This article was downloaded by: [128.122.253.

212] On: 10 May 2015, At: 22:10

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

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

Leasing and Selling: Optimal Marketing Strategies for a

Durable Goods Firm
Preyas Desai, Devavrat Purohit,

To cite this article:

Preyas Desai, Devavrat Purohit, (1998) Leasing and Selling: Optimal Marketing Strategies for a Durable Goods Firm.
Management Science 44(11-part-2):S19-S34. http://dx.doi.org/10.1287/mnsc.44.11.S19

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 [email protected].

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.

© 1998 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
 Leasing and Selling: Optimal Marketing
Strategies for a Durable Goods Firm
Preyas Desai j Devavrat Purohit
Fuqua School of Business, Duke University, Durham, North Carolina 27708
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

T his paper analyzes the problems associated with marketing a durable through leases and
sales. Academic research in this area has argued that in a monopolistic environment, leasing
dominates selling. Hence, leasing and selling should not co-exist and the firm should concentrate
its efforts solely on leasing. We show that the relative profitability of leasing and selling hinges
on the rates at which leased and sold units depreciate. In particular, we find that leasing does
not dominate selling in all cases; if sold units depreciate at a significantly higher rate than leased
units, a monopolistic firm is better off by only selling its product. In addition, we find that if
leaded and sold products depreciate at different rates, then the optimal strategy for the firm
involves a combination of both leasing and selling. We conclude the paper with an empirical
analysis of the depreciation rates of leased and sold units of a popular car model. We find that
the depreciation rate of leased cars has been significantly lower than the depreciation rate of
sold cars.
(Leasing; Selling; Durable Goods; Automobiles)

1. Introduction needs to maintain an optimal balance in the number of

products leased and sold. To highlight the importance
Most research in the marketing of durables has taken one of this issue, we note that the recent increase in leases
of two familiar tacks. First has been an emphasis on strat- in the automobile market has created an intriguing mar-
egies used by firms when their only option is to sell their keting problem. In particular, automobile manufactur-
products to consumers. Although this assumption of sales ers now have to concern themselves with marketing ex-
is reasonable for nondurable products, it is less appropri- leased cars that compete with their new cars. Thus, the
ate with durables that either can be leased or sold to con- issue they have to resolve is the following: Is it possible
sumers. Examples of product categories where firms em- that a large number of leases may improve a manufac-
ploy a mix of selling and leasing include photocopiers, turer’s current profitability but lower its long-term prof-
mainframe and mid-range computers, PCs, airplanes, etc. itability? Said differently, what are the long-term impli-
Second, given this observation of markets with both leases cations of leasing to some customers and selling to an-
and sales, the academic literature, beginning with Coase other set of customers? In this paper, we provide
(1972), has suggested that leasing can be more profitable insights into marketing durable goods by developing a
than selling. In this paper, we address both these issues. model that accounts for the strategic effects of simulta-
First, we explore the strategy of leasing and selling in a neously leasing and selling.
marketing context, and second, we offer an explanation A feature in common with leases and sales is that they
for why the coexistence of leasing and selling is an optimal both lead to future competition for the manufacturer. In
strategy for a durable goods firm. particular, a used or an ex-leased product can compete
From a managerial perspective, this paper shows how with the firm’s sales and leases of new products. How-
leasing creates strategic issues that are different from ever, a crucial difference between leases and sales is that
those associated with selling. This suggests that a firm leases create a secondary ex-leased market that is in the

0025-1909/98/4411/0S19$05.00
Copyright q 1998, Institute for Operations Research
and the Management Sciences MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S19

3b2f no07 Mp 19 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

firm’s control, while sales lead to a secondary used mar- petition for the manufacturer’s new cars.2 The auto-
ket that is outside the firm’s control. In the limit, because mobile industry clearly is concerned about this issue as
a leasing strategy allows the firm to eliminate the mar- noted by an executive from General Motors: ‘‘We don’t
ket for secondhand goods, it can even be more profita- want to create the kinds of problems with leasing re-
ble for the firm (Waldman 1997). Thus, a manufacturer turns created in the early 1990s when low-mileage for-
needs to evaluate the relative benefits of leasing and mer rental cars undercut sales of new vehicles’’ (Serafin
selling its product by making two inter-related deci- 1994). On the other hand, some automobile industry
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

sions: (1) Taking into account the future competition observers also highlight the flip side of the above prob-
created by its current marketing strategy, is there an lem: Because consumers may not maintain them ade-
optimal mix of leases and sales? (2) Which consumers quately, the returning lease vehicles may have high de-
should be targeted through leases and which ones preciation rates. Thus, if consumers return a large num-
through sales? That is, should the leases (or sales) be ber of poor quality ex-leased cars, the brand name may
targeted to the low-end or the high-end of the market? be affected adversely and result in a drop in used car
These issues represent ongoing concerns in any durable prices (Rechtin 1994).
product category where the product can be leased or While purchases of durables are commonplace, leas-
sold to consumers. ing has become popular only more recently. In order to
To further motivate the marketing issues associated understand the rationale behind this popularity and to
with leases and sales, consider the automobile indus- develop a model that captures the essence of the prob-
try which currently has to deal with a problem of its lem, it is important to be clear about the consequences
own creation. In particular, aggressive leasing in this of leasing from the perspectives of both the consumers
industry has reached the point where the number of and the firm. In particular, the academic literature has
off-lease vehicles will increase from 1.9 million in found leasing to be more profitable than selling in a
1993 to about 3.3 million in 1997 (Automotive News, monopolistic environment. The reason for this result is
1994).1 Competition from such large numbers of ex- that, unlike a monopolist who leases its product, a mo-
leased cars can have serious consequences for new car nopolist seller cannot commit itself to restrict quantities
prices and manufacturer profitability (e.g., Business and keep prices high. Thus, the seller has a problem
Week 1994, Thomas, 1994). For example, a jump in off- with time-consistency which then hurts its profitability.
lease vehicles could drive down the value of used Coase (1972) conjectured that this inability to commit
cars. This decrease in used car prices would have the would instantly lower prices to the competitive level.3
inadvertent effect of increasing the cost of leasing a This conjecture was subsequently formalized by Bulow
new car (Wall Street Journal 1994a). This occurs be- (1982) and Stokey (1981). Later research has pointed out
cause lease rates are set based on expected residual conditions under which Coase’s conjecture does not
values, and any decrease in the residual value in-
creases the capitalized cost of the car. 2
This problem is similar to the competition between new car sales
The nature of competition between ex-leased cars and
from dealers and ‘‘program car’’ sales from rental agencies analyzed
new cars clearly depends on the perceived quality dif- by Purohit and Staelin (1994). In the late 1980s and early 1990s, rental
ferences between these cars. This quality difference is agencies began selling their used (‘‘almost new’’) rental cars directly
determined by the rate at which the market perceives to consumers, greatly cutting into dealers’ sales of new cars. The car
the ex-leased cars to have depreciated. If the cars re- manufacturers dealt with the competition from almost new rental cars
by first requiring rental agencies to hold on to their cars for longer
turning from short-term leases have not depreciated
periods of time and later by instituting buyback programs.
much and are ‘‘almost new,’’ they represent strong com- 3
This time-inconsistency leads the monopolist to market a quantity
that is ‘‘too high.’’ An interesting analogy to time-inconsistency that
leads to a ‘‘too high’’ quantity comes from time-inconsistency in terms
1
Throughout this paper, we use the terms off-lease and ex-leased cars of R&D expenditures. In particular, a durable goods monopolist can
to refer to cars that were initially leased and then re-marketed by the practice planned obsolescence by investing in ‘‘too high’’ a level of
firm. R&D (Waldman 1996).

S20 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 20 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

hold, e.g., if there is a constant inflow of new customers prefers leases and another prefers sales; therefore, the
(Conlisk, Gerstner and Sobel 1984), if there are replace- firm should lease to the former and sell to the latter.
ment sales (Bond and Samuelson 1984), if the monop- However, because of the secondary markets created
olist has increasing marginal production costs (Kahn through such a strategy, it is not clear whether it is in
1986), or if demand is discrete (Bagnoli, Salant, and the firm’s interests to have concurrent leases and sales.
Swierzbinski 1989). The goals of this paper are to (1) understand the stra-
Essentially, the reason that leasing dominates selling tegic issues associated with concurrently leasing and
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

is that selling leads to an incompleteness of contracts. selling a product, and (2) determine the conditions un-
In particular, a lease ties-in consumers for a certain pe- der which this concurrent strategy is optimal. In this
riod during which prices are assumed not to change. sense, whatever the antecedents for the current increase
On the other hand, if market prices can change over the in leases in the auto industry, the question of whether
course of the lease, then leasing would also have a prob- a concurrent leasing and selling strategy is optimal still
lem with time-consistency (DeGraba 1994). In general, remains to be answered.4 This paper presents a first step
however, leasing does not tie-in consumers, thus mak- in that direction. In particular, we model a market
ing it more profitable than selling. An exception to this which potentially can allow both leases and sales. The
is if there is a threat of entry and the monopolist uses basic structure of our model parallels previous research
its sales as a credible entry deterrent (Bucovetsky and in the durable goods literature. That is, we assume the
Chilton 1986). However, does the coexistence of leasing firm markets a durable product in a two-period struc-
and selling occur only for the above reasons, or are there ture (e.g., Bulow 1982, Levinthal and Purohit 1989). As
other factors that may lead to a similar result? In tack- is the practice with most durables, we assume that if
ling this problem in a monopolistic environment where customers choose to use the product, they can do so by
sold and leased products depreciate at different rates, either buying or leasing the product. We note the fact
we find that (1) leasing dominates selling only under that any units of the product marketed in period 1 enter
certain conditions, and (2) a combination of leasing and the market in period 2 as either a used product or an
selling is the optimal strategy. ex-leased product. Either of these compete with any
Turning our attention to consumers’ incentives to new products that the firm tries to sell or lease in period
lease rather than buy, we find that the most frequently 2. To be consistent with consumer perceptions of how
mentioned benefit of leasing in the popular press is af- leased and sold cars depreciate and to model the poten-
fordability. In particular, because leasing involves fi- tial problems associated with the quality of leased cars,
nancing only the capitalized cost of the car, a con- we allow leased and purchased goods to depreciate at
sumer’s monthly payments are lower with leasing than different rates.
with buying. Therefore, if some consumers have severe In terms of the consumer side of the market, we
monthly budget constraints, they can benefit from leas- model three commonly observed consumer usage pat-
ing cars. Upon reflection, however, this argument is not terns: a group of consumers who buy new cars and fre-
convincing, because, as with any financial contract, the quently replace them; another group of consumers who
‘‘affordability’’ benefit of leasing can easily be replicated buy new cars and hold on to them as long as they last;
by selling. For example, a seller can have a series of and a third group of consumers who buy only not-new
‘‘low’’ monthly payments followed by a single large (i.e., used or ex-leased) cars. To endogenously develop
payment. In fact, Mitsubishi goes a step further—it of- these consumer strategies, we consider customers who
fers a balloon loan with low initial payments such that differ in their valuations of the product (as in Mann
a consumer has the option of selling the car to the firm 1992, Moorthy 1984). Based on this simple structure, we
at a predetermined price and time (Wall Street Journal find that the firm’s strategy to either lease or sell to any
1995a). Importantly, however, affordability does not ex-
plain whether concurrent leasing and selling is optimal
from the manufacturer’s long-term perspective. At best, 4
As an aside, the auto executives are clearly unsure about the long-
the argument of affordability asserts that one segment term consequences of using aggressive leasing to market their cars.

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S21

3b2f no07 Mp 21 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

group of consumers depends upon the relative depre- rioration interchangeably to reflect the physical wear
ciation rates of sold and leased cars. If leased cars are and tear of the product. We capture this difference in
likely to depreciate more than sold cars, the firm should depreciation through the parameters db and dl , where 0
direct leases to the high willingness-to-pay consumers, ° db ° 1 and 0 ° dl ° 1 (we elaborate on this issue
who tend to replace their cars each period. On the other subsequently).5 Finally, note that in period 1 only new
hand, if leased cars are expected to depreciate less, the cars are available, while in period 2, new cars as well as
firm should direct its leases at lower willingness-to-pay used and ex-leased cars are available.
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

consumers who are more inclined to purchase their ex-

leased cars at the end of the lease period. 2.2. Consumers
The rest of the paper is organized as follows. Section For most products, we observe that consumers differ in
2 describes our model. In §3, we analyze the benchmark their willingness to pay. To capture this fact, we model
cases of pure leasing and pure selling to see how they consumers who are heterogeneous in their valuations
are affected by depreciation. Subsequently, in §4, we of the car. Given that the product is a durable, note that
analyze the implications of concurrently selling and consumers value the car for the services it provides for
leasing to different consumers. In §5, we conduct an multiple periods. Thus, it is easier to think about a con-
empirical analysis to check on the robustness of our as- sumer’s valuation of the services provided in each pe-
sumption on depreciation. Finally, in §6, we conclude riod. More formally, we use the parameter u √ [0, 1] to
the paper and present some ideas for future research. represent a consumer’s valuation of the services pro-
vided by a car in any period, where a consumer with a
higher u values the car more than a consumer with a
2. Model lower u. We assume that u is distributed uniformly be-
In order to formulate the model, we first detail our as- tween 0 and 1, and in any period each customer uses at
sumptions about the product and the players, i.e., con- most one car. This form of consumer heterogeneity has
sumers and the manufacturer. Note that although we been widely employed in marketing and economics
motivate this discussion by assuming that the product (e.g., Mussa and Rosen 1978, Moorthy 1984, Moorthy
being marketed is a car, the model applies to any du- and P’ng 1992). Note that other forms of differences
rable product category. To capture the durable nature among consumers, e.g., income or price sensitivity, can
of the product and to highlight the inter-temporal in- be mapped into the differences in valuations.
centives of consumers and the firm, we use a two-period Recall from our discussion earlier that the product
model in our analysis. depreciates as it ages. This, in turn, suggests that in any
2.1. Product period, a new car is more valuable than a not-new (i.e.,
The product that we consider is a car and to model its ex-leased and used) car. Thus, we assume that a con-
longevity we assume that it lasts for two periods. As- sumer’s gross per-period utility, G, from using either a
suming it lasts for two periods is not critical; it is only new or a not-new car is given by
important to assume that it lasts for a finite amount of G( u, n) Å u(1 0 n dH i ), (1)
time. In addition, we assume that there can potentially
be three types of cars available in the market: new cars, where the parameters n and dH i (i Å b, l) together repre-
ex-leased cars, and used cars. Ex-leased cars were pre- sent the deterioration of the car with age. More specif-
viously leased by consumers and are now offered for ically, n is an indicator variable such that n Å 0 if the
sale by the manufacturer. Used cars are owned and thus product is new and n Å 1 if the product is old, and dH i is
offered for sale by consumers. Thus, the key difference a random variable that represents depreciation of the
between ex-leased and used cars is the form of prior product with age. dH i has a mean of db when the car is
‘‘ownership,’’ i.e., whether the car was leased or bought. bought and a mean of dl when the car is leased. Thus,
This difference in the form of ownership can lead to
leased and sold cars depreciating at different rates. 5
Throughout the paper, subscripts b and l denote buying and leasing
Throughout this paper, we use depreciation and dete- respectively.

S22 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 22 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

the expected gross per-period utility from a new car is ticular, there is considerable disagreement in the auto
u and from a not-new car is u(1 0 di ), i Å (b, l). Our industry about the relative rates of depreciation of
assumed form of per-period utility implies that any in- leased and sold vehicles. Some industry observers be-
crease in depreciation lowers utility. In addition, note lieve that leased cars depreciate at a higher rate than
that this decrease in utility is higher for consumers with sold cars, while others believe that sold cars depreciate
higher valuations, dG/dud d õ 0. Intuitively, this is rea- at a higher rate than leased cars (Wall Street Journal
sonable, because any specific deterioration of the prod- 1994b, 1995b). Because large scale leasing is a relatively
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

uct should result in a greater loss in utility for those recent phenomenon in this industry, there is no publicly
consumers who have higher valuations for quality, i.e., available information on the relationship between these
consumers with higher u. depreciation rates. Therefore, in order to capture either
Equation (1) assumes a particular form of vintage-use situation, we model two deterioration rates, dl and db ,
depreciation that we believe is appropriate for modeling for leased and bought cars, respectively. We acknowl-
durables. This form of depreciation assumes there are edge that the depreciation process could be endoge-
effects associated with both vintage and use. In partic- nous, resulting in equilibrium rates of db and dl . Our
ular, as a product becomes dated, it also loses some of model does not determine the equilibrium rates; how-
its appeal to consumers and this effect persists indepen- ever, it determines the firm’s optimal strategy given any
dent of usage or whether the car was sold or leased. For depreciation rates.6 As a check on the robustness of our
example, a 1995 model year car is valued less than an assumption of different depreciation rates, we estimate
equivalent 1996 model year car. An alternative method the depreciation rates for leased and sold units of a pop-
of modeling depreciation is to assume that some pro- ular car model. We find that compared to sold units,
portion (0 ° g ° 1) of the products depreciate fully and leased units depreciate at a significantly lower rate (ap-
the remaining (1 0 g ) units are of the same quality as proximately 2.4% per year).
new ones. This form of depreciation, often referred to The most popular argument for a higher depreciation
as a one-hoss shay type, does not allow for any differ- for leases relies on adverse selection and moral hazard.
entiation between new and used units. Given our inter- That is, leasing may attract a disproportionately large
est in product differentiation and our focus on markets number of ‘‘poor’’ drivers who may also have less in-
with vintage effects, it is more appropriate to assume a centives to adequately maintain the car. If this indeed
vintage-use depreciation. is the case, then one may expect lessees’ poor driving
Our assumptions about depreciation imply that, habits or their lack of adequate maintenance to result in
holding all else fixed, consumers prefer a new car to a a higher deterioration of leased vehicles. On the other
not-new car. This preference arises because, regardless hand, some observers make an argument for leased cars
of the degree of maintenance, any car depreciates as it depreciating at a lower rate than sold cars. In particular,
ages and provides less value to consumers. For exam- leasing contracts typically restrict the usage of leased
ple, older cars suffer from a lack of newness or a ‘‘new cars to an annual mileage limit. If usage exceeds the
car smell,’’ they tend to have more squeaks and rattles, specified annual limit, then lessees are penalized by a
and even a higher probability of breaking down. At the
time of purchase or lease of the new car, consumers do 6
Note that even with endogenous depreciation, the equilibrium rates
not know how much the car will deteriorate over time,
could go either way. For example, if consumers are plagued by moral
but they know the mean and distribution of the random hazard, then in equilibrium dl ú db . However, if firms are effective in
deterioration variable. The rate of deterioration could imposing restrictions on lessees, we also have to consider the possi-
also depend on the maintenance the consumer chooses bility of dl õ db . One way of modeling leasing restrictions and charges
to provide (e.g., Mann 1992; Padmanabhan and Rao for excess wear and tear is by creating a penalty function that becomes
effective if the leased car depreciates more than a certain amount (say)
1993).
dU . However, if the maintenance effort is costly for consumers, we
As alluded to earlier, an important aspect of our would expect all consumers to ensure a depreciation of dU . Our exog-
model is that we allow for the possibility that a leased enously specified dl could be viewed as an approximation of this
car may depreciate differently from a sold car. In par- process.

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S23

3b2f no07 Mp 23 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

per-mile charge. In addition, leasing contracts also pro- Figure 1 Consumer Strategies
vide for an ‘‘excessive wear and tear’’ charge that the
lessor can levy at the end of the leasing period. Thus, it
is possible that restrictions on usage or lessees’ fear of
extra fees may in fact result in leased cars deteriorating
less than bought cars. Also note that car manufacturers
(e.g., Cadillac, Jaguar, etc.) are attempting to influence
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

the perception of leased vehicles’ depreciation by ‘‘cer-

tifying’’ the quality of the car. Finally, we note one ad-
ditional factor that suggests that leased cars may depre-
ciate at a lower rate than sold cars. That is, an ‘‘almost
new’’ (e.g., two-year old) used car is more likely to be
sold because it is a ‘‘lemon’’ than is an ex-leased car of
the same age which is available merely because the lease 2. Middle: Get a new car in period 1 and then hold
expired. onto this car in period 2. That is, consumers either buy
As mentioned earlier, a reasonable assumption in our a new car and then hold onto it (BH) or lease a new car
model is that, ceteris paribus, all consumers prefer a and then purchase the ex-leased car (LX). These strat-
new car to a not-new car. Ultimately, however, what egies typify traditional consumers who tend to hold on
leads a customer to decide between a new and a not- to their cars for longer periods of time.
new car is the price. More specifically, we assume that 3. Bottom: Do not use any car in period 1 and get a
a consumer’s net per-period utility, U, is given by not-new car in period 2, i.e., remain inactive in period
1 and then purchase either a used (IU) or an ex-leased
U Å u(1 0 n di ) 0 p, (2) car (IX). These strategies are followed by first-time buy-
ers such as college students who enter the market from
where the first part of Equation (2) is the gross utility the bottom of the car spectrum.
defined in Equation (1) and p is the price paid by the 4. Inactive: Do not use a car in either period and stay
consumer. Depending on the strategy chosen by the out of the market, i.e., remain inactive in both periods (II).
firm, we show later that p can be either a lease or a sales Note that the grouping of consumers outlined above
price. corresponds directly to their overall valuations of the
Consumers in our model are aware that although the car. That is, consumers with higher valuations for the
product is homogeneous in period 1, it is differentiated services provided by a car will tend to be in the top
in period 2. That is, in period 1, only new cars are avail- group, followed by consumers with lower valuations in
able in the market, while in period 2, both new and not- the middle group, and so on. Note that the ‘‘true’’ bot-
new cars are available. Thus, at the beginning of period tom consumers are actually the ones who remain inac-
1, when consumers evaluate their needs over the two- tive in both periods. Here we use ‘‘bottom’’ to refer to
period horizon, we assume they follow one of the strat- the lowest valuation consumers who actually use the
egies laid out in Figure 1. These strategies broadly clas- product. Finally, because of our assumption that each
sify consumers into one of the four following groups: consumer uses at most one car in any period, the num-
1. Top: Get a new car in period 1 and another new ber of consumers in any group is given by the length of
car in period 2. Thus, consumers either lease a new car that group. It is important to recognize that the number
in each period (LL) or buy a new car in each period of consumers in any group will be determined by the
(BB). Consumers who buy a new car in period 2 will prices charged by the manufacturer. In other words,
also sell their used car in the open market. Consumers based on the firm’s marketing decisions, the sizes of
who follow BB or LL strategies can be thought of as new these groups emerge endogenously.
car lovers in the sense that they prefer to have a new As is the case with most durables, consumers’ expec-
car in each period. tations of future prices play an important role in their

S24 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 24 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

purchase decisions. In particular, in acquiring a durable, facturer leases cars in periods 1 and 2. As outlined ear-
consumers form expectations about two related future lier, cars that are leased in period 1 are perceived by the
prices. First is the price of future new products—the market to depreciate at a rate 0 ° dl ° 1. Thus, if the
more that consumers expect future prices to fall, the less manufacturer leases q1l new cars in period 1, then in
they are willing to pay for their current purchase. The period 2, it can lease the q1l ex-leased cars carried over
second expectation has to do with the future price of the from the prior period and, if it chooses, it can lease q2l
current product. That is, consumers are more likely to new cars. In terms of consumer behavior from Figure 1,
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

pay a higher price for products that tend to retain more we only have LL, LX, IX, and II strategies. This implies
of their value over time. Within our model, we capture the following inverse demand system:
these characteristics by assuming that consumers are
l2x Å (1 0 dl )(1 0 q1l 0 q2l ),
able to forecast future prices perfectly. Although this
assumption may at first appear strong, note that all we l2n Å l2x / dl (1 0 q2l ),
need to assume is for consumers to assume that prices
l1n Å 1 0 q1l , (3)
decrease over time. Rather than assuming an arbitrary
price reduction process, we assume that consumers where l2x is the price of an ex-leased (x) car in period 2,
have perfect expectations. l2n is the price of a new car (n) in period 2, and l1n is the
price of leasing a new car in period 1. Note that period
2.3. Manufacturer 2 demand for cars is such that the price of a new car is
We assume the firm can manufacture and market the higher than the price of an ex-leased car, l2n ¢ l2x . This
car at a constant marginal cost of c. For analytical sim- is reasonable because older cars have depreciated and
plicity and because the presence of positive costs does thus provide a lower level of service. If the cars do not
not affect the nature of our results, we set c Å 0.7 We depreciate (i.e., dl Å 0) there is no difference between
assume that the manufacturer does not buy back or new and ex-leased cars, and l2n Å l2x .
destroy any of the units it markets in period 1. Note Based on these prices, the manufacturer maximizes
that a sufficiently high marginal cost would ensure profits over both periods by choosing optimal quantities
that this assumption holds. We first analyze the of q1l and q2l . Because leasing does not entail a problem
benchmark cases where the manufacturer exclusively with time inconsistency, the manufacturer chooses both
uses either selling or leasing. Subsequently we enrich quantities simultaneously. Thus, it maximizes
the model to capture concurrent leasing and selling
in our model. The central question that this allows us l1nq1l / r [l2nq2l / l2xq1l ], (4)
to address is whether a combination of leasing and where 0 ° r ° 1 is the discount factor (common to all
selling can be an optimal strategy. players). This maximization problem leads to the fol-
lowing sets of optimal quantities.
3. The Benchmark Strategies 1
In this section, we consider the pure strategies of leasing q*1l Å (5)
2(1 / dl r 0 d l2 r )
and selling and analyze how they are affected by de-
preciation. That is, we first look at the case where the 1
q *2l Å *.
0 (1 0 dl )q 1l (6)
firm only leases to consumers and then consider the case 2
where it only sells to consumers. If the car does not depreciate, i.e., dl Å 0, then the man-
3.1. Pure Leasing ufacturer sets q2l Å 0 and leases q1l in both periods. Fi-
In this section, we assume that the only way for the firm nally, note that a myopic monopolist would choose to
to market its product is through leases and the manu- lease a larger quantity of the product in period 1.

3.2. Pure Selling

7
Results with c ú 0 are shown in an appendix available from the In this section, we consider the case where the manu-
authors. facturer can only sell its product to consumers. Thus,

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S25

3b2f no07 Mp 25 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

the manufacturer sells q1b new cars in period 1, and an- Given the optimal q *2b (q1b ), the manufacturer opti-
other q2b new cars in period 2. In addition, cars sold in mizes profits over both periods by choosing the optimal
period 1 depreciate at a rate 0 ° db ° 1 and are available quantity to sell in period 1. That is, it maximizes
in the secondhand market as used cars. In terms of con-
sumer behavior from Figure 1, this implies that consum- p1nq1b / r [p2nq *2b (·)], (9)
ers use only BB, BH, IU, and II strategies. yielding
The set of equations that characterize this model are
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

given by: 2
q *1b Å . (10)
4 / r / 2dbr 0 3d b2 r
p2u Å (1 0 db )(1 0 q1b 0 q2b ),
3.3. Comparing Pure Leasing and Selling
p2n Å p2u / db (1 0 q2b ), The general result of previous research in this area has
p1n Å (1 0 q1b ) / rp2u , (7) been that a monopolist firm would prefer to lease rather
than sell its product. This stems from the firm’s inability
where p2u is the price of a used car in period 2, p2n is the to commit to a sales strategy. Yet it is not entirely clear
price of a new car in period 2, and p1n is the cost of if this were to hold true if the leased and sold goods
buying a car in period 1. depreciated at different rates. In particular, we have the
Note that because new and used cars compete against following proposition: 8
each other, any increase in sales of new cars (q2b ) lowers
PROPOSITION 1. A pure selling strategy has higher prof-
the price of a used car. In addition, as in the case of pure
its than a pure leasing strategy if and only if
leasing, note that the price of a new car is higher than
the price of a used car. This price difference is affected q_ ____________
entirely by the rate at which the used car has depreci- 1 / 2 1 0 3dl (1 0 dl )
db ú . (11)
ated. If the depreciation rate is high, then the used car 3
is less of a substitute for the new car and the price dif-
From Equation (11), note that if sold cars depreciate
ference between the two is also high. On the other hand,
at a sufficiently higher rate than leased cars, then pure
if there is no depreciation then used and new cars are
selling is more profitable than pure leasing. While the
identical and hence their prices are also identical. Fi-
extant literature finds that theoretically leasing is more
nally, note how the model incorporates consumers ex-
profitable than selling, it also concedes that they both
pectations of future prices. In Equation (7), the first part
co-exist. In fact, Tirole (1988) suggests that leasing may
of the equation for p1n reflects the cost of owning the car
become less attractive for a firm because of moral hazard
for one period; this cost is similar to the lease price, l1n
and adverse selection. That is, lessees are harder on their
in Equation (3). However, a consumer who purchases
cars and leasing attracts poor drivers—in terms of our
a durable in period 1 is concerned not only with the
model, this implies that dl should be much higher than
current value but also with the future value of the used
db . However, Proposition 1 shows exactly the opposite
durable. Hence the selling price in period 1 reflects the
result. That is, leasing is less attractive than selling only
current value and the expected price in the future.
when db is significantly higher than dl .
To solve this model, we first solve the manufacturer’s
For the intuition behind Proposition 1, consider Fig-
problem in period 2 and then solve the period 1 prob-
ure 2 that displays the manufacturers profits under both
lem. This backward induction is required because con-
the pure leasing and selling cases. In the case of pure
sumers expectations are rational and we are solving for
leasing, note that whether the car does not depreciate at
a subgame perfect Nash equilibrium. The manufacturer
all (i.e., dl Å 0) or the car depreciates completely (i.e., dl
maximizes profits, p2nq2b , in period 2 by choosing the
Å 1), the profits are identical and the maximum that can
optimal number of cars to sell, q *2b . This yields:

1 0 (1 0 db )q1b 8
Proofs of all propositions are in an appendix available from the au-
q *2b Å . (8)
2 thors.

S26 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 26 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

Figure 2 Optimal Profits Under Pure Leasing and Selling be top, middle, and bottom groups of consumers in the
market, but there will also be a larger number of strat-
egies. That is, one might expect to see a combination of
the strategies under the pure cases—LL, BB, LX, BH, IU
and IX—as outlined Figure 1. In addition, one may note
that we could potentially have a combination of leasing
and buying strategies—BX and LU. However, BX and
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

LU strategies are absent because they are dominated by

LX and BH, respectively.
In comparing the difference in utilities of a given con-
sumer following LL and BB strategies, we note that this
difference depends only on the prices. That is,

U LL 0 U BB Å [ u 0 l1n / r( u 0 l2n )]

0 [ u 0 p1n / r( u / p2u 0 p2n )]

be achieved. That is, the extreme values of dl allow the
Å p1n 0 l1n / r(p2n 0 p2u 0 l2n ). (12)
firm to market a restricted quantity of leases to keep
prices and profits high. Under a pure selling strategy, Therefore, depending on equilibrium prices, all con-
the firm maximizes profits when db Å 1 or the car de- sumers will find either BB or LL to be more profitable
preciates fully (i.e., time-inconsistency is not an issue). than the other. As a result, BB and LL do not co-exist in
With this extreme depreciation, there is no secondhand equilibrium. In the case of the middle group, both LX
market created and, hence, no competition for the firm. and BH may be observed in equilibrium. The bottom
In addition, when the car depreciates fully, then it is no group of consumers follow IU or IX depending on the
longer a durable and the distinction between leasing type of old cars available in the market.
and selling is moot. For selling to dominate leasing, a Given this range of consumer strategies, we consider
necessary condition is that the sold car depreciate at a two marketing strategies for the firm: (i) Premium lease
sufficiently high rate of db ¢ 23 . This implies that only in which the firm’s quantity decisions make an LL strat-
when the sold product is almost a non-durable does egy dominate a BB strategy for all top consumers; and
selling become strictly more profitable than leasing. (ii) Value lease in which the firm’s quantity decisions
make a BB strategy dominate an LL strategy for all top
4. Concurrent Leasing and Selling consumers. This allows us to answer the question of
In prior sections we analyzed the cases where the firm whether leases should be targeted to the high or the low
exclusively sold or leased its product. In both these willingness-to-pay consumers.
cases, the basic behavior of the top, middle, and bottom In summary, under premium and value leasing, the
groups of consumers was given by the following: consumer strategies that emerge are defined as follows:

Top Middle Bottom Top Middle Bottom

Pure Leasing LL LX IX Premium Lease LL LX BH IX

Pure Selling BB BH IU Value Lease BB LX BH IU

As in the pure selling and pure leasing cases, the top

where the strategies are laid out in Figure 1. group uses a new car in each period, the middle group
A strategy with concurrent leases and sales suggests uses a new car in period 1 and a not-new car in period
that we will see a combination of the consumer strate- 2, and the bottom group is inactive in period 1 and uses
gies under pure leasing and selling. Thus, there will still a not-new car in period 2. The key difference in the

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S27

3b2f no07 Mp 27 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

concurrent case is that by allowing leases and sales, we Finally, in all cases, we assume the firm determines
now have two subgroups in the middle. These two the quantities of cars to lease and sell in each period.
groups are different only in the method by which they These quantities determine consumers’ utilities and, as
obtained a car in period 1 (i.e., leased or bought), be- a consequence, their willingness to pay. Thus, rather
cause they both hold on to it in period 2. Finally, be- than assuming reduced form demand functions, our ap-
cause the middle group holds onto its cars in period 2, proach is more general and determines the (inverse) de-
the bottom group of consumers can only obtain cars that mand functions endogenously. These endogenous de-
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

are turned over from the top group in period 2. mand functions have some appealing properties that we
4.1. Effect of Depreciation highlight briefly. First, the price of any product de-
A characteristic of our model is that for any given car, creases as the quantity of substitute products increases,
the top group has a higher valuation than the middle e.g., an increase in the number of used cars decreases
group, and the middle group has a higher valuation the price of new cars. And second, the higher the de-
than the bottom group. These valuations become im- preciation rate, the lower the price of the used (or ex-
portant when it comes to determining the demand func- leased car). In addition, any increase in depreciation of
tions. If each group has only one strategy associated a used or an ex-leased car increases the price of a new
with it, then it is easy to develop the demand functions. car. This occurs because an increase in depreciation
As shown earlier, such is the case in the pure leasing makes the used and ex-leased cars weaker substitutes
and selling cases. However, as we show below, the pro- for the new car. In solving the model, we ensure that
cedure under the concurrent leasing and selling strate- the equilibrium is sub-game perfect and the firm is se-
gies is slightly more complicated. In particular, the mid- quentially rational. Thus, we first solve the period 2
dle group is comprised of two subgroups, each with its problem and then solve the period 1 problem.
own strategy. As shown earlier, consumers can either
4.2. Premium Lease
lease a new car in period 1 and hold onto it in period 2
Under a premium lease strategy, in period 1, the firm
(an LX strategy), or they can buy a new car in period 1
leases q1l new cars and sells q1b new cars. In period 2,
and hold onto it in period 2 (a BH strategy). Thus, in
the manufacturer markets q2l ° q1l new cars. This im-
modeling concurrent leases and sales, we need to de-
plies that q2l ex-leased cars enter the market and are pur-
termine the ordering of these two subgroups, i.e., which
chased by the bottom group of the market. In addition,
of these two subgroups has a higher valuation for the
the remaining ex-leased cars (q1l 0 q2l ) are purchased by
product.
consumers who previously leased their car and now de-
The ordering of these two subgroups depends en-
cide to exercise their option. Thus, under a premium
tirely upon the rates of depreciation of leases and sales,
lease strategy, we need to determine demand functions
dl and db . In particular, if db ú dl , then ex-leased cars
for new (p2n ) and ex-leased (p2x ) cars in period 2, and
have a higher market value than used cars. This, in turn,
period 1 new car sales (p1b ) and leases (p1l ).
implies that LX consumers will have a higher valuation
For both cases ( db ú dl and dl ú db ), the (inverse) de-
for the car than will BH consumers. Similarly, if db õ dl ,
mand functions under this scenario are provided in an
then used cars will have a higher market value than ex-
appendix. As before, we solve this model by first solv-
leased cars, implying that BH consumers will have a
ing the period 2 problem and then the period 1 problem.
higher valuation for the car than will LX consumers. In
That is, the firm maximizes profits in period 2, p2nq2l
addition, to ensure that equilibrium prices entail no ar-
/ p2xq1l , by choosing the optimal amount q *2l . Given this
bitrage opportunities for consumers, it is important to
determine the ordering of the middle group. Thus, de- optimal choice of q *2l , the firm maximizes its total profits
pending on the ordering of the subgroups, the model in period 1:
generates demand functions that depend on the relative P Å p1nq1b / l1nq1l / r(p2nq *2l (·) / p2xq1l ) (13)
rates of depreciation of leases and sales. Thus, with both
premium and value leases, we analyze two cases. In the by choosing the optimal quantities to lease (q *1l ) and sell
first, db ú dl , and in the second, db õ dl . * ) in period 1.
(q 1b

S28 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 28 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

4.3. Value Lease PROPOSITION 2. If the depreciation rates of leased cars

Under a value lease strategy, in period 1, the firm sells and sold cars are equal, then the profits of the value lease,
q1b new cars and leases q1l new cars. In period 2, the firm premium lease, and pure leasing strategies are also equal. In
markets q2b new cars and q1l ex-leased cars carried over addition, both the concurrent strategies collapse to the pure
from period 1. In addition, the q1b used cars (turned over leasing strategy.
from the top group) will be available in the market.
Now consider the case where db ú dl or sold cars de-
Thus, under a value strategy, we need to determine de-
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

preciate at a higher rate than leased cars. As discussed

mand functions for new (p2n ), ex-leased (p2x ), and used
earlier, this might occur if the firm places restrictions on
cars (p2u ) in period 2, and in period 1, new car sales (p1b )
mileage, maintenance requirements for lessees to fol-
and leases (l1n ).
low, or if the firm ‘‘fixes’’ up ex-leased cars before it
For both cases ( db ú dl ) and ( dl ú db ), as before, we
markets them again to consumers, etc. This suggests
first solve the firm’s period 2 problem. That is maximize
that any consumers who buy a car in period 1 will own
period 2 profits, p2nq2b / p2xq1l , by choosing an optimal
‘‘lower’’ quality products at the beginning of period 2.
* . Given this optimal q 2b
q 2b * , we now solve the problem
This future decrease in quality reduces their willingness
in period 1. That is, maximize profits
to pay for the car in period 1. This might suggest to the
reader that the strategy the firm should follow is to tar-
P Å p1bq1b / l1nq1l / r [p2nq *2b / p2xq1l ] (14)
get the ‘‘higher’’ quality car to the higher willingness to
by choosing an optimal level of leases and sales. pay group. In other words, the lease should be targeted
to the top group and the sales to the middle group.
4.4. Comparing Value and Premium Leases However, this initial intuition is incorrect and our anal-
In this section, we explore the effects of premium lease ysis suggests the following proposition:
and value lease strategies on the firm’s profits. In par-
ticular, the differences between these strategies are af- PROPOSITION 3. When db ú dl , a value lease strategy is
fected by the relative depreciation rates of leased and observed in the sub-game perfect equilibrium.
sold cars, respectively. As noted earlier, these depreci- Note that the above proposition implies that when db
ation rates have an interesting effect on the future qual- ú dl : (i) the firm’s profits are higher with a value lease
ity of the car. That is, new cars are of the same quality strategy and (ii) all consumers adopt one of BB, BH, LX
in period 1, regardless of whether they are leased or and IU strategies. The second part in particular requires
bought by the consumer. However, a difference in de- that no consumer adopt an LL strategy in the above
preciation rates leads to a difference in ‘‘quality’’ only equilibrium. Indeed, given the optimal quantities and
in period 2. prices under value lease, a BB strategy dominates an LL
When leased and sold cars depreciate identically, strategy. In addition, we also show that no consumer
then db Å dl and consumers do not differentiate between will use either BX or LU strategies.
an ex-leased car and a used car and there is nothing to The higher profitability of a value lease strategy for
suggest that one is of higher quality than the other. the firm means that when db ú dl , it is more profitable
Thus, from the consumer’s perspective, leased and sold for the firm to target the middle group with a lease and
cars are identical and market prices should reflect that the top group with a sales strategy. This implies that the
fact. That is, if consumers buy a new car in period 1, the firm should target the product of ‘‘lower’’ quality to the
residual value of their used car is p2u , and if they lease higher willingness to pay (top) group through a sales
a new car, their option price is p2x . Given that the resid- strategy. This occurs mainly because these consumers
ual values are equal under both value lease and pre- care less about the future quality of the product. On the
mium lease strategies, period 1 lessees and buyers face other hand, the middle consumers are more concerned
identical choices in period 2. As a result, they face iden- about the future quality, and the firm should target the
tical choices in period 1. In terms of profits, this leads ‘‘higher’’ quality product to this group with a lease
to the following proposition: strategy.

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S29

3b2f no07 Mp 29 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

Finally, consider the case where dl ú db and leased exception. In particular, as long as depreciation rates are
cars depreciate at a higher rate. In this case, an ex-leased different, we find that it is optimal to market through
car has ‘‘lower’’ quality than a used car in period 2, concurrent leases and sales. This result holds despite the
which implies that the price of an ex-leased car will be fact that the products marketed through leases and sales
lower than that of a used car. Thus, a consumer who have the same marginal cost. The earlier research would
buys a car gets a higher ‘‘quality’’ car at the end of pe- suggest that the firm would be better off marketing only
riod 1 than a consumer who leases a car. the highest quality (i.e., lowest depreciation) product—
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

we find that the firm is better off marketing both quality

PROPOSITION 4. When db õ dl , a premium lease strategy
levels.
is observed in the sub-game perfect equilibrium.
Finally, we note that while the firm generally does
Proposition 4 asserts that when db õ dl , the firm op- not have control over the rate at which sold cars depre-
timizes profits with a premium lease strategy. For this ciate, it can influence the rate of depreciation of leased
strategy to work, quantities and prices have to be such cars. As discussed earlier, this might come about
that no consumer adopts BB, BX, or LU strategies. Prop- through restrictions on the number of miles allowed on
osition 4 suggests that the firm should target the lower the car or requiring maintenance at regularly scheduled
quality car (i.e., a leased car) to the higher willingness intervals. Note however that there are limits to which
to pay or the top group. That is, the top group should the firm can influence this rate and that compliance on
be targeted with a lease strategy and the middle group the part of the lessees should be observable. Recently,
with a sales strategy. In summary, the insight from some manufacturers (e.g., Audi) have began offering
Propositions 3 and 4 is that the product that depreciates leases where the manufacturer pays for all regularly
more should be marketed to the highest willingness to scheduled maintenance over the lease period. Not only
pay consumers. does this reduce the lessee’s perceived cost of maintain-
Now we are in a position to address the question ing the car, we believe that it strongly influences the
posed at the beginning of this paper: Given that a firm value of the car when it is returned to the firm. Thus,
marketing a durable product creates its own future the question from the firm’s perspective is if it can in-
competition, can this firm be better off by following a fluence the rate of depreciation of its leases, how would
strategy of concurrent leasing and selling? We answer it choose to do so? In general, we find that if the firm is
as follows: operating under a premium lease strategy, then it
should try and increase dl to the highest level possible
PROPOSITION 5. If leased and sold cars depreciate at dif-
(e.g., minimizing the number of restrictions on lessees’
ferent rates, db x dl , and db , dl x 1 then a combination of
contracts). On the other hand, if the firm is under a
leasing and selling is more profitable than pure leasing or
value lease strategy, it should try and decrease dl to the
pure selling.
lowest level possible (e.g., increasing the level of restric-
In other words, a combination of leasing and selling tions).
with db x dl allows the firm to discriminate across con-
sumers. However, note that if the product depreciates
fully (i.e., d Å 1), then it is no longer a durable and the 5. Robustness Check
distinction between selling and leasing is moot. A simplifying assumption of our model is that the de-
Earlier research has shown that a monopolist can im- preciation rates of leased and sold units are determined
prove its profitability by marketing products at varying exogenously. Endogenizing depreciation in the current
quality levels (e.g., Mussa and Rosen 1978, Moorthy model, although a laudatory task, makes the framework
1984). However, this result holds specifically when mar- intractable in studying a strategy that allows concurrent
ginal cost increases sufficiently with the quality of the leasing and selling. However, as pointed out earlier,
product. In fact, if cost were independent of quality, even though our model does not determine the depre-
then the monopolist would be better off marketing only ciation rates endogenously, it does determine the firm’s
the highest quality level. Our result offers a noteworthy optimal strategy given any depreciation rates. In partic-

S30 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 30 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

ular, we find that as long as the depreciation rates of addition, note that by analyzing a specific model, we
leased and sold units are different from each other, then have the advantage of comparing cars that are similar
a firm can benefit from a strategy that involves in terms of features and optional equipment. Our sam-
concurrent leasing and selling. Given this result, the is- ple contains the average auction prices of sold and off-
sue is whether leased and sold units indeed depreciate lease cars on a monthly basis from January 1994 to Sep-
at different rates? We address this empirical question tember 1996. Thus the sample begins from the point
by comparing depreciation rates based on actual market where the 1993 car is approximately one-year old. This
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

prices for a sample of off-lease and sold cars.9 is typical of the secondary market because most con-
sumers tend to hold onto their cars for at least one year.
5.1. Market Depreciation Rates In other words, the secondary market becomes large
The automobile market provides a rich arena in which enough to obtain accurate market prices usually after a
to test the robustness of an important assumption in our minimum of one year of sales. In addition, we have the
model. However, obtaining the prices of leased and sold average odometer reading of these cars in each month.
cars proves to be considerably complicated. The prob- To test whether leased and sold cars depreciate at dif-
lem arises because all publicly available price data on ferent rates, we estimated the following regression
used cars do not differentiate between ex-leased and equation
sold units. For example, the N.A.D.A. Official Used Car
Guide gives the average retail and wholesale prices by Ln Pricet Å a / b1TIMEt / b2TIME 1 LEASE
vintage of automobile, but it does not make any adjust-
/ b3 ln MILESt / et . (15)
ments for whether the car is used or off-lease. Although
the automobile manufacturers have this information, where PRICEt is the price at time t; TIMEt is a trend
they are reluctant to part with these data because of variable; TIME 1 LEASE is an interaction between the
proprietary concerns. These constraints limit research- trend variable (TIMEt ) and a dummy for off-lease cars;
ers’ access to these data. MILESt is the average mileage of the car at time t; and
However, we were able to get preliminary data from et is the error term.
A.D.T. Corporation, one of the largest wholesale auto In our model of depreciation (Equation 15), we as-
auction houses in the country. The number of cars that sume that the intercept ( a ) is the same for leased and
cycled through A.D.T.’s auctions in 1995 were approx- sold cars. This assumption reflects the fact that our his-
imately two million. Thus, because these average tory of prices is from the point where the car is approx-
wholesale prices are based on a large number of sales, imately one year old. At this point, there is no difference
we are assured that these prices are reflective of the between the prices of used and off-lease vehicles in our
overall market. sample in that both are considered ‘‘almost new’’; thus,
Our sample reflects the market prices of a 1993 model they should have the same intercept. We also tried to
car. For reasons of confidentiality, we can not disclose include the interaction between LEASE and MILES in
the name of the manufacturer or the model. Although the model. Unfortunately, including this interaction cre-
a single model limits the generalizability of our results, ates severe multicollinearity problems. Finally, we use
we emphasize that the car model under study repre- a log linear functional form to capture the depreciation
sents one of the largest selling nameplates in the U.S. In of the car. This form captures the notion that the dollar
amount of depreciation decreases with the age of the
9
We have also analyzed this issue in an experimental setting with a 2 car. In addition, this functional form has been used ex-
1 2 design in which car segment (luxury and non-luxury) and age tensively in other research on automobile pricing (e.g.,
(two- and four-year old) were the two factors. Subjects were asked for Agarwal and Ratchford 1980; Feenstra 1988).
their preferences between identical used and ex-leased cars (they were
Table 1 presents the estimates of the regression equa-
also given an option to indicate if they were indifferent between them).
We found that 84% of our subjects did not view ex-leased and used
tion after correction for first-order autocorrelation. First,
cars as being similar (i.e., they did not choose the indifferent option). we comment on the face validity of the model. The co-
Details of this study are available from the authors. efficient of TIME ( 00.015, p õ 0.01) suggests that on an

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S31

3b2f no07 Mp 31 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

Table 1 Estimates of Depreciation more control over the secondary market. Second,
given the long life of a durable, consumers may per-
Parameter Standard
ceive the depreciation of a durable to depend on
Variable Estimate Error T-Value
whether it was originally leased or purchased, which
INTERCEPT 11.306 0.48 0.0001 then affects market prices. And third, in choosing
TIME 00.015 0.002 0.0001 durables, consumers try to forecast their long-term
TIME 1 LEASE 0.002 0.001 0.05 needs which have different implications for their will-
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

Ln MILES 00.193 0.048 0.0002

ingness to pay for leases and purchases.
N 66
R-Square 0.9
Given our focus on depreciation, we abstract away
Durbin-Watson 1.88 other issues that can potentially play an important
role in durables markets. While this abstraction is nec-
essary to get a handle on the problem, an important
annual basis, the log of this model’s price decreases by goal for future research is make the model more gen-
00.015. This translates to a depreciation of 16.47% eralizable. In particular, we see intriguing possibili-
per year. This is consistent with other studies on ties for conducting further empirical analyses and ex-
depreciation rates that find an average depreciation of tending the analytic model. On the empirical side, we
15–18% per year across a wide range of models (Purohit plan to explore how leases and sales are related to the
1992). Second, note that price also decreases with the quality of the product (Desai and Purohit 1998). On
odometer reading or the number of miles (MILES, the modeling front, there are three important exten-
00.193, p õ 0.01) on the car. This is reasonable because sions that can help overcome some of the limitations
we would expect depreciation to increase with usage— of the current model. First, we need to incorporate
in this case, mileage. Thus, increasing mileage by 10% products with multiple levels of quality. This can help
decreases the price of the car by 1.9%. us address the question of how leases and sales vary
The null hypothesis is that off-lease and sold cars de- by product quality. Second, a limitation of our current
preciate at identical rates. The variable, TIME captures approach is our use of a two-period model (in prin-
the shape of the depreciation curve. The interaction ciple, an n-period model). Thus, we need to explore
term, TIME 1 LEASE (0.002, p õ 0.05) tests whether ways of modeling this problem in an infinite horizon.
off-lease cars depreciate at a different rate from sold And third, we need to study the consequences of al-
cars. Based on our results, we reject the null hypothesis lowing depreciation to be endogenous. One such pos-
of equal depreciation. In particular, we find that off- sibility would model a consumer’s choice between a
lease cars depreciate at an annual rate that is 2.4% lower lease and a purchase in a random utility framework.
than sold cars. In this case, a demand shock could affect consumer
preferences, that in turn, could affect depreciation
rates.10 These caveats notwithstanding, we believe
6. Conclusions that this model represents an important step in un-
In this paper, we address the firm’s problem in choos- derstanding the idiosyncrasies of marketing durables.
ing between leases and sales to market its durable We began the paper by asking whether a strategy
product. This general choice between leases and sales of concurrent leasing and selling may hurt a manu-
is crucial because of three issues that affect the firm’s facturer’s profitability. An important result of our
marketing strategy over the long term. First, leases analysis is that a combination of leasing and selling
and sales lead to different forms of competition in the is better for the manufacturer than either selling alone
future. That is, once a firm sells a durable, the product or leasing alone. Our analysis also suggests that in
exists in a competitive secondhand market that com-
petes with the firm’s sales of new products. On the 10
For a further discussion of problems of endogeneity in choice mod-
other hand, if the product is leased, it is returned to els, see Villas-Boas and Winer (1997) and Besanko, Gupta, and Jain
the firm at the end of the lease; hence, the firm has (1997).

S32 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 32 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

terms of profits, it is better to target the ‘‘lower’’ qual- References

ity product (i.e., higher depreciation product) to the Automotive News 1994. Auto auctions prepare for the onslaught of off-
lease vehicles.
higher willingness to pay (top) group of the market,
Bagnoli, S., S. Salant, J. Swierzbinski. 1989. Durable goods monopoly
and the ‘‘higher’’ quality product to the lower will-
with discrete demand,’’ J. Political Economy 97 , 1459–1478.
ingness to pay (middle) group. Because the top group Besanko, D., S. Gupta, D. Jain 1997. Logit demand estimation under
always ends up using a new, high quality car in each competitive pricing behavior: An equilibrium framework. Forth-
period, these consumers are somewhat less concerned coming Management Sci.
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

about the future degradation in quality. Although a Bond, E., L. Samuelson 1984. Durable good monopolies with rational
expectations and replacement sales. Rand J. Econom. 15 , 336–345.
higher depreciation rate lowers the total price these
Bucovetsky, S., J. Chilton 1986. Concurrent renting and selling in a
consumers are willing to pay, it is still optimal for the durable goods monopoly under threat of entry. Rand J. Econom.
firm to target this group with the ‘‘lower’’ quality car. 17 , 261–278.
Traditional wisdom suggests that the higher quality Bulow, J. 1982. Durable-Goods Monopolist. J. Political Economy 90 (2)
product should be targeted to the higher willingness 314–332.
Business Week 1994. Leasing fever. (February 7) 92–97.
to pay group. However, we find that when high will-
Coase, R. 1972. Durability and monopoly. J. Law and Econom. 15 (April)
ingness to pay consumers like to turn their products 143–149.
over frequently, then it is optimal to target them with Conlisk, J., E. Gerstner, J. Sobel. 1984. Cyclic pricing by a durable goods
the lower quality product. monopolist. Quarterly J. Econom. 99 489–505.
An important result of our analysis is that we are Connelly, M. 1994. Ford coaches Taurus turn-ins. Automotive News
(August 1) 1.
able to extend the standard results in the literature on
DeGraba, P. 1994. No lease is short enough to solve the time inconsis-
durable goods. In particular, starting with Coase’s tency problem. J. Indus. Econom. 42 361–374.
seminal analysis, it has been well accepted that selling Desai, P.D. Purohit. (1998). Competition in durable goods markets: the
poses particular problems for the durable goods mo- strategic consequences of leasing and selling. Forthcoming Mar-
nopolist. Hence, the firm should follow a pure leasing keting Sci.
strategy. In this paper, we show that the former ar- Feenstra, R. 1988. Gains from trade in differentiated products: Japa-
nese compact trucks. Robert C. Feenstra, ed. Empirical Methods for
guments rely on the important assumption that leases
International Trade, MIT Press, Cambridge, MA. 119–136.
and sales do not differentially affect the quality of the Gruber, G. 1994. Recovering hidden values in retail lease portfolios. J.
used product. By allowing leases and sales to depre- of Retail Banking, 14 ,(3) 35–42.
ciate at different rates, we show that a strategy of con- Gul, F., H. Sonnenschien, R. Wilson. 1986. Foundations of dynamic
current leasing and selling is preferable for the man- monopoly and the Coase conjecture. J. Econom. Theory, 39 155–
190.
ufacturer. Intuitively, one can expect different depre-
Henry, J. 1994. The Party’s Over. Automotive News (August 22) 1i.
ciation rates to give the manufacturer a broader Kahn, C. 1986. The durable goods monopolist and consistency with
product line, and this broader line can lead to higher increasing costs. Econometrica 54 , (2) 275–294.
profitability. However, this argument is only partially Levinthal, D., D. Purohit 1989. Durable goods and product obsoles-
correct. In particular, with two levels of ‘‘used’’ goods cence. Marketing Sci. 8 , (1) Winter, 35–56.
Mann 1992. Durable goods monopoly and maintenance. Internat. J.
(i.e., ex-leased and used), the manufacturer is better
Indust. Organization 10 65–79.
able to protect the market for the new product. That
Moorthy, K. S. 1984. Market segmentation, self selection, and product
is, the ex-leased product can be thought of as a fight- line design. Marketing Sci., 3 , (4) 288–307.
ing brand that takes on the competition from the used I. P. L. Png 1992. Market segmentation, cannibalization, and
market.11 the timing of product introductions. Management Sci., 38 , (3)
345–359.
Mussa, M. S. Rosen 1978. Monopoly and Product Quality. J. Econom.
Theory 18 , 301–317.
11
We thank Dipak Jain, an associate editor, and two anonymous re- Purohit, D. 1992. Exploring the relationship between new and used
viewers for their comments on earlier versions of this paper. We are durable goods: The case of automobiles. Marketing Sci., 11 (2) 154–
also grateful to Jim Anton, Kyle Bagwell, Miguel Villas-Boas and Rich 167.
Staelin for helpful discussions. We thank Tom Kontos and A.D.T. Cor- , R. Staelin 1994. Rentals, sales and buybacks: Managing second-
poration for providing us with the data for our study. ary distribution channels. J. Marketing Res. 31 (3) 325–338.

MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998 S33

3b2f no07 Mp 33 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07
 DESAI AND PUROHIT
Optimal Marketing Strategies

Rechtin, M. 1994. Japanese gamble residuals for share. Automotive Waldman, M. 1996. Planned obsolescence and the R&D decision. Rand
News (July 18), 4. J. Econom. 27 (3) (Autumn), 583–595.
Serafin, R. 1994. Automakers look for new lease on loyalty. Automotive 1997. Eliminating the market for secondhand goods: An alter-
News (July 18). native explanation for leasing. J. Law and Econom. XL (1) (April)
Stokey, N. 1981. Rational expectations and durable goods pricing. Bell 61–92.
J. Econom. 12 (Spring), 112–128. Wall Street Journal 1994a. Automobiles. Dow Jones & Co., February 18,
Thomas, C. 1994. Dealer scramble for used cars; demand soars. Auto- B1.
motive News (May 2) 44. 1994b. Automobiles. June 3, B1.
Downloaded from informs.org by [128.122.253.212] on 10 May 2015, at 22:10 . For personal use only, all rights reserved.

Tirole, J. 1988. The Theory of Industrial Organization, The MIT Press, 1995a. Autos: New car leases come loaded with options. April
Cambridge, MA. 12, B1.
Villas-Boas, J. M., R. Winer. 1997. Endogeneity in brand choice models. 1995b. Autos: Leased-car glut may lift costs, deflate values. June
Working Paper, University of California, Berkeley. 3, B1.

Accepted by Dipak C. Jain; received March 27, 1996. This paper has been with the authors 18 months for 3 revisions.

S34 MANAGEMENT SCIENCE/Vol. 44, No. 11, Part 2 of 2, November 1998

3b2f no07 Mp 34 Wednesday Dec 09 12:36 PM Man Sci (November, Part 1) no07