price.
However, a lower deterioration rate also increases the
future competition from used cars. To address these
issues, we let the focal firm’s deterioration (d 4 1 1 c)
52
be greater than the competitor’s deterioration (D 4 1
1 C). The other aspects of the model remain un-
changed. We study the asymmetric duopoly for two
additional reasons: (1) It allows us to verify that our
duopoly results are not solely due to the symmetry
assumption, and (2) we can examine whether or not
the firm with a higher deterioration rate has an incen-
tive to market a greater number of units.
With the differences in c and C, the relevant demand
functions are obtained from equations similar to Equa-
tions (1) through (10). For the sake of brevity, we give
more details in the appendix. Following the order of
the earlier sections, we first solve the problem in period
2. This yields:
a(2 1 e) 1 cq1n (2f ` 2 1 e2) 1 eCQ1n(1 1 F)
2n 4
q*** ,
4 1 e2
(16)
a(2 1 e) 1 CQ1n(2F ` 2 1 e2) 1 ecq1n (1 1 f)
2n 4
Q*** ,
41 e2
(17)
where the superscript *** denotes optimal values in the
present case.
The optimal second-period quantities are very simi-
lar to those in the symmetric duopoly, with one im-
portant difference: Other things being equal, the focal
firm’s first period decisions (the first-period quantity
and the fraction of leased cars) have a smaller impact
on the competitor’s second-period quantity than the
competitor’s first-period decisions’ impact in the focal
firm’s second-period quantity. The reason is that, all
else being equal, the competitor’s used cars are “bet-
ter” than the focal firm’s used cars: C . c.
In the first period, each firm chooses its optimal frac-
tion of leased cars and the first-period quantity. We
derive the equilibrium values by simultaneously solv-
ing each firm’s maximization problem. Table 1 gives
the details of the solution.
Proposition 5. In the presence of competition and
asymmetric substitutability, neither firm leases its entire
output. Both firms’ equilibrium fractions of leased cars in-
crease as (i) the level of competition between firms decreases,
and (ii) the own rate of deterioration decreases.
Proof. See the appendix. M
Thus, we are able to generalize our results in a sym-
metric duopoly (Propositions 2, 3, and 4) to the case of
Marketing Science/Vol. 18, No. 1, 1999
 DESAI AND PUROHIT
Competition in Durable Goods Markets

Table 1 Equilibrium Values in the Asymmetric Duopoly

Fraction of Leased Cars First Period Quantity

14e2 ` e4 ` 4e2 (C 1 1)Cq ` c(8 1 4e2 ` e3 1 4e2q ` 4e2cq) ` w1 a[2(1 ` Cq 1 C2q) 1 e]

Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.

Focal Firm
c(4 ` 2e 1 e2)(2 1 e ` 2Cq 1 2C2q) e2 ` 4(1 ` cq 1 c2q)(1 ` Cq 1 C2q)

Competitor 14e2 ` e4 ` 4e2 (c 1 1)cq ` C(8 1 4e2 ` e3 1 4e2q ` 4e2Cq) ` w2 a[2(1 ` cq 1 c2q) 1 e]
C(4 ` 2e 1 e2)(2 1 e ` 2cq 1 2c2q) e2 ` 4(1 ` cq 1 c2q)(1 ` Cq 1 C2q)

w1 4 2c(1 1 C)Cq(4 ` 2e 1 e2 1 2e2q ` 2e2cq),

w2 4 2C(1 1 c)cq(4 ` 2e 1 e2 1 2e2q ` 2e2Cq).
Note: The above values are for only those parameter values for which f *** . 0 and F*** . 0.

an asymmetric duopoly. The intuition for the above

results is similar to that in the symmetric duopoly case
and is not repeated here.
Comparing the equilibrium fractions of leased cars
for the two firms, we get the following result.
Proposition 6. The focal firm, with a higher rate of
deterioration, chooses a smaller fraction of leased cars than
the competitor who has a lower rate of deterioration.
Proof. See the appendix. M
The optimal fraction of leased cars is determined by
the trade-off between the competitive pressures and
the need for time-consistency. Because the firm with
the smaller rate of deterioration derives more revenues
from the second period, the time-consistency problem
is more serious for this firm. As a result, it chooses a
higher fraction of leased cars.
As pointed out earlier, the affordability argument is
often raised when we consider the extent of leasing in
the auto industry. To reiterate, this line of reasoning
argues that because a lease tends to have lower
monthly payments than a purchase, leases allow con-
sumers to afford more expensive cars than purchases.
Thus, the hypothesis is that the proportion of leases
should increase with the price of the car. Indeed, an-
ecdotal empirical evidence does bear this out. How-
ever, the question remains: Is affordability the reason
why more expensive cars are leased more often? We
answer as follows.
Corollary. When manufacturers’ cars deteriorate at
different rates, then the higher priced car has a higher frac-
tion of leasing than the lower priced car.
Proof. See the appendix.
Thus, based on the model developed in this section,
we find that the extent of leasing is higher with higher
priced cars. The interesting aspect of this result is that
it occurs without having any credit-constrained con-
sumers who prefer leasing to buying. In our model,
leasing and buying are identical options for consumers
and no segment prefers one over the other. However,
even under this condition, we find that the more ex-
pensive car is leased more often. To understand this
result, note that (1) the firm with the higher percentage
of leasing also has the lower deterioration rate, and (2)
the lower deterioration rate allows the firm to charge
higher prices. Taken together, this leads one to the con-
clusion that leasing increases with the price of the car.
However, this conclusion—similar to the affordability
argument—misses the role of deterioration as an in-
tervening variable. This suggests to us that in consid-
ering the auto market, more expensive cars may be
leased more often primarily because of their higher
“quality” and not because of reasons related to
affordability.
6. Empirical Analysis
In this section, we test our theoretical results using
1993 U.S. passenger car and light truck market data.
Based on the propositions developed in the earlier sec-
tions, we have the following three testable hypotheses.
Hypothesis 1. In a competitive market, firms will
choose either a combination of leasing and selling or only
selling.

Marketing Science/Vol. 18, No. 1, 1999 53

 DESAI AND PUROHIT
Competition in Durable Goods Markets
Hypothesis 2. For any firm, as the level of competition
in the market increases, the fraction of leased cars decreases.
Hypothesis 3. For any firm, as its own rate of deteri-
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
oration increases, the fraction of leased cars decreases.
6.1. Data and Measures
We use 1993 calendar year data for the U.S. automobile
market. Our unit of analysis is a car line (plate), such
as Chevrolet Corsica, Ford Taurus, Honda Accord, etc.
We use the following measures collected from leading
sources of automobile data.
Fraction of Leased Cars. The fraction of leased cars
is the percentage of a model’s production that was
leased to consumers. Note that this fraction is at the
model and not the manufacturer level, e.g., 34.7% of
1993 Ford Tauruses and 27.9% of Ford Mustangs were
leased. These data are provided to us by CNW/Mar-
keting Research.
Deterioration. For our measure of deterioration of
1993 model year cars, we use predicted reliability from
Consumer Reports. Predicted reliability for each model,
reported by Consumer Reports in its 1993 auto issue, is
based on frequency-of-repair records of earlier vin-
tages of the same model. For example, the predicted
reliability of a 1993 Ford Mustang is based on the re-
pair history of 1988–1992 Mustangs. Thus, predicted
reliability is the expected reliability based on the
model’s past performance. We also use an alternate
measure for deterioration from Consumer Reports. This
variable, trouble index, is an aggregate measure of how
well the car has performed over time. We obtain this
measure from Consumer Reports 1996 auto issue after
the 1993 models have been in use for three years. Thus,
trouble index is the actual deterioration at the end of
three years. Since our analysis is essentially an ex-ante
one, reliability is closer to our deterioration than is
trouble index which is an ex-post revelation of a car’s
deterioration. On the other hand, using the trouble in-
dex in our model can also test whether consumers have
rational expectations about their cars’ deterioration
rates.
Level of Competition. We first divide all car and
light truck lines into a number of submarkets or seg-
ments, using the classification scheme used by the Au-
tomobile News 1994 Market Data Book. For each of these
54
segments, we obtain unit sales and a manufacturer’s
market share within that submarket. We use the well-
known Herfindahl index to measure the extent of com-
petition within each segment. For a submarket with n
products having market shares mi (i 4 1, . . . n), the
Herfindahl index, H, is given by
n
H4 o
i41
mi2.
Note that H is higher for markets with higher concen-
tration (lower competitive intensity) and is lower for
markets with lower concentration (higher competitive
intensity). As another measure of competition within
a segment, we also use the total market share of the
top three cars in a submarket.
6.2. Empirical Model
Since the first hypothesis does not need any formal
testing (none of the car lines leased its entire output),
we develop an empirical model that allows us to test
the other two hypotheses cross-sectionally. In partic-
ular, we estimate the following equation:
ln 11 1LEASE
LEASE2
4b 0 ` b1TRUCK
` b2REL ` b3HINDEX, (18)
where
TRUCK 4 1 for light trucks and 4 0 for passenger
cars,
LEASE 4 fraction of cars leased for a given car line,
REL 4 predicted reliability numbers for the car from
Consumer Reports,
HINDEX 4 Herfindahl index for a class of cars.
We test three alternative specifications in which we
replace REL by TROUBLE, the trouble index from Con-
sumer Reports, and replace HINDEX by TOP3, the total
market share of the top three cars in each submarket.
Since REL and TROUBLE, and HINDEX and TOP3 are
highly correlated, we cannot use them together. There-
fore we run separate regressions in which the two vari-
ables are sequentially replaced by the alternate mea-
sures. Table 2 gives summary statistics for these
variables.
Note the absence of a price term in our empirical
Marketing Science/Vol. 18, No. 1, 1999
 DESAI AND PUROHIT
Competition in Durable Goods Markets

model (Equation (18)). There are several reasons for
this. Since our objective here is simply to test our theo-
retical results, we do not use price as an explanatory
variable because our results do not describe any im-
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.
shows that the sample is well suited to test our other
two hypotheses.
The results of our analysis are presented in Table 3.
Overall, the data support our hypotheses. Both REL

pact of price on lease fractions.11 More importantly, (0.24, p , 0.01) and HINDEX (14.63, p , 0.01) have the
according to the papers in the durable goods literature, predicted sign and are statistically and economically
leasing allows a firm to keep its prices high. That is, significant, supporting Hypotheses H2 and H3. We
these papers show that the market price of a product also note that the negative estimate for TRUCK (11.15,
is a function of lease fraction and not vice versa. There- p , 0.01) suggests that the extent of leasing is lower in
fore, using price as an explanatory variable leads us to the truck market. Hypotheses H2 and H3 are also sup-
a serious endogeneity problem. Finally, in models with ported in the three other alternative specifications.
log-odds dependent variables, heteroskedasticity can Note that given the nonlinearity of our model, we can-
frequently be a concern. However, we have tested for not describe the effects of a unit change in HINDEX or
heteroskedasticity and have found it not to be a prob- REL on the fraction of cars leased. Finally, note that
lem. As a result, we estimate the above model by OLS we have also estimated the model by replacing TOP3
rather than WLS. first with TOP4 (total market share of the largest four
firms in the segment) and then with TOP5 (total mar-
ket share of the largest five firms in the segment). These
6.3. Results
results also support our hypotheses.
While Hypothesis H1 does not need a formal test, the
descriptive statistics given in Table 2 show that LEASE,
the fraction of leased cars, ranges from 0.5% to 86.2%,
with the mean being 21.53%. Figure 1 gives a histo-
7. Summary and Conclusions
While leasing is commonly believed to be simply a
gram of LEASE. As Figure 1 shows, all the cars in our
way of financing or a pricing arrangement, our anal-
sample employ a mix of leasing and selling. This is
ysis shows that a manufacturer’s choice of the leasing-
consistent with our Hypothesis H1. In addition, it also
selling balance has strong strategic effects in a com-
11 petitive market. These strategic effects arise because of
Can using price as an explanatory variable help us test the afford-
ability argument? The answer is no—the affordability arguments is the dual nature of the competition. First, there is com-
based more on lessees’ limited ability to obtain loans. Therefore, a petition between the firms. And second, there is com-
good test of the affordability argument necessarily involves com- petition between used and new products from the
paring the income, existing debts, and borrowing capacity of con- same firm. It is in handling both these types of com-
sumers who lease a car model with the corresponding variables of
petition that a leasing-selling strategy becomes useful.
consumers who buy the same car model.
The essential findings of this research are as follows:
• A competitive environment forces firms to adopt
Table 2 Summary Statistics
strategies where they only sell their products or they
use a combination of leasing and selling. We find that
Variable Mean Std. Dev. Minimum Maximum no firm chooses a strategy of only leasing, which is
what would occur in a monopoly. This occurs because
LEASE 21.5250 18.7024 0.5000 86.2000 a pure leasing strategy leaves room for the competition
In LEASE 11.6797 1.2998 15.2933 1.8320
to take a larger share of the market. A combination of
TRUCK 0.2558 0.4376 0.0000 1.0000
HINDEX 0.1034 0.0403 0.0572 0.2299
leases and sales allows the firm to be competitive and
TOP3 0.4611 0.1370 0.3114 0.9128 yet keeps prices relatively higher than a pure selling
TOP4 0.5508 0.1458 0.3805 0.9808 strategy.
TOP5 0.6222 0.1403 0.4480 0.9965 • The extent of leasing that occurs in a market de-
REL 3.1709 1.3220 1.0000 5.0000 pends on the degree of competitive intensity between
TROUBLE 3.8828 0.5465 3.0000 5.0000
the competitors. As the competitive intensity increases,

Marketing Science/Vol. 18, No. 1, 1999 55

 DESAI AND PUROHIT
Competition in Durable Goods Markets

Figure 1 Histogram of Lease Fractions Table 3 Regression Results

Regression 1:
Variable Estimate Standard Error t-statistic p-value
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.

INTERCEPT 13.5858 0.4471 18.0200 0.0001

TRUCK 11.1465 0.3498 13.2780 0.0013
REL 0.2427 0.0760 3.1940 0.0017
HINDEX 14.6350 3.8140 3.8370 0.0002

N 4 158, R 2 4 13.99, adjusted-R 2 4 12.31.

Regression 2:
Variable Estimate Standard Error t-statistic p-value

INTERCEPT 13.8469 0.8649 14.4480 0.0001

TRUCK 11.3144 0.3378 13.8910 0.0002
TROUBLE 0.3507 0.1834 1.9120 0.0578
HINDEX 12.4501 3.8118 3.2660 0.0014

N 4 145, R 2 4 11.84%, adjusted-R 2 4 9.96%.

Regression 3:
Variable Estimate Standard Error t-statistic p-value
we see the competitors decrease their level of leasing.
In extreme cases of competition, we find that the com- INTERCEPT 14.4530 0.5957 17.4750 0.0001
TRUCK 11.2781 0.3558 13.5920 0.0004
petitors choose to use pure selling strategies.
REL 0.2331 0.0751 3.1020 0.0023
• The extent of leasing is also related to the per- TOP3 5.3815 1.2842 4.1910 0.0001
ceived rate of deterioration of the product. As a prod-
N 4 158, R 2 4 15.41%, adjusted R 2 4 13.77%.
uct’s rate of deterioration decreases, the firm chooses
Regression 4:
to increase its level of leasing. Interestingly, the firm
Variable Estimate Standard Error t-statistic p-value
that has the higher level of deterioration chooses a
lower level of leasing. This occurs principally because INTERCEPT 13.9948 0.8968 14.4550 0.0001
the firm with the higher deterioration is, in a sense, TRUCK 11.3494 0.3497 13.8590 0.0002
less competitive; thus, it has to use a more aggressive TROUBLE 0.2873 0.1804 1.5920 0.1135
TOP3 3.6688 1.1423 3.2120 0.0016
sales strategy and less accommodating lease strategy.
• We verify the implications of our theoretical model N 4 145, R2 4 11.63%, adjusted R2 4 9.75%.
with data from the automobile market.
These results offer an explanation for why lease rates
are different across markets and why they change over
there is some evidence of this in the automobile mar-
time. These results can also explain why in a given
ket, it is less clear that this deception could occur in
market, some firms lease more units and others sell business environments, e.g., office copy machines,
more. While industry observers have argued for the computers, etc., where the lessors are more
affordability aspect of leasing, we show that it is the knowledgeable.
strategic benefit of leasing that may play a bigger role. We believe that an important arena for future work
We caution that in our model, consumers are fully in- in this area is to explore the consumer behavior impli-
formed; thus, from the perspective of a consumer, the cations of leasing and selling. In particular, we have
leasing and buying options are financially equivalent. not addressed issues related to adverse selection or
One could argue that in the real world, consumers are moral hazard that could arise in leasing situations. For
deceived by the fine print in lease contracts. While example, a greater proportion of bad drivers would

56 Marketing Science/Vol. 18, No. 1, 1999

 DESAI AND PUROHIT
Competition in Durable Goods Markets

choose leases (adverse selection) or once drivers lease l1n 4 a 1 q1n 1 eQ1n,
a car, they have less of an incentive, compared to buy- l2n 4 a 1 q2n 1 cq2u 1 e(Q2n ` CQ2u),
ers, to take care of the car (moral hazard). If these in-
l2u 4 c(a 1 q2n 1 q2u 1 e(Q2n ` CQ2u)), (A5)
deed are problems in the market, then the issue from
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.

Competitor:
the perspective of the firm is to understand how such L1n 4 a 1 Q1n 1 eq1n,
behavior affects firm profitability, and to develop con-
L2n 4 a 1 Q2n 1 CQ2n 1 e(q2n ` cq2u),
tracts that would minimize these issues from arising
in the first place.12 L2u 4 C(a 1 Q2n 1 Q2u 1 e(q2n ` cq2u)). (A6)
We first show that f *** , 1 for all values of e in the (0,1] interval.
f *** $ 1 when e 4 0 or when
Appendix Proofs
e . !4 ` 4cq 1 4c2q ` 4(1 1 C)Cq(1 ` cq ` c2q)
Proof of Proposition 3 or when
The equilibrium fraction of leased cars is given by e , !4 ` 4cq 1 4c2q ` 4(1 1 C)Cq(1 ` cq ` c2q).
e2(2 ` e 1 2cq ` 2c2q)
5 6
Since
f ** 4 Max 0,1 1 . (A1)
(4 ` 2e 1 e2)c 4 ` 4cq 1 4c2q ` 4(1 1 C)Cq(1 ` cq ` c2q) . 4,
It is easy to see that when
!4 ` 4cq 1 4c2q ` 4(1 1 C)Cq(1 ` cq ` c2q) . 2.
e (2 ` e 1 2cq ` 2c q)
2 2
Therefore, f *** , 1 only when e 4 0 or when e . 2 or e , 12. None
. 1, f ** 4 0.
(4 ` 2e 1 e2)c of these values falls within the feasible range of values of e [ (0,1].
Therefore, for all feasible values of e, f *** , 1. Similarly it can be
Restricting our attention to the positive values of f **,
shown that F*** , 1 for all feasible values of e.
e2(2 ` e 1 2cq ` 2c2q) We now show that f *** decreases as e increases:
f ** 4 1 1 , (A2)
(4 ` 2e 1 e2)c ]f *** n1 ` n2 ` n3 ` n4
4 , (A7)
]e c(4 ` 2e 1 e2)2 (2 1 e ` 2Cq 1 2C2q)2
]f** 1e(16 ` 16e ` 4e2 1 e3 ` 16cq ` 4ecq 1 16c2q 1 4ec2q)
4 where
]e c(4 ` 2e 1 e2)2
(A3) n1 4 1e(64 1 322 1 4e3 ` 8e4 1 e5),

Since e [ (0,1], and c [ (0,1], e2 . e3, 16cq . 16c2q, and 4ecq . n2 4 14e(1 1 C)C(32 ` 4e 1 8e2 1 4e3 ` e4
4ec2q. Therefore, the numerator of ]f **/]e is negative. It is easy to ` Cq(16 ` 4e 1 2C 1 4eC)).
see that the denominator is positive. Therefore, ]f **/]e , 0. n3 4 14e(16 1 e3)(1 1 c)cq,

Proof of Proposition 4 n4 4 14e(1 1 c)c(1 1 C)C(32 ` 4e 1 e3

When ` Cq(16 ` 4e 1 16C 1 4eC)).
It can be easily verified that the denominator of ]f ***/]e is positive
e2(2 ` e 1 2cq ` 2c2q) e2(2 ` e 1 2cq ` 2c2q)
, 1, f ** 4 1 1 . and that n1, n2, n3, and n4 each is negative. Therefore, ]f ***/]e , 0.
(4 ` 2e 1 e )c
2
(4 ` 2e 1 e2)c
The proof for ]F***/]e , 0 is similar.
After simplification, We now show that f *** increases as c increases:
]f *** e2(4 1 e2 ` 4c2q ` 4Cq 1 4C2q ` 4c2Cq2 1 4c2C2q2)
]f ** e2(2 ` e ` 2c2q) 4 .
4 2 . 0. (A4) ]c c2(4 ` 2e 1 e2)(2 1 e ` 2Cq 1 2C2q)
]c c (4 ` 2e 1 e2)2
(A8)
Proof of Proposition 5 Since e, c and C are in the (0,1] interval, the numerator and the de-
The demand functions in the asymmetric case are as follows. nominator are both positive, and ]f ***/]c . 0. The proof of ]F***/
]C . 0 is similar and is omitted.
Focal Firm:
Proof of Proposition 6
12
We need to show that F*** . f ***.
We are grateful to CNW/Marketing Research, Brandon, Oregon
for providing us the leasing data. We thank Brian Ratchford and F *** 1 f ***
three anonymous reviewers for helpful comments. We also thank e2(C 1 c)(g1 ` g2 ` g3 ` g4)
4 ,
Gary Erickson, Bob Jacobson, Martin Lariviere, Chak Narasimhan, cC(4 ` 2e 1 e )(2 1 e ` 2Cq 1 2C2q)(2 1 e ` 2cq 12c2q)
2

Rick Staelin, Nikos Vettas, and Russ Winer for comments on earlier (A9)
versions of the paper. The usual disclaimer applies. where

Marketing Science/Vol. 18, No. 1, 1999 57

 DESAI AND PUROHIT
Competition in Durable Goods Markets

g1 4 8 1 4e 1 2e2 ` e3, References

g2 4 4(2 1 e)(1 1 C)Cq, Brenner, D. 1992. Law and Regulation of Common Carriers in the Com-
munications Industry. Westview Press, Boulder, CO.
g3 4 4(2 1 e)(1 1 c)cq, Bresnahan, T. 1981. Competition and collusion in the American au-
Downloaded from informs.org by [128.122.253.228] on 02 June 2015, at 07:39 . For personal use only, all rights reserved.

g4 4 2cCq(4 1 e ` 4q 1 2eq ` 2ecq 1 4c2q ` 2eCq

2 tomobile industry. J. Econometrics 35 457–482.
` 4cCq 1 2ecCq 1 4C2q ` 4cCq2 1 4cC2q2 1 4c2C2q2). Bond, E., L. Samuelson. 1984. Durable good monopolies with ra-
tional expectations and replacement sales. Rand J. Econom. 15(3)
Clearly the denominator of F*** 1 f *** is positive. Therefore, F*** 336–345.
. f *** if the numerator is positive. Note that as c, C and e are all Bucovetsky, S., J. Chilton. 1986. Concurrent renting and selling in a
(weakly) less than 1, g1, g2 and g3 are positive. In g4, the sum of all durable goods monopoly under threat of entry. Rand J. Econom.
terms except 6c2C3q3(c 1 1) is positive. 17 261–278.
g1 . 2 . 2c2C3q3(1 1 c); and g2 . 4c(1 1 c)q . 4 c2C3q3(1 1 c). Bulow, J. 1982. Durable-goods monopolist. J. Political Econom. 90(2)
Therefore, (g1 ` g2) . 6c2C3q3(1 1 c). Therefore, (g1 ` g2 ` g3 ` 314–332.
g4) . 0 and F*** . f*** for any value of C . c. ——. 1986. An economic theory of planned obsolescence. Quart. J.
Proof of Corollary Econom. 51 729–750.
1n . p***
We need to show that P*** 1n . After simplification,
Business Week. 1994. Leasing fever. February 7, 92–97.
Coase, R. 1972. Durability and monopoly. J. Law and Econom. 15 143–
D 4 P***
1n 1 p***
1n
149.
a(C 1 c)(w1 ` w2 ` w3) Coughlan, A., B. Wernerfelt. 1989. On credible delegation by oligo-
4 ,
(4 ` 2e 1 e2)[4(1 ` cq 1 c2q)(1 ` Cq 1 C2q) 1 e2] polists: a discussion of distribution channel management. Man-
where agement Sci. 35 226–239.
Desai, P., D. Purohit. 1996. Leasing and selling: optimal strategies
w1 4 e4 1 4e(1 1 C 1 c) ` e3(1 1 C 1 c), (A10)
for a durable goods firm. forthcoming in Management Sci.
w2 4 8(1 ` cq 1 c q)(1 ` Cq ` C q),
2 2
(A11) ——, ——. 1998a. Leasing and selling; optimal marketing strategies
for a durable goods firm. Management Sci. 44 (November, Part
w3 4 2e (14 ` C 1 2Cq ` 2C q)
2 2
2) 19–34.
` 2c2q(1 ` Cq 1 C2q) ——, ——. 1998b. Appendix to competition in durable goods mar-
kets. Working Paper. Duke University, Durham, NC.
` c(1 1 2q 1 2Cq2 ` 2C2q2). (A12) Gul, F., H. Sonnenschein, R. Wilson. 1986. Foundations of dynamic
Since the denominator of D is positive, its sign is the same as the monopoly and the Coase conjecture. J. Econom. Theory 39 155–
sign of its numerator, which in turn is the same as the sign of w1 ` 190.
w2 ` w3. Lal, R. 1990. Improving channel coordination through franchising.
After further simplification, Marketing Sci. 9(4) 299–318.
McGuire, T., R. Staelin. 1983. An industry equilibrium analysis of
w1 ` w2 ` w3 4 l***
2u w4 ` (4 ` 2e 1 e )
2
downstream vertical integration. Marketing Sci. 2(2) 161.
• [2(1 1 c)(1 ` Cq 1 C2q) 1 e(2 1 2c 1 C)], (A13) Menezes, M., J. Serbin. 1993. Xerox Corporation: the customer sat-
isfaction program. Harvard Business School Case Series 591-
where
055. Harvard Business School Publishing Division, Cambridge,
(4 ` 2e 1 e2)(4(1 ` Cq 1 C2q)(1 ` cq 1 c2q) 1 e2) MA.
w4 4 . 0.
ac Purohit, D., R. Staelin. 1994. Rentals, sales and buybacks: managing
secondary distribution channels. J. Marketing Res. 31(3) 325–338.
The expression
——. 1997. Dual distribution channels: the competition between
2(1 1 c)(1 ` Cq 1 C2q) 1 e(2 1 2c 1 C) rental agencies and car dealers. Marketing Sci. 16(3) 228–245.
is decreasing in e, and equals Shooshan, H. 1984. Disconnecting Bell, Elmsford, Pergamon Press,
New York.
C(1 ` 2(1 1 c)(1 1 C)q) . 0 at e 4 1. Stokey, N. 1981. Rational expectations and durable goods pricing.
Therefore, Bell J. Econom. 12 112–128.
Tirole, J. 1988. The Theory of Industrial Organization. MIT Press, Cam-
2(1 1 c)(1 ` Cq 1 C2q) 1 e(2 1 2c 1 C) . 0,
bridge, MA.
and w1 ` w2 ` w3 . 0, which in turn implies that D 4 P***
1n 1 Wall Street Journal. 1995. Autos: new car leases come loaded with
1n . 0.
p*** options. April 12, B1.

This paper was received May 13, 1997, and has been with the authors 11 months for 2 revisions; processed by Brian T. Ratchford. To avoid a conflict of
interest, Ratchford handled this paper as acting editor while Richard Staelen was editor.

58 Marketing Science/Vol. 18, No. 1, 1999