Omega 54 (2015) 134–146

Contents lists available at ScienceDirect

Omega
journal homepage: www.elsevier.com/locate/omega

The comparison between trade-in and leasing of a product

with technology innovations$
Kate J. Li a,n, Susan H. Xu b,†
a
Department of Information Systems and Operations Management, Sawyer Business School, Suffolk University, Boston, MA 02108, USA
b
Department of Supply Chain and Information Systems, Smeal College of Business, The Pennsylvania State University, University Park, PA 16802, USA

art ic l e i nf o a b s t r a c t

Article history: Companies can adopt trade-in and/or leasing to shorten consumers' upgrade cycle and gain control over
Received 12 June 2014 secondary markets. In this paper, we consider a monopolistic manufacturer who offers a technology
Accepted 31 January 2015 product to a market consisting of heterogeneous consumers. We focus on an exogenous, stochastic
Available online 9 February 2015
innovation process that determines the availability of new technology and consequently, residual value
Keywords: of the current product. We derive the optimal pricing strategy of trade-in and leasing, respectively,
Trade-in examine its impact on the manufacturer's expected profit, and compare the performance of the two
Leasing strategies. Trade-in protects the manufacturer against residual value risk and allows the flexibility of
Durable goods offering the option at different innovation states separately. Leasing, on the other hand, provides the
Product reuse
manufacturer an opportunity to circumvent low new product prices and thus increases expected profit
Stationary equilibrium
when product reuse profitability is high. The interplay between the two forces, product reuse profit-
ability and new product price, determines the preference between trade-in and leasing. Our findings
provide monopolistic manufacturers guidance on how to optimally employ the trade-in and leasing
strategies.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction
Trade-in and leasing are two widely adopted strategies by indivi-
dual consumers and companies in a variety of industries. Consumers
regularly trade in automobiles, electronics, power tools, video games,
golf clubs, etc. towards the purchase of new products. In the business-
to-business (B2B) markets, many companies trade in large-scale
communication devices, CT scanners, photocopies, computers, etc.
when they replace their current equipment. On the leasing side,
equipment leasing has a long history and has been adopted by
companies around the world [41]. It accounts for about $827 billion
of business each year in the United States, and will continue to be a
dynamic and growing business [11]. It has been used by companies of
different sizes to acquire a broad range of assets, such as aircraft,
automobiles, IT, medical, manufacturing and construction equipment.
According to ELFA, 72 percent of U.S. companies lease some or all of
their equipment. In recent years, in spite of the challenges posed by a
slow economy, low interest rates and falling equipment prices, IT
leasing remains a viable and vibrant industry [39]. Leasing has also
been widely used in business-to-consumer (B2C) markets, mainly for
acquiring automobiles and technology products. For example, new car
☆
This manuscript was processed by Associate Editor Wang.
n
Corresponding author. Tel.: þ 1 6175571586.
†
Deceased.
E-mail addresses: kjli@suffolk.edu (K.J. Li), shx@psu.edu (S.H. Xu).
leasing basically did not exist in the early 1980s, but experienced
steady growth in the 1980s and 1990s with reaching a peak in around
2000 [20]. Recently battery leasing has been introduced by automobile
manufacturers and power suppliers in order to reduce the purchase
cost of electronic vehicles [24].
Manufacturers are increasingly offering trade-ins and leasing,
and using the returned products for remanufacturing, recycling,
and other types of product reuse. Xerox is one of the best-known
examples. Its remanufacturing facility is based partly on returns
from trade-ins and off-lease products, which has generated cost
savings of several hundred million dollars each year [14]. Interface
Inc., a carpet manufacturer, provides Evergreen Lease with the goal
of reducing the environmental impact of its operations and
describing it as a “new workable business model for sustainable
development” [29]. Other examples include IBM, HP, Herman
Miller, Bosch, and Eletrolux [33].
With the ever-increasing rate of technology innovation and
product replacement, trade-in and leasing play a crucial role in
helping companies to stay competitive, increase consumer satis-
faction by providing more flexible acquisition methods, and be
environmentally responsible. The examples above show that
companies have been implementing trade-in and/or leasing, in
combination with outright selling. Trade-in and leasing enable
companies to achieve two basic goals. First, the two strategies can
be used to incentivize consumers to upgrade to products featuring
newly introduced technology or simply to replace used products

http://dx.doi.org/10.1016/j.omega.2015.01.018
0305-0483/& 2015 Elsevier Ltd. All rights reserved.
 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146
with brand-new ones. Second, since trade-in and off-lease pro-
ducts are returned to the companies, they can gain better control
over product reuse and secondary market activities in order to
achieve higher profit, fight counterfeits, and engage in sustainable
operations.
In this paper, we intend to study the implementation of trade-in
and leasing under stochastic technology innovation, examine their
impact on consumer behavior and firm profitability, and compare
the two strategies under different pricing and product reuse
scenarios. Specifically, we consider a setting where a monopolistic
manufacturer offers a technology product to a heterogeneous
consumer population who differ in their willingness-to-pay (WTP)
for the product. The technology product is designed to last for two
periods functionally, i.e., after it has been used for two periods, it
fully depreciates functionally and has no residual value remained.
We assume that there is an exogenous, stochastic innovation
process governing the availability of the next-generation technol-
ogy. Whenever a new technology becomes available, the manufac-
turer introduces a new product featuring the latest technology to
the market and stops offering new products featuring the previous-
generation technology. A product thus is subject to two types of
value decay: functional depreciation (FD), caused by wear and tear,
and technological obsolescence (TO) caused by technological inno-
vation. Trade-in and leasing can be offered by the manufacturer to
encourage consumers to replace products suffering FD and/or TO to
brand-new products featuring the latest technology.
In the trade-in model, the manufacturer may offer trade-in credit to
buy back consumers' on-hand products, which can be applied towards
the purchase new products. Meanwhile, consumers who have a one-
period-old product on hand decide whether to take advantage of the
trade-in offer or to keep their product for one more period until it fully
depreciates functionally. If a consumer chooses to trade in when
innovation does not occur, he essentially replaces a used, one-
period-old product suffering only FD by a brand-new product with
the same technology. This decision is made based on his own
willingness-to-pay and the trade-in credit provided by the manufac-
turer. Since such used products are still technologically current, the
manufacturer sells them in the marketplace to consumers with lower
willingness-to-pay. If a consumer chooses to trade in when innovation
occurs, the consumer is able to upgrade to the new technology
immediately after it becomes available. In essence, the consumer can
enjoy a brand-new product featuring the latest technology instead of a
used product suffering both FD and TO. However, because now these
trade-in products are used and obsolete, it is usually not to the best
economic interest of the manufacturer to resell them given her costs of
inspection, remanufacturing, and remarketing; instead, the manufac-
turer is better off disassembling, reusing or recycling materials, or
donating these products to charity.
Evidence in practice supports the manufacturer's different dispos-
ing methods of trade-in products at different innovation states. Apple
has a “Reuse and Recycling Program”1. Currently, the Reuse Program
accepts iPhones as old as iPhone 4 and gives customers gift cards
based on model number, capacity, and condition of the device. But for
older models (including iPhone, iPhone 3G, and iPhone 3GS), custo-
mers can only participate in the Recycling Program which does not
provide any monetary compensation, but simply recycles for free for
customers. Besides Apple, other companies, such as Cannon, Best Buy,
and Costco, offer programs operated in a similar way. A trade-in value
search conducted in October 2014 using Best Buy's online trade-in
calculator shows that the trade-in values of iPhone 3G, iPhone 3GS,
iPhone 4, iPhone 4s, iPhone 5, and iPhone 5s (all with 16 GB memory)
are: $0, $5, $60, $110, $201, and $300, respectively.2 In the B2B markets,
1
See www.apple.com/recycling/gift-card/for details
2
See www.bestbuy.com/site/Electronics-Promotions/Online-Trade-Infor details
135
asset recovery services are a popular offering from HP Financial
Services (HPFS) that take back HP and non-HP equipment from a
wide range of industrial customers. HPFS took in roughly one million
units in 2006 [7]. Remarketing older assets is a key part of HPFS's
business; however, if a piece of equipment is too old (severe
technology obsolescence), or in such a bad shape that it no longer
has any useful life left (severe functional depreciation), HPFS will mine
the equipment for any useful parts and then manage the disposition of
the remainder in accordance with applicable environmental regula-
tions. These B2B- and B2C-market examples demonstrate that when
the resell value of trade-in products is higher than the associated costs,
companies sell refurbished and/or used products in the market;
otherwise, they dispose these products by reusing or recycling
materials or donating to charity.
Note that though operated in the same way, trade-ins offered
when innovation occurs and when innovation does not occur
serve different purposes to the manufacturer and the consumers.
When innovation occurs, the manufacturer uses trade-in to buy
back technologically obsolete, used products in order to accelerate
new technology adoption; when innovation does not occur, the
manufacturer uses trade-in to buy back technologically current,
used products and sells them alongside new products featuring
the same technology. As such, the manufacturer essentially serves
as an intermediary to facilitate the transaction between consumers
who want to replace their functionally depreciated, one-period-
old products with brand-new ones and consumers who demand
used products. The manufacturer decides trade-in credits in
anticipation of consumer choice to maximize her expected profit;
while consumers choose consumption strategy based on their own
willingness-to-pay and trade-in credits offered by the manufac-
turer in order to maximize their own utility. From the manufac-
turer's perspective, prices are her enabler to induce consumer
behavior that results in her maximum expected profit. Since we
are interested in analyzing the long-run behavior of the manu-
facturer and the consumers, we characterize the optimal station-
ary prices.
The leasing model is set up in a similar way where consumers
may be offered the option of leasing a new product. All leases are
designed to last for one period and it is assumed that no call
option is embedded. If a consumer leases a new product, he
returns the off-lease product to the manufacturer after one period
when the lease ends and assumes no further responsibility. Same
as in the trade-in model, the manufacturer sells off-lease products
in the marketplace when innovation does not occur, and disposes
them outside the primary marketplace otherwise. The manufac-
turer decides lease prices, given which the consumers choose their
consumption strategy to maximize their own utility.
The trade-in and leasing models are solved separately and their
results are compared. We find that the manufacturer's trade-in
decision is affected by two factors, namely new product price and
product reuse profitability. When new product price is low, a relatively
low trade-in credit is enough to get consumers interested. Therefore,
the manufacturer can increase her profit by encouraging more
frequent product replacement and upgrade through trade-in. Product
reuse profitability is the value that the manufacturer can recover from
traded-in products. The higher this is, the more likely that the
manufacturer offers trade-in. An important characteristic of the
trade-in model is the manufacturer's flexibility to make separate
decisions of offering trade-in when innovation occurs and when it
does not. This is because consumers participating in trade-in return
their used product when they purchase a new one, and the manu-
facturer can dispose these used products promptly under observed
innovation condition. Therefore, for example, if selling used products
to lower-valuation consumers is lucrative, but recycling materials is
not, the manufacturer can choose to offer trade-in only when
innovation does not occur.
136 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146
The manufacturer loses this flexibility when leasing is
employed. The distinction is that when leasing is offered, off-
lease products are returned to the manufacturer a period later
when the lease term ends. Given the stochastic innovation process,
the manufacturer has to resort to the expected gain of product
reuse (considering disposing used products both when innovation
occurs and when it does not) to aid her decision. Consequently, the
expected gain of product reuse has to be high enough for the
manufacturer to offer leasing. The difference between trade-in and
leasing can also be understood by examining the risk-bearing
situation. As discussed earlier, a one-period-old, functionally
depreciated product may also be technologically obsolete if
innovation occurs. We term the risk caused by the uncertain
innovation process as the residual value risk. The ownership of
the one-period-old products determines the party who assumes
this risk. It is clear to see that when trade-in is offered, it is the
consumers who bear the residual value risk because they buy the
product outright first and the manufacturer decides later whether
she wants to buy back these products; while under leasing, it is the
manufacturer who carries this risk because off-lease products go
back to the manufacturer nonetheless. Indeed, Pierce [30] finds
that for automobile leasing, industry-wide failures in effectively
managing this residual value risk led to losses of $10–11 billion
between 2000 and 2001, and the average loss per leased returned
vehicle was $3269 in 2002. Although from the manufacturer's
perspective, leasing is disfavored because of the risk-bearing
situation, it may still be preferred when the current new product
prices are too low. Raising prices usually is not a viable option, so
leasing gives the manufacturer an opportunity to offer the product
in another way to increase her expected profit. The interplay of the
two factors, new product price and product reuse profitability,
determines the preference between trade-in and leasing.
The rest of the paper is organized as follows. In Section 2, we
review related literature. Section 3 details model setup, solution
procedure, and results of the trade-in model. In Section 4, we
present the leasing model with a focus on its difference from the
trade-in model. We compare trade-in and leasing strategies and
discuss the managerial implications in Section 5, and provide
concluding remarks and future research opportunities in Section 6.
2. Literature review
Our paper is primarily related to the streams of literature in
economics studying durable-goods monopolist offering sequen-
tially improving products, and various studies about trade-in and
leasing in operations management and marketing.
Coase [6] famously postulates the time inconsistency problem
for a durable-goods monopolist of which the monopolist might
not be able to capture any monopoly profits if she can change
product price over time. An extensive body of research has been
built upon [6] and the stream that is most relevant to our paper
considers monopolists supplying improving durable goods. In the
two-period model set up by [10] and a followup work by [22], the
authors assume that a low-quality version of a product exists in
the first period and a high-quality version becomes available in the
second period. The monopolist chooses first- and second-period
prices and second-period quality level to maximize her discounted
profit, while consumers decide whether and when to purchase the
product to maximize their utility. Silver [37] is concerned with the
pricing of similar goods of different vintages coexisting in a
market. It examines the differential pricing over the life of the
update and its effect on the existing vintages as well. No second-
hand markets exist in [10,22,37]. Waldman [45] studies a monopo-
list's R&D decision where her investment amount determines the
probability that an improved product becomes available in the
second period. It is shown that if the monopolist sells rather than
rents her output, she has an incentive to overinvest. Fudenberg
and Tirole [16] analyze pricing for sequentially improving products
under different market information conditions. Two types of
products are considered: those with an active secondhand market
and anonymous consumers, and those with no secondhand
market and identified consumers. In the former case, it is shown
that the monopolist may choose to produce, buy back, or keep the
earlier version stock constant once the new version becomes
available. In the latter case, the monopolist chooses to offer
upgrade discounts.
In the papers discussed above, new product pricing is the key
decision made by a monopolist. Besides pricing, there are papers
examining the timing of introducing improved products. Levinthal
and Purohot [23] consider a setting where a monopolist may
introduce an improved version of the current product. The authors
consider various policy measures available to the monopolist and
identify the optimal sales strategy under different product
improvement scenarios. The authors find that, for modest levels
of product improvement, the monopolist's optimal policy is to
phase out sales of the old product; while for large improvements, a
buy-back policy is more profitable. Different from using aggregate
demand function as in [23,28] considers a monopolist who sells to
two groups of consumers with different willingness-to-pay. The
monopolist has a high- and a low-quality version of a product
available simultaneously and needs to decide how to introduce
and price the two versions. The authors find that if the monopolist
decides to introduce the two versions sequentially, the high-
quality version should be introduced first.
This body of literature mainly adopts a two-period model setup
where innovation occurs with certainty in the second period or is
subject to the monopolist's investment or introduction decision.
Given the complexity of today's business environment, the devel-
opment and introduction of new technology usually is not deter-
mined by one company. We are interested in examining how a
monopolist can use trade-in and leasing to cope with and take
advantage of technology innovation. Therefore, in our study, we
use an infinite-horizon model to consider a case where innovation
is governed by an exogenous stochastic process. The availability of
a new technology is uncertain and is not determined by the
monopolist. We examine the long-run behavior of a monopolistic
company and consumers maximizing their utilities. Moreover, in
order to focus on the aspect of product improvement, this stream
of literature generally ignores physical wear and tear. We, on the
other hand, are interested in studying the impact of both techno-
logical obsolescence and functional depreciation on trade-in and
leasing decisions.
Trade-in and leasing encourage product replacement, promote
product reuse, and enable companies to gain control over the
secondary markets, which potentially can impact a range of
closed-loop supply chain issues that have been extensively studied
such as supply chain network design (e.g., [38,15,36,46,25]),
product design (e.g., [5,8,26,12,47]), consumer return process and
policy (e.g., [40,21,17,18,34]), and pricing decisions of reusable
components and remanufactured products (e.g., [13,44,27,42,31]).
In operations management and marketing literature, the most
relevant stream of literature examining the application of trade-in
and leasing. For trade-in, [43] studies quasi-durable goods, i.e.,
products that are infinitely lasting, but with diminishing service,
and therefore consumers do not necessarily leave the market after
their first purchase. The authors compare two commonly used
marketing practices: offering discounts to either repeat buyers
(trade-ins) or new buyers (introductory offers). The authors find
that trade-ins are optimal for durable products and introductory
offers are optimal for consumable products when consumers are
myopic. Ray et al. [33] examine three trade-in pricing schemes
 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146
under a single-period model setting: uniform pricing, age-
independent trade-in pricing, and age-dependent trade-in pricing.
In both [43,33], no technology innovation is considered and trade-
in is used exclusively to increase purchase frequency. Another
paper related to ours is [32]. Motivated by automobile trade-ins,
lemon problems and adverse selection are major focuses of the
paper and trade-in is proved to be an intervention by the
monopolist to reduce inefficiencies caused by lemon problems.
Rao et al. [32] assume constant technology as well. On the
contrary, Agrawal et al. [1] assume that a product is subject to
full technological obsolescence, i.e., consumers need to replace
their product in every period. Hence, in their model, trade-in is not
used as an incentive for consumers to repurchase, but purely a
mechanism for price discrimination.
On the leasing side, Desai and Purohit [9] analyze the co-
existence of leasing and selling. The paper shows that the relative
profitability of leasing and selling hinges on the rates at which
leased and sold units depreciate, which results in the different
optimal strategy that a firm should implement. Huang et al. [19]
construct a dynamic game to simultaneously explain concurrent
leasing and selling, active secondary markets for used goods,
heterogeneous consumers, and nontrivial transaction costs. Leas-
ing is considered as a facilitator of price discrimination, which de-
bundles a durable good into new and used portions. Instead of
different rates of depreciation, Huang et al. [19] cite the compara-
tive advantage of the monopolist disposing used goods as the
explanation of concurrent leasing and selling. Bhaskaran and
Gilbert [3,4] also study the use of leasing, but extend the scope
to considering complementary products and channel structure.
Agrawal et al. [2] investigate whether leasing is environmentally
superior to selling. The authors find that leasing can be envir-
onmentally worse than selling despite full remarketing, or greener
than selling despite the premature disposal of off-lease products,
depending on firm, product, and consumer characteristics.
The aforementioned papers examine a wide array of questions
regarding trade-in and leasing. Our paper complements this body
of literature from the following two aspects. First, all these papers
consider a single product with no technology innovation involved.
In our study, we examine the role that trade-in and leasing play
when product offering depends on a stochastic innovation process.
Second, we not only study trade-in and leasing separately, but also
compare their performance under different new product pricing
and product reuse profitability scenarios.
3. The trade-in model
3.1. The model setup
We consider an infinite-horizon, discrete-time model in which
a monopolistic manufacturer offers a technology product to
heterogeneous, utility-maximizing consumers, to maximize her
profit. Specifically, our model consists of the following constructs.
3.1.1. Innovation process and product lifetime
The functional life of each new product is designed to last for
two periods, after which the product is fully depreciated function-
ally. We assume that there is an exogenous innovation process that
governs the introduction of the next-generation technology of the
product to the market. The innovation process follows a geometric
distribution, with q0 being the probability that a new technology
becomes available in the next period, and q1 being the probability
otherwise, where q1 ¼ 1 q0 . The state of the innovation process in
each period is denoted by a binary variable j, where innovation
state 0 means that a new generation of technology just becomes
available and innovation state 1 means otherwise.
137
3.1.2. The manufacturer and the trade-in option
At any time, the monopolistic manufacturer, who faces no
production capacity constraint and lead time, sells new products
featuring the latest technology to consumers. We assume that new
product prices at states 0 and 1 are fixed, denoted as pN N
0 and p1 ,
respectively, in order to focus on the comparison between trade-in
and leasing for a manufacturer who has already been in the
market selling new products outright, but is interested in accel-
erating product replacement due to more frequent technology
innovations and gaining control over secondary market.
The manufacturer may offer two trade-in opportunities to
consumers: state-0 trade-in when a new generation of technology
becomes available, and state-1 trade-in otherwise. During periods
when innovation does not occur, sold products undergo wear and
tear caused by normal usage while its technology remains the
latest. We refer to product value decay caused by usage as
functional depreciation (FD). The manufacturer may allow consu-
mers to trade in their one-period-old products and will resell
these trade-in units in the marketplace because their technology is
still current. Consumers who have a one-period-old product on
hand may decide to take the trade-in offer and use the given
trade-in credit as a partial payment to purchase a new product, or
keep the unit for one more period until it depreciates fully
functionally.
When innovation occurs, the new technology will be launched
immediately, and the previous-generation technology will become
obsolete and will no longer be adopted by new products. Under
this situation, besides functional depreciation, one-period-old
products also suffer technological obsolescence (TO), which refers
to product value decay caused by the innovation process. The
manufacturer may also offer trade-in, but she will dispose the one-
period-old products outside of the marketplace by disassembling
and reusing parts, recovering materials, donating to charity, etc.
That is because due to both FD and TO, the value retained by these
products may not worth the cost of reselling in the secondary
market. Consumers with one-period-old products again may
choose to trade in or continue using their products for one more
period. In essence, after a product is sold and used for one period,
if innovation does not occur, the value of the product is reduced
because of functional depreciation; however, if innovation occurs,
the value of the product is decreased due to both functional
depreciation and technological obsolescence.
The manufacturer controls trade-in credits, pTj;t Z 0, where t is
the time period index, to maximize her expected profit. For
convenience, we call a used product a state-j used product when
the innovation process is in state j. When the manufacturer resells
one-period-old products in the marketplace, she charges pU1;t ,
which is determined endogenously by the market clearance
condition. That is, pU1;t is charged such that available trade-in
products are sold out. Selling state-1 used products incurs a per-
unit cost γ to the manufacturer, which includes costs of inspection,
repair, refurbishing, and remarketing. Note that this cost applies
only in state 1 when the manufacturer resells used products, but
does not apply in state 0 when she disposes them outside of the
market. We further assume that when the manufacturer disposes
state-0 used products, she obtains a per-unit net profit, β. The
manufacturer also has a unit production cost, c. We assume that
γ o c and β o c.
Though operated in the same way, state-0 and state-1 trade-in
programs serve different purposes to the manufacturer. At innova-
tion state 0, the manufacturer buys back technologically obsolete
products in order to accelerate new technology adoption; while at
innovation state 1, since new and used products have the same
technology and co-exist in the marketplace, the manufacturer
essentially serves as an intermediary who facilitates the transac-
tion between consumers who want to replace their functionally
138 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146

Table 1
Notation summary.

Parameters
q0 Probability that a new technology becomes available in the next period
j Innovation state
c Production cost
γ Remanufacturing and remarketing cost
β Net profit of disposing used products outside of the market
θN N U U
0 , θ1 , θ0 , θ1 H-consumers' valuations of state-0 new, state-1 new, state-0 used,
and state-1 used products, respectively
wH, wL H- and L-consumers' sensitivity measures
nH, nL Sizes of H- and L-consumers
a A consumer's action taken in the current period
a0 A consumer's action taken in the previous period
ilk Consumer states, k ¼ 1; …; 4, l ¼ TI; LS
π il Limiting probability of consumer state ilk
k

us ðpl Þ Consumers' long-run average utility of adopting consumption

strategy s given pl , l ¼ TI; LS
ul ðpl Þ The manufacturer's long-run average profit given pl , l ¼ TI; LS
pU1;l Used product price at state 1, l ¼ TI; LS (endogenously determined)
Decision variables
pTj Stationary trade-in price at state j
pLj Stationary leasing price at state j

in state-1 new products). Therefore, θ0  θ1 can be considered as a

depreciated, one-period-old products with brand-new ones and
consumers who demand used products.
3.1.3. The consumers
The population of consumers is constant and a consumer can
own at most one unit of the product in any period. Consumers
have universally agreed, common attribute values θ~ 0 , θ~ 1 , θ~ 0 , and
N N
N N
measure of technological obsolescence. Moreover, wear and tear
caused by product usage deteriorates the value of state-1 used
products even though their technology is still up-to-date. Hence,
θN1  θU1 , the valuation of functional difference between a state-1
new and a state-1 used product, measures functional depreciation.
U

~ U
θ 1 , which represent the one-period utility of consuming a state-0
new product, a state-1 new product, a state-0 used product, and a
state-1 used product, respectively. Because of the memoryless
property of the innovation process, these attribute values only
depend on the innovation state but are independent of the age of
the technology. We assume that 0 o θ~ 0 o θ~ 1 o θ~ 1 o θ~ 0 to reflect
U U N N
that new technology is valued more than existing technology, and
a brand-new product is valued more than a used product featuring
the same technology.
Although consumers agree upon these common attribute
values, they differ in their sensitivity to them, which captures
consumer heterogeneity. We assume that the consumer popula-
tion consists of two types of consumers: H-consumers, with
sensitivity measure w ~ H , and L-consumers, with sensitivity mea-
sure w ~ L , where w ~ H Zw ~ L Z 0. Let wL ¼ w ~ L =w
~ H , which is L-consu-
mers' sensitivity defined as a fraction of that of H-consumers. The
value of wL captures the degree of heterogeneity of the consumers.
A lower wL means that H- and L-consumers are more hetero-
geneous in terms of their willingness-to-pay. Therefore, it can be
expected that in general, wL is lower for B2C markets than for B2B
markets. Each consumer's sensitivity type is known only to the
consumer himself. A consumer's valuation for a product is his
sensitivity measure multiplied by the common attribute value of
the product. For example, an H-consumer values a state-0 new
~ H θ~ 0 . For H-consumers, we define θN
N N
product at w 0 ¼ wHθ0 ,
θN1 ¼ w~ H θ~ 1 , θU0 ¼ w~ H θ~ 0 , and θU1 ¼ w~ H θ~ 1 . Consequently, a L-con-
N U U
sumer's valuations of a state-0 new, state-1 new, state-0 used, and
state-1 used products can be represented as wL θ0 , wL θ1 , wL θ0 ,
N N
and wL θ1 , respectively. The sizes of H- and L-consumers are nH and
U
3.2. Rule of the game and solution characterization
The manufacturer and the consumers are assumed to be
rational and to maximize their own expected payoff. We study
an infinitely repeated game with complete information by assum-
ing that all payoff-relevant information is common knowledge.
Uncertainty in this repeated game stems from the stochastic
innovation process captured by innovation probability q0. At the
beginning of each period, the manufacturer announces the trade-
in credit, and at innovation state 1, used product price is deter-
mined by the market clearance condition. The utility-maximizing
consumers then choose a consumption strategy based on their
own type, H or L, the trade-in credit offered by the manufacturer,
and used product price determined by the market.
Figure 1 illustrates the sequence of events. In this game, the
consumers play strategically against the manufacturer, but not
against each other.
In our analysis, we focus on stationary pricing strategies that can
be deployed by the manufacturer. A stationary pricing strategy
depends only on the innovation states, but is independent of time
and market share [35]. Let the stationary price vector be
pTI ¼ ðp0;TI ; p1;TI Þ ¼ ðpT0 ; pT1 ; pU1;TI Þ, which are state-0 trade-in credit,
state-1 trade-in credit, and the price of state-1 used product,
respectively, where pU1;TI is endogenously determined so is not a
~ ~
decision variable of the manufacturer. A consumer's action, a, in
each period depends on innovation state j, the corresponding price
vector pj;TI , and his action in the previous period a0 (because the
U functional lifespan of a new product is two periods, we only need

nL, respectively, and we assume that nH r nL . A table of notations is

provided in Table 1. 0,t
T
1,t
T

Note that innovation state affects the universally agreed attri-

bute values and consumer valuations of the products. The differ-
ence, θ0  θ1 , reflects the valuation of technological difference
N N

between the latest technology (i.e., featured in state-0 new

products) and the previous-generation technology (i.e., featured Fig. 1. Sequence of events.
 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146
to track a consumer's action in the previous period). Therefore, a
consumer's state variable, iTI, can be expressed as ða0 ; pj;TI Þ.
3.3. Consumer behavior—stationary consumption strategies
and long-run average payoffs
If an H-consumer took action a0 ¼ N in the previous period,
where N denotes purchasing a new product, he can choose to
(i) trade in and purchase a new product at a discounted price of
pNj  pj , also denoted by N, or (ii) keep his on-hand product for one
T
more period, denoted by action K.3 If he took action a0 a N in the
previous period, denoted as a0 ¼ N (i.e., he does not have a usable
unit on hand), he purchases a new product. Note that we do not
consider the case where an H-consumer does not make a purchase
when he does not have a usable unit on hand because it results in
the entire market being inactive which is a trivial case. A
consumer's single-period payoff is denoted by rða; iTI Þ. We have
four distinct consumer states, labeled as iTI
1 ¼ ðN; p0;TI Þ, i2 ¼ ðN
; p0;TI Þ, iTI ¼ ðN; p Þ, and iTI
¼ ðN ; p Þ. The single-period
3 1;TI 4 1;TI
of an H-consumer are summarized in Table 2. We assume that it is
never profitable for the manufacturer to sell new products to L-
consumers. Hence, L-consumers buy used products when they are
available at state 1 (if H-consumers decide to trade in) and obtain
payoff wL θ1  pU1;TI , denoted by action U, or remain inactive other-
U
wise, denoted by action I.
Given that we are interested in analyzing the long-run behavior
of the manufacturer and the consumers, we choose expected long-
run average payoff as the optimization criterion. In responding to
the manufacturer's stationary pricing vector pTI , a consumer
chooses a stationary consumption strategy that maximizes his
long-run average utility per period. A stationary consumption
strategy of a consumer is a function that maps each possible
consumer state i to a particular feasible action a. From our earlier
discussion, for H-consumers, aiTI ; aiTI A fN; Kg, and aiTI ¼ aiTI ¼ fNg;
1 3 2 4
for L-consumers, aiTI ¼ I and aiTI A fU; Ig (note that iTI
1 and i3 are
2 4
unreachable states for L-consumers). Subsequently, there are a
total of four feasible stationary consumption strategies for H-
consumers: AH;TI ¼ fS1TI ; S2TI ; S3TI ; S4g, and two for L-consumers:
AL;TI ¼ fS5; S6g. We label these strategies as S1TI S6 (see Table 3 for
details).
The limiting probabilities, denoted by π iTI , k ¼ 1; …; 4, and the
k
long-run average utilities, denoted by us ðpTI Þ, s ¼ 1; …; 6, are shown
in Table 3. We use S2TI as an example to explain the calculation of
the limiting probabilities. When S2TI ¼ ðK; N; N; NÞ is deployed, the
transition probability matrix is
2 3
0 q0 0 q1
6 7
6 q0 0 q1 0 7
P¼6 7 ¼ max
6 q 0 0 q 1 0 7:
4 5
q0 0 q1 0
Limiting probabilities, π iTI , π iTI , π iTI and π iTI , can be obtained by
1 2 3 4
solving a system of linear equations. The objective of consumers is
to maximize their long-run average utility per period, which is the
difference between the reward obtained and the price paid by a
consumer. We explain the computation of u1 ðpTI Þ as an example.
The long-run average utility of an H-consumer adopting S1TI
ðN; N; N; NÞ, which stipulates the purchase of a new product in
each period regardless of innovation state and the action taken in
3
We assume that due to high transaction cost, it is not economically plausible for
consumers to sell their used products directly to other consumers or to dispose
them elsewhere.
139
the previous period, is
u1 ðpTI Þ ¼ π 1iTI rðN; iTI
1 Þ þ π iTI rðN; i2 Þ þ π iTI rðN; i3 Þ þ π iTI rðN; i4 Þ
1 TI 1 TI 1 TI
1 2 3 4
¼ ðq0 θ0 þ q1 θ1 Þ ðq0 ðpN
N N T N T
0  p0 Þ þ q1 ðp1  p1 ÞÞ:
Intuitively, the above expression states that, over the long run, an
H-consumer adopting S1TI earns a payoff of ðθ0  pN
N
0 þ p0 Þ in q0 
T
100% of periods and a payoff of ðθ1  p1 þ p1 Þ in q1  100% of
N N T
periods because he consumes a new product in each period by
paying the discounted price after trading in.
Taking into account of the clearance condition required at
innovation state 1, there are four feasible strategy pairs for
H- and L-consumers: ðS1TI ; S5Þ, ðS2TI ; S5Þ, ðS3TI ; S6Þ, and ðS4; S6Þ,
where the first and second entries in a set of parentheses denote
H- and L-consumers' consumption strategies, respectively. The
formation of these feasible strategy pairs guarantees that when
H-consumers choose to trade in their one-period-old products at
innovation state 1, L-consumers will buy these used products from
TI the manufacturer. Recall that we assume the size of L-consumers
payoffs exceeds that of H-consumers, so the clearance condition can
always be satisfied.4 Hence, the resulting price of state-1 used
products determined by the clearance condition is pU1;TI ¼ wL θ1 .
n U
3.4. The manufacturer's optimization problem
3.4.1. Problem setup
The manufacturer decides the stationary pricing scheme pTI , in
anticipation of the consumers' reaction, to maximize her expected
long-run average profit per period. By choosing different pricing
schemes, the manufacturer essentially induces the consumers to
adopt different consumption strategies. Consumers' choices of
consumption strategies in turn decide the aggregate market shares
of new and used products, which determine the manufacturer's
overall profit.
In essence, the optimal stationary price vector pnTI is the
TI
manufacturer's enabler to induce consumer behavior that results
in her maximum long-run average profit. We first formulate the
manufacturer's long-run average profit maximization problem for
each of the four feasible strategy pairs: ðS1TI ; S5Þ, ðS2TI ; S5Þ,
ðS3TI ; S6Þ, and ðS4; S6Þ. We define uTI ðpTI Þ as the manufacturer's
long-run average profit, which is the sum of the long-run average
payments from the two consumer groups playing different con-
sumption strategies, subtracted by the manufacturer's long-run
average cost of enabling such strategies. We use the strategy pair
ðS1TI ; S5Þ as an example. When the manufacturer induces ðS1TI ; S5Þ,
her maximization problem is
uTI
n
fuTI
ðS1;S5Þ ¼ maxpTI ðS1;S5Þ ðpTI Þg
0  p0 Þ þ q1 ðp1  p1 Þ  c þ q0 β þq1 ðwL θ 1  γ ÞÞg
U
fnH ðq0 ðpN T N T
pTI
ð1Þ
s:t: u1 ðpTI Þ Z u2 ðpTI Þ; ð2Þ
u1 ðpTI Þ Zu3 ðpTI Þ; ð3Þ
u1 ðpTI Þ Zu4 ðpTI Þ; ð4Þ
where Inequalities (2)–(4) ensure that H-consumers choose S1TI .
The optimization problems based on ðS2TI ; S5Þ and ðS3TI ; S6Þ can be
formulated in a like manner. We provide only their objective
4
Note that this assumption simplifies exposition, but is not crucial for the
analysis. If the size of H-consumers is greater than that of L-consumers, we can
simply assume that the manufacturer will dispose excess used products outside of
the market as she does at innovation state 0.
140 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146

functions here: We solve the manufacturer's profit maximization problem in

( !) two steps. First, we examine the constraint set of each of the four
q1 T c  q1 ðwL θ1  γ Þ
U
n q20 N optimization problems to determine whether dominated strategy
uTI
ðS2;S5Þ ¼ max nH p0 þ q1 pN
1  p  ;
pTI 1 þ q0 1 þ q0 1 1 þ q0 pairs exist. If so, we exclude these strategy pairs from further
ð5Þ consideration. Second, we solve for the optimal stationary price
vector pnTI for inducing each strategy pair and calculate the
   manufacturer's maximum expected long-run average profit. We
n q q2 c  q0 β
uTI
ðS3;S6Þ ¼ max nH q0 pN
0  0 pT0 þ 1 pN 1  ; ð6Þ then compare the obtained maximum profits and identify the
pTI 1 þ q1 1 þ q1 1 þq1 optimal strategy pair under different parameter value scenarios.
and the manufacturer's average profit given ðS4; S6Þ is nH ððq0 =
2ÞpN0 þ ðq1 =2Þp1  c=2Þ.
N
3.4.2. Dominant consumption strategies given different new product
Before presenting the solution procedure and the result, we
price regions
discuss two factors, new product price and product reuse profit-
As shown in Table 3, new product prices affect consumers'
ability, which determine the manufacturer's decision. When new
long-run average utility, hence the choice of consumption strate-
product price at an innovation state is low, trade-in at that
gies. We first examine the existence of dominant strategies under
innovation state is favored for two reasons. First, because of the
low price, the manufacturer does not make considerable profit
Table 4
from each sale; therefore she needs to encourage more frequent Two factors driving trade-in decisions.
purchases by offering trade-ins to increase profit. Second, the
manufacturer does not need to offer a high trade-in credit to New product price (pN
j )

incentivize consumers to buy a new unit because its price is

Profitability of product reuse (vj) High Low
already low. Now consider the other factor, product reuse profit-
ability at innovation states 0 and 1, defined as v0 ¼ β  θ0 and
U
High DEPENDS Favors state-j TI
v1 ¼ ðwL θ1  γ Þ  θ1 ¼ ðwL  1Þθ1  γ , respectively. v0 is the value
U U U
Low Disfavors state-j TI DEPENDS
difference between disposing state-0 used products outside of the
market (i.e., β) and letting H-consumers keep them for one more
period (i.e., θU0 ); and v1 is this value difference at state 1. A higher
product reuse profitability implies that the manufacturer can
obtain higher value from disposing used products, therefore
trade-in is more favored; otherwise, trade-in should be discour-
aged. Table 4 summarizes the impact of the two factors on the
manufacturer's trade-in decision. When new product price is low
and product reuse profitability is high, trade-in is clearly favored;
when the conditions are reversed, trade-in is not a preferable
option. When the two factors pull trade-in decision in opposite
directions, the final outcome will depend on their relative
strength.

Table 2
The single-period payoff matrix of an h-consumer in the trade-in model.

Action a Consumer state

iTI
1 ¼ ðN; p0;TI Þ iTI
2 ¼ ðN ; p0;TI Þ iTI
3 ¼ ðN; p1;TI Þ iTI
4 ¼ ðN ; p1;TI Þ

N θN
0  p0 þ p0
N T
θN
0  p0
N
θN
1  p1 þ p1
N T
θN
1  p1
N

K θU
0 a θU
1 
Fig. 2. Dominant consumption strategy pairs under different new product pricing
a
 denotes infeasible actions. scenarios.

Table 3
Consumption strategies, limiting probabilities, and long-run average utilities in the trade-in model.

Strategy Consumer state Limiting probabilities Long-run average utility

iTI
1 iTI
2 iTI
3 iTI
4 π iTI π iTI π iTI π iTI us ðpTI Þ
1 2 3 4

   
S1TI N N N N q0 0 q1 0 q0 θN N
0 þ q1 θ1  q0 ðp0  p0 Þ þ q1 ðp1  p1 Þ
N T N T
!  
S2TI K N N N q0 q20 q1 q0 q1 q20 θN U
0 þ q0 θ 0 q20 q1
1 þ q0 1þ q0 1 þ q0 q1 θN
1 þ  0 þ q1 p1 
pN N
pT1
1 þ q0 1 þ q0 1 þ q0 1 þ q0
!  
S3TI N N K N q0 q0 q1 q1 q21 q21 θN U
N 1 þ q1 θ 1 q0 q2
1 þ q1 1 þ q1 1þ q1 q0 θ0 þ  q0 p0 N
p þ 1 pN
T
1 þ q1 1 þ q1 1 þ q1 0 1 þ q1 1
S4a K N K N q0 q0 q1 q1 q0 θN N U U q
0 þ q1 θ 1 þ q0 θ0 þ q1 θ1 q
2 2 2 2  0 pN þ 1 pN
2 2 0 2 1
S5a – I – U 0 q0 0 q1 U
q1 wL θ1  q1 pU1;TI
S6a – I – I 0 q0 0 q1 0

a
S4–S6 do not have “TI” as subscript because as will be shown later that they are also applicable to the leasing model.
 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146 141

different new product pricing scenarios. The result is summarized three regions shown in Fig. 2 and determine their corresponding
by Theorem 1 and Fig. 2. dominant strategies. □

Theorem 1. The existence of dominant consumption strategy pairs
depends on the new product prices at both innovation states and the
magnitude of their difference. The dominant strategy pairs in each of
the four new product pricing regions, as shown in Fig. 2, are:
Fig. 2 shows the dominant consumption strategy pairs in
different new product pricing regions. In practice, companies
usually either keep the price of a product constant throughout
its life cycle (i.e., pN0 ¼ p1 ) or reduce price after the product has
N
been introduced for some time (i.e., pN 0 4p1 ). Hence, we focus on
N

 In Region 1, the Low Price Region, ðS1TI ; S5Þ is the only dominant
strategy pair for H- and L-consumers.
 In Region 2, the Balanced Pricing Region, none of the four strategy
pairs is dominated.
 In Region 3, the Low Introductory Pricing Region, ðS2TI ; S5Þ is
dominated.
 In Region 4, the Deep Discounting Region, ðS3TI ; S6Þ is dominated.
Proof of Theorem 1. Table 3 contains the long-run average utility
of the four consumption strategies, S1TI –S4, among which
H-consumers can choose. We compare the long-run average
utilities to determine whether dominant strategies exist under
different new product pricing scenarios. We use the optimization
the area under the 45-degree line. Note that we keep ðS4; S6Þ, the
outright-selling-only strategy pair, available in all four regions
because it represents the status quo of the market.
In Region 1, the Low Price region, the impact of new product
price outweighs that of product reuse profitability; thus ðS1TI ; S5Þ
is the only dominant strategy pair. In essence, unless the manu-
facturer (or a legislature) can demand a disposal fee, which is
equivalent to a negative trade-in credit, H-consumers always
prefer trade-in. Region 2 is labeled as the Balanced Pricing region
where new product prices at the two innovation states are
commensurate with each other. No dominated strategies exist
because neither state's trade-in is definitively preferred to the
other. We label the region where pN N
1 is relatively close to p0 the

problem defined for ðS1TI ; S5Þ, Eqs. (1)–(4), as an example. It is Low Introductory Pricing region, i.e., Region 3 in Fig. 2. In this
obvious that if (2)–(4) hold, S1TI dominates S2TI , S3TI , and S4. By region, ðS2TI ; S5Þ is dominated. Recall that S2TI is the state-1-trade-
rearranging the three inequalities, we have in-only strategy. When pN 1 is relatively high, state-1 trade-in
becomes less attractive compared to state-0 trade-in; therefore,
1 þq0 T pN  θ0 H-consumers favor S1TI or S3TI . In other words, when state-1
pT1 Z  p þ 0 ; ð7Þ
q1 0 q1 trade-in is worthwhile to H-consumers, state-0 trade-in is at least
as worthwhile. Region 4, the Deep Discounting region where pN 1 is
q0 T pN  θ1 much lower than pN 0 , represents a scenario where the manufac-
pT1 Z  p þ 1 ; ð8Þ
1 þq1 0 1 þ q1 turer discounts heavily once the product is not newly introduced.
Intuitively, contrary to Region 3, ðS3TI ; S6Þ is dominated in this
0 þ q1 p1  ðq0 θ 0 þ q1 θ 1 Þ
q0 T q0 pN N
pT1 Z  p þ ; ð9Þ region.
q1 0 2q1

where θ0 ¼ θ0  θ0 and θ1 ¼ θ1  θ1 . When pN 0 r θ 0 , the right-

N U N U
hand side of Inequality (7) is non-positive and therefore (7) holds
regardless of parameter values. Similar argument can be applied to
Inequalities (8) and (9). Hence, when pN 0 r θ 0 and p1 r θ 1 , S1TI
N
dominates the other three strategies, i.e., it is the only strategy that
should be induced by the manufacturer when new product prices
fall in Region 1. Following similar analysis, we can define the other
3.4.3. Optimal consumption strategy given different product reuse
scenarios
Now we derive and discuss the optimal consumption strategy
pair that the manufacturer should induce in each new product
pricing region. We solve for optimal trade-in credits, pT0n and pT1n ,
and the resulting maximum expected long-run average profit
given each dominant strategy pair in each region. We then

Fig. 3. The optimal consumption strategy pairs in the trade-in model. (a) the balanced pricing region, (b) the low introductory pricing region and (c) the deep discounting
region.
142 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146

q0 θ0 þ q1 θ1 cÞ=q1 (which is the derivation of Line 3), ðS1TI ; S5Þ

compare the maximum profits and the strategy pair that yields the
maximum profit is the optimal pair. The results are summarized in
Theorem 2 and Fig. 3.
Theorem 2. The optimal strategy pair that maximizes the manufac-
turer's expected long-run average profit depends on cost parameters,
consumer valuations, and innovation frequency. The stationary
equilibrium can be specified as follows. The optimal trade-in credits
and the manufacturer's maximum long-run average profit are pro-
U U
results in higher maximum average profit; otherwise, ðS2TI ; S5Þ is
preferred. Such pairwise comparisons are conducted and the result is
summarized by Fig. 3(a). The other three regions can be analyzed
analogously. □
The result of Region 2, the Balanced Pricing region (as sum-
marized by Fig. 3(a)) demonstrates the flexibility of trade-in.
When pN N
0 and p1 are balanced, i.e., neither is exceedingly high or

vided in Table 5.
 In Region 1, the Low Price region, when c r q0 pN0 þq1 pN1 þ
2q0 β þ 2q1 ðwL θ1  γ Þ, ðS1TI ; S5Þ is the optimal strategy pair;
U the manufacturer should induce ðS1TI ; S5Þ to maximize her profit
otherwise, the manufacturer should induce ðS4; S6Þ.
 In Region 2, the Balanced Pricing region, the optimal strategy pair
given different conditions of product reuse profitability is illu-
strated by Fig. 3(a). When v0 and v1 are both large, ðS1TI ; S5Þ is the
optimal strategy pair; when v0 is large and v1 is small, ðS3TI ; S6Þ is
optimal; when the condition is reversed, ðS2TI ; S5Þ is optimal;
otherwise ðS4; S6Þ is optimal.
 In the Low Introductory Pricing region (Region 3), the optimal
strategy pair is illustrated by Fig. 3(b). When v0 and v1 are both
large, ðS1TI ; S5Þ is the optimal strategy pair; when v0 is large and
v1 is small, ðS3TI ; S6Þ is optimal; otherwise ðS4; S6Þ is optimal.
 In the Deep Discounting region (Region 4), the optimal strategy
pair is illustrated by Fig. 3(c). When v0 and v1 are both large,
ðS1TI ; S5Þ is the optimal strategy pair; when v0 is small and v1 is
large, ðS2TI ; S5Þ is optimal; otherwise ðS4; S6Þ is optimal.
Proof of Theorem 2. It is obvious that the manufacturer's long-run
average profit maximization problem of any of the feasible strategy
pairs is a linear program and therefore the optimum is always attained
on one or more vertices of the feasible region, which can be solved by
standard methods. Here we use Region 2, the Balanced Pricing region
as an example. In this region, all four consumption strategy pairs are
non-dominated, so we first solve for the optimal trade-in credits that
maximize the manufacturer's long-run average profit, as defined by
Eqs. (1)–(6). The results are provided in Table 5. Then we compare the
maximum long-run average profit achieved under each strategy pair
and determine the optimal strategy pair given different product reuse
profitability scenarios (i.e., different values of v0 and v1). For example,
by comparing the maximum profits obtained by inducing ðS1TI ; S5Þ,
and ðS2TI ; S5Þ, we have that when v1 4  ð1 þ q0 Þ=q1 v0  ðθ0 þ
N
low compared to the other, product reuse profitability plays the
predominant role in determining the manufacturer's optimal
trade-in decision. Not surprisingly, when both v0 and v1 are high,
because she can recover more value from product reuse; when
both are low, the manufacturer remains to sell outright only.
Otherwise, she can exploit different product reuse situation at
different innovation states independently. In essence, when pro-
duct reuse is rather profitable at state 1 but not at state 0 (i.e., v0 is
small but v1 is large), the manufacturer chooses to induce ðS2TI ; S5Þ
where H-consumers only trade in at state 1; when the situation is
reversed, she encourages H-consumers to pursue trade-in only at
state 1. Since the manufacturer decides trade-in offer after obser-
ving the innovation state, she can completely shield herself from
the residual value risk of used products.
Recall that in Region 3, the Low Introductory Pricing region
ðS2TI ; S5Þ, the state-1 trade-in only strategy pair, is always domi-
nated. Hence, as shown in Fig. 3(b), when v0 is small and v1 is
large, the manufacturer sells outright instead of offering state-1
trade-in as she does in the Balanced Pricing region. This is because
even though the reuse of state-1 used products is profitable, state-
1 trade-in is disfavored because of the prevailing effect of the
relatively high new product price at state 1. This exemplifies the
interplay of the two factors driving the trade-in decision, new
product price and product reuse profitability, discussed in Section
3.4.1. Region 4, the Deep Discounting region, is the opposite to
Region 3 and can be understood analogously.
4. The leasing model
One of the objectives of this paper is to compare trade-in and
leasing, so we adopt the same basic model setup and solution
concept, as explained in Sections 3.1 and 3.2, for the leasing model.

Table 5
Optimal consumption strategy and maximum average profit in different new product pricing regions.

Region Optimal trade-in credits Optimal strategy & maximum profit

Region 1 pT0n ¼ 0 ðS1TI ; S5Þ : nH ðq0 ðpN U

0 þ βÞþ q1 ðp1 þ wL θ 1  γÞ cÞ
N

(Low price Region) pT1n ¼ 0

 
Region 2 ð1 þ q1 ÞpN
0  q1 p1  ð1þ q1 Þθ 0 þ q1 θ 1
N
0 þ q1 p1 þ q0 θ0 þ q1 θ 1
q0 pN N
pT0n ¼ ðS1TI ; S5Þ : nH þ q0 β þ q1 ðwL θU1  γÞ  c
2 2
(Balanced Pricing Region) Tn  q0 pN
0 þ ð1 þ q0 Þp1 þ q0 θ 0  ð1 þ q0 Þθ1
N
p1 ¼
2
 
q0 pN q pN q q θ0  q1 ð1 þ q0 Þθ1 q1 ðwL θU1  γÞ c
ðS2TI ; S5Þ : nH 0
þ 1 1  0 1 þ 
2 2 2ð1þ q0 Þ 1þ q0 1 þ q0
 
q pN q pN q ð1þ q1 Þθ0  q0 q1 θ1 q β c
ðS3TI ; S6Þ : nH 0 0 þ 1 1 þ 0 þ 0 
2 2 2ð1þ q1 Þ 1 þ q1 1 þ q1
 
0 þ q 1 p1  c
nH q0 pN N
ðS4; S6Þ :
2
 
Region 3 pT0n ¼ 0 q21 q
ðS1TI ; S5Þ : nH q0 pN0 þ pN þ 1 θ1 þ q0 β þ q1 ðwL θU1  γÞ  c
1 þ q1 1 1 þ q1
 
(Low Introductory Pricing Region) 1  θ1
pN q21 1  
pT1n ¼ ðS3TI ; S6Þ : nH q0 pN0 þ pN þ q βc
1þ q1 1 þ q1 1 1 þ q1 0
 2 
Region 4 pN  θ 0 q0 q0
pT0n ¼ 0 ðS1TI ; S5Þ : nH pN0 þ q1 p1 þ
N
θ0 þ q0 β þ q1 ðwL θU1  γÞ  c
1þ q0 1 þ q0 1 þ q0
 2 
(Deep Discounting Region) pT1n ¼ 0 q0 1
ðS2TI ; S5Þ : nH pN þ q1 p N
þ ðq ðwL θU1  γÞ cÞ
1 þ q0 0 1
1 þ q0 1
 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146 143

In the leasing model, the manufacturer provides the option of problem is

leasing, along with outright selling, to consumers where all leases n
n o
last for one period and do not have embedded call options. We uLS
ðS3;S5Þ ¼ max uLS
ðS3;S5Þ ðpLS Þ
pLS
present consumer behavior in the leasing model, solve the ( !)
c  q20 β  q0 q1 ðwL θ1  γ Þ
U
q1 N q
manufacturer's long-run average profit maximization problem, ¼ max nH p1 þ 0 pL0 
pLS 1 þ q1 1 þ q1 1 þ q1
and discuss the results in this section.
s:t: u3 ðpLS Þ Zu1 ðpLS Þ;
u ðpLS Þ Z u2 ðpLS Þ;
3

4.1. Consumer behavior—stationary consumption strategies and u3 ðpLS Þ Z u4 ðpLS Þ; ð10Þ

long-run average payoffs
In the leasing model, the stationary price vector offered by the
manufacturer is: pLS ¼ ðp0;LS ; p1;LS Þ ¼ ðpL0 ; pL1 ; pU1;LS Þ, which consists of
new product leasing prices at states 0 and 1, respectively, and the
used product price at state 1. pU1;LS is determined endogenously by
the market clearance condition. If an H-consumer took action
where the constraints ensure that H-consumers choose S3LS from
the choice set. The derivation of the objective function is similar to
that in the trade-in model. The optimization problem of the other
two strategy pairs can be formulated in the same way. We provide
their objective functions here:
n

ðS1;S5Þ ¼ maxfnH ðq0 p0 þq1 p1  c þ q0 β þ q1 ðwL θ 1  γ ÞÞg;

U
uLS L L
ð11Þ
a0 ¼ N in the previous period, regardless of the current innovation pLS
state, he keeps his one-period-old product for another period, K. If ( !)
c  q0 q1 β  q21 ðwL θ1  γ Þ
U
he took action a0 a N in the previous period, he can choose to n q0 N q
uLS
ðS2;S5Þ ¼ max nH p þ 1 pL  :
(i) purchase a new product, N, or (ii) lease a new product, denoted pLS 1 þ q0 0 1 þ q0 1 1 þq0
by L. Table 6 shows the single-period payoff matrix of an
ð12Þ
H-consumer in the leasing model. There are a total of four feasible
consumption strategies for H-consumers, labeled as S1LS –S3LS and Comparing to the objective functions of the trade-in model,
S4, the outright buying strategy as defined in the trade-in model. given by Eqs. (1), (5), and (6), there is a significant difference: in
the leasing model, the terms β and wL θ1  γ , which reflect the
U
Similarly, we assume that L-consumers' sensitivity measure wL is
so low that it is not profitable for the manufacturer to sell or lease profitability of product reuse at innovation states 0 and 1 respec-
new products to L-consumers. Therefore, they have the same tively, enter all objective functions regardless of the innovation
choice set, fS5; S6g, as that in the trade-in model. The limiting state at which leasing is offered; however, in the trade-in model,
probabilities and long-run average utilities of S1LS –S3LS are pre- β only appears in the objective function of strategies involving
state-0 trade-in, and wL θ1  γ only appears in those involving
U
sented in Table 7.
state-1 trade-in. Essentially, trade-in decision can be made by
considering product reuse profitability at different innovation
states separately; but leasing decision has to be made by jointly
4.2. The manufacturer's optimization problem considering both states. This reflects the inherently different risk-
bearing situation of the two strategies: consumers assume the
The manufacturer uses the stationary price vector pLS to induce
consumer behavior that maximizes her long-run average profit.
Following the same argument, the clearance price of state-1 used
products is pU1;LS ¼ pU1;TI ¼ wL θ1 . There are four feasible strategy
n n U

pairs for H- and L-consumers: ðS1LS ; S5Þ, ðS2LS ; S5Þ, ðS3LS ; S5Þ, and
ðS4; S6Þ. We now formulate the manufacturer's long-run average
profit maximization problems and use ðS3LS ; S5Þ as an example.
When inducing ðS3LS ; S5Þ, the manufacturer's maximization

Table 6
The single-period payoff matrix of an H-consumer in the leasing model.

Action Consumer state

a iLS iLS iLS iLS

1 ¼ ðN; p0;LS Þ 2 ¼ ðN ; p0;LS Þ 3 ¼ ðN; p1;LS Þ 4 ¼ ðN ; p1;LS Þ

N a θN
0  p0
N  θN
1  p1
N

L  θN
0  p0
L  θN
1  p1
L

K θU
0  θU
1 

a
 denotes that the action is infeasible. Fig. 4. The optimal consumption strategy pairs in the leasing model.

Table 7
Consumption strategies, limiting probabilities, and long-run average utilities in the leasing model.

Strategy Consumer state Limiting probabilities Long-run average utility

iLS
1 iLS
2 iLS
3 iLS
4 π iLS π iLS π iLS π iLS us ðpLS Þ
1 2 3 4

S1LS K L K L 0 q0 0 q1 ðq0 θN N
0 þ q1 θ1 Þ  ðq0 p0 þ q1 p1 Þ
L L

q0 q0 q1 q1
S2LS K N K L q20
1 þ q0 1 þ q0 1 þ q0
q0 θ N N 2 U U
0 þ q1 θ 1 þ q0 θ 0 þ q0 q1 θ 1 q pN þ q pL
 0 10 þ q 1 1
1 þ q0 1 þ q0 0
q0 q1 q0 q1
S3LS K L K N 1 þ q1 1 þ q1
q21
1 þ q1
q0 θ N N U 2 U
0 þ q1 θ 1 þ q0 q1 θ 0 þ q1 θ 1 q pN þ q pL
 1 11 þ q 0 0
1 þ q1 1 þ q1 1
144
Fig. 5. The comparison between trade-in and leasing in the balanced pricing region
(Region 2).
residual value risk associated with one-period-old products when
trade-in is adopted; when leasing is employed though, it is the
manufacturer who bears this risk.
The result of the leasing model is summarized by Theorem 3
and Fig. 4. The proof is similar to that of Theorem 2 and therefore
is omitted.
Theorem 3. The optimal consumption strategy pair given difference
product reuse profitability scenarios is shown in Fig. 4. The optimal
leasing prices are
0  q1 p1 þ q0 θ 0 þ q1 θ 1
ð1 þ q1 ÞpN
n
N
pL0 ¼ ;
2 product reuse profitability is high at both innovation states, the
n  q0 p0 þ ð1 þ q0 Þp1 þ q0 θ0 þ q1 θ1
N N
pL1 ¼ ;
2 ðS1TI ; S5Þ) in the trade-in model, and obtains the same expected
and the manufacturer's maximum expected profit given strategy pair
ðS1LS ; S5Þ is
0 þ q1 p1 þq0 θ 0 þ q1 θ 1 þ 2q0 β þ 2q1 ðwL θ 1  γ Þ  2c
U
n q0 pN N
uLS
ðS1;S5Þ ¼
2 considered, it is still more profitable to choose ðS1LS ; S5Þ than
Line 7 in Fig. 4 delineates two regions: one region where
product reuse profitability is high at both innovation states (i.e.,
the region above Line 7), and the other region where both are low
compared to how H-consumers value the corresponding used
products (i.e., the region below Line 7). Recall that both β and
wL θ1  γ are in the objective functions of the three strategy pairs
U
(Eqs. (10)–(12)) because off-lease products are returned to the
manufacturer one period later when innovation state is uncertain.
Therefore, it is preferred by the manufacturer to offer leasing only
when off-lease products can be reused profitably under both
innovation states; and when this criterion is indeed satisfied, the
manufacturer is better off offering leasing at both innovation
states, instead of offering it at just one of them. Hence, strategy
pairs ðS2LS ; S5Þ and ðS3LS ; S5Þ are never optimal.
5. The comparison between trade-in and leasing and the
managerial implications
We discuss the comparison between trade-in and leasing given
the four new product pricing regions defined in Fig. 2. Theorem 4
summarizes strategy preference in the different pricing regions.
Theorem 4. Leasing is the preferred strategy in Region 1, the Low
Price region; in Region 2, the Balanced Pricing region, trade-in is
always at least as good as leasing; in Region 3, the Low Introductory
Pricing region and Region 4, the Deep Discounting region, the
K.J. Li, S.H. Xu / Omega 54 (2015) 134–146
preference between trade-in and leasing depends on parameter
values.
Proof of Theorem 4. The comparison between trade-in and
leasing can be done by calculating the difference in the manufac-
turer's maximum expected profits obtained under both strategies
(shown in Table 5 and Theorem 3). We use Region 1, the Low Price
region as an example. The manufacturer chooses between
ðS1TI ; S5Þ and ðS4; S6Þ in the trade-in model; and between
ðS1LS ; S5Þ and ðS4; S6Þ in the leasing model. We have
0 þ q1 p1  q0 θ 0  q1 θ 1
q0 pN
n n
N
uTI LS
ðS1;S5Þ  uðS1;S5Þ ¼ r0
2
n
n
ðS1;S5Þ r uðS4;S6Þ in
in Region 1. In addition, ðS4; S6Þ is chosen when uTI
the trade-in model. Hence, leasing yields the same or higher
maximum expected profit than trade-in in this region. The
comparison in the other three new product pricing regions can
be performed in the same way. □
The preference of leasing in Region 1 is obvious. When new
product prices are low, leasing enables the manufacturer to
achieve higher profit because it can be priced independently from
the current selling price. The comparison in Region 2 is summar-
ized by Fig. 5. The shaded areas are where trade-in is preferred to
leasing; while the rest is where trade-in and leasing yield the
same expected profit. Therefore, trade-in always performs at least
as good as leasing in this region. Line 7, as defined in Fig. 4, divides
the optimal choice between ðS1LS ; S5Þ and ðS4; S6Þ in the leasing
model. First consider the area above Line 7, where in the leasing
model, the manufacturer should offer leasing at both innovation
states (i.e., ðS1LS ; S5Þ). When both v0 and v1 are large, i.e., when
manufacturer chooses to offer trade-in at both states (i.e.,
profit as leasing. When either v0 or v1 is smaller, trade-in exhibits
its advantage over leasing. For instance, when v0 is smaller, the
manufacturer stops offering trade-in at state 0 and switches to
ðS2TI ; S5Þ, the state-1 trade-in only strategy. When leasing is
:
outright selling, but the manufacturer's expected profit is ham-
pered by less lucrative product reuse at state 0. Hence, in the
shaded area above Line 7 and below Line 3 (marked as
“S2TI 4 S1LS ”), trade-in outperforms leasing. The same explanation
applies to the shaded area above Line 7 and below Line 4 where v1
is smaller. In the area below Line 7, when jointly considering v0
and v1, outright selling is preferred to leasing. However, by
offering state-specific trade-ins, the manufacturer can still benefit
from encouraging product replacement and reuse. In summary,
the flexibility of acting independently at different innovation
states makes tarde-in a preferred strategy in Region 2, the
Balanced Pricing region.
The comparison between trade-in and leasing in Regions 3 and
4 is presented in Figs. 6 and 7, respectively, and can be understood
as follows. Recall that in Region 3, the Low Introductory Pricing
region, the current new product price is relatively low when it is
first introduced to the market, i.e., pN0 is not much higher than p1 .
N
As summarized by Table 4, new product price and product reuse
profitability are the two factors driving the manufacturer's trade-
in decision, which in turn affects the preference between trade-in
and leasing. Compared to Region 2, the Balanced Pricing region,
new product price plays a dominant role in the Low Introductory
Pricing region. By comparing Fig. 6 to Fig. 5, two main differences
are observed. First is the dotted area in Fig. 6 where leasing yields
higher expected profit than trade-in, which does not exist in Fig. 5.
Trade-in is always at least as good as leasing in the Balanced
Pricing region; however, in the Low Introductory Pricing region,
although trade-in is more flexible, leasing may prevail because it
 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146 145

enables the manufacturer to circumvent the relatively low new
product price pN 0 by offering leasing at an independent price when
product reuse profitability is high at both innovation states.
Second is that under Balanced Pricing, when v0 is small and v1
is large, the state-1 trade-in only strategy S2TI is the most
preferred. Nonetheless, under Low Introductory Pricing, S2TI is
always dominated because of the relatively high pN 1 . Consequently,
when v0 is small, the manufacturer can no longer exploits the
flexibility of trade-in by resorting to the state-1 trade-in only
strategy, but has to sell outright. Analogous to that in the Low
Introductory Pricing region, reversing the situation at states 0 and
1 is true in Region 4, the Deep Discounting region and the result is
presented by Fig. 7. In short, when new product prices at the two
innovation states are disproportional, the flexibility of trade-in is
restrained and leasing may be favored because it can be priced
separately to better capture consumer valuation.
6. Conclusions and future research
In this paper, we study the problem of a monopolistic manufac-
turer offering a technology product to a heterogeneous consumer
population where trade-in or leasing can be provided either as an
incentive to accelerate innovation adoption or as a means to facilitate
replacing physically depreciated used products with brand-new ones.
We characterize the optimal stationary pricing strategy for the
manufacturer and the optimal consumption strategies that hetero-
geneous consumers should choose to maximize their own utility
under both the trade-in and leasing models. We then compare trade-
in and leasing in terms of the manufacturer's profitability.
Several key insights gained from our analysis are summarized
as follows. First, a main difference between trade-in and leasing is
the timing of the return of used products. Used products are
returned to the manufacturer immediately after consumers choose
to trade in. However, for leasing, consumers return their off-lease
products one period later when their leases end. Therefore when
trade-in is considered, the manufacturer has the flexibility to make
trade-in decision at different innovation states independently. The
manufacturer loses this flexibility when leasing is offered, which
exposes her to residual value risk caused by the stochastic
innovation process. Second, when current new product prices
are low, leasing may enable the manufacturer to increase profit
by charging a separate leasing price, as opposed to trade-in where
consumers still purchase at the current selling price. Third, a
manufacturer who considers using trade-in or leasing to promote
product upgrading and replacement can use our models to assist
her decision. As a first step, she should determine the new product
pricing region that she belongs to. If her current pricing scheme is
in the Low Price or the Balanced Pricing region, she knows
immediately that leasing or trade-in is her optimal strategy,
respectively. If her current pricing scheme is in the Low Introduc-
tory Pricing or the Deep Discounting region, she needs to further

Fig. 6. The comparison between trade-in and leasing in the low introductory pricing region (Region 3).

Fig. 7. The comparison between trade-in and leasing in the deep discounting region (Region 4).
146 K.J. Li, S.H. Xu / Omega 54 (2015) 134–146
examine the profitability of product reuse at different innovation
states. When new product price is relatively low at an innovation
state and product reuse profitability at that state is much higher
than that at the other state, trade-in is the preferred strategy
because of its flexibility and risk-free nature to the manufacturer.
Otherwise, leasing is the optimal strategy because it enables the
manufacturer to charge a higher price by introducing a new
buying option.
For future research, it is interesting to consider an endogenous,
stochastic innovation process where the manufacturer's invest-
ment and effort level can affect the success of innovation. It will
also be useful to examine how this endogenous innovation process
interplays with the manufacturer's pricing decision. Instead of the
two groups of consumers characterized in this paper, one can
study a continuous spectrum of consumers to better capture the
interaction between consumers and consequently, between the
primary and secondary markets. Moreover, we observe in practice
the coexistence of trade-in and leasing offered by the same
manufacturer. Our modeling framework can be adapted to study
the phenomenon.
References
[1] Agrawal V, Ferguson M, Souza G. Trade-in rebates for price discrimination and
product recovery. Working paper, 2014.
[2] Agrawal V, Ferguson M, Toktay LB, Thomas V. Is leasing greener than selling?
Management Science 2012;58(3):523–33.
[3] Bhaskaran SR, Gilbert SM. Selling and leasing strategies for durable goods with
complementary products. Management Science 2005;51(8):1278–90.
[4] Bhaskaran SR, Gilbert SM. Implications of channel structure for leasing or
selling durable goods. Marketing Science 2009;28(5):918–34.
[5] Chen C. Design for the environment: a quality-based model for green product
development. Management Science 2001;47(2):250–63.
[6] Coase R. Durability and monopoly. Journal of Law and Economics 1972;15
(1):143–9.
[7] Correa M. Treasure hunters. 〈www.hp.com〉; 2007.
[8] Debo LG, Toktay LB, van Wassenhove LN. Market segmentation and product
technology selection for remanufacturerable products. Management Science
2005;51(8):1193–205.
[9] Desai P, Purohit D. Leasing and selling: optimal marketing strategies for a
durable goods firm. Management Science 1998;44(11):S19–34.
[10] Dhebar A. Durable-goods monopolists, rational consumers, and improving
products. Marketing Science 1994;13(1):100–20.
[11] Equipment Leasing and Finance Association. 〈www.elfaonline.org〉.
[12] The compatibility of durable goods with contingent generic consumables.
Omega 41;2013:574–85.
[13] Ferguson ME, Toktay LB. The effect of competition on recovery strategies.
Production and Operations Management 2006;15(3):351–68.
[14] Fishbein BK, McGarry LS, Dillon PS. Leasing: a step toward producer respon-
sibility. Report from INFORM Inc., New York; 2000. 〈www.informinc.org〉.
[15] Francas, Minner DS. Manufacturing network configuration in supply chains
with product recovery. Omega 2009;37(4):757–69.
[16] Fudenberg D, Tirole J. Upgrades, tradeins, and buybacks. The RAND Journal of
Economics 1998;29(2):235–58.
[17] Hu W, Li Y, Govindan K. The impact of consumer returns policies on consign-
ment contracts with inventory control. European Journal of Operational
Research 2014;233(2):398–407.
[18] Huang X, Gu J, Ching W, Siu T. Impact of secondary market on consumer
return policies and supply chain coordination. Omega 2014;45:57–70.
[19] Huang S, Yang Y, Anderson K. A theory of finitely durable goods monopoly
with used-goods market and transaction costs. Management Science 2001;47
(11):1515–32.
[20] Johnson J, Schneider HS, Waldman M. The role and growth of new-car leasing:
theory and evidence. Journal of Law and Economics 2014;57(3):140–54.
[21] Ketzenberg M, Zuidwijk R. Optimal pricing, ordering and return policies for
consumer goods. Production and Operations Management 2009;18(3):344–60.
[22] Kornish LJ. Pricing for a durable-goods monopolist under rapid sequential
innovation. Management Science 2001;47(11):1552–61.
[23] Levinthal DA, Purohit D. Durable goods and product obsolescence. Marketing
Science 1989;8(1):35–56.
[24] Li Z, Ouyang M. A win-win marginal rent analysis for operator and consumer
under battery leasing mode in China electric vehicle market. Energy Policy
2011;39(6):3222–37.
[25] Longinidis P, Georgiadis MC. Integration of sale and leaseback in the optimal
design of supply chain networks. Omega 2014;47:73–89.
[26] Maukhopadhyay SK, Setoputro R. Optimal return policy and modular design for
build-to-order products. Journal of Operations Management 2005;23:496–506.
[27] Mitra S. Revenue management for remanufactured products. Omega 2007;35
(5):553–62.
[28] Moorthy KS, Png IPL. Market segmentation, cannibalization, and the timing of
product introductions. Management Science 1992;38(3):345–59.
[29] Olivia R, Quinn J. Interface's evergreen services agreement. HBS Case 9-603-
112. Cambridge MA: Harvard Business School; 2003.
[30] Pierce L. Organizational structure and the limits of knowledge sharing:
incentive conflict and agency in car leasing. Management Science 2012;58
(6):1106–21.
[31] Qiang Q, Ke K, Anderson T, Dong J. The closed-loop supply chain network with
competition, distribution channel investment, and uncertainties. Omega
2013;41(2):186–94.
[32] Rao RS, Narasimhan O, John G. Understanding the role of trade-ins and
durable goods markets: theory and evidence. Marketing Science 2009;28
(5):950–67.
[33] Ray S, Boyaci T, Aras N. Optimal prices and trade-in rebates for durable,
remanufacturable products. Manufacturing & Service Operations Management
2005;7(3):208–28.
[34] Ruiz-Benitez R, Ketzenberg M, van der Laan EA. Managing consume returns in
a high clockspeed industries. Omega 2014;43:54–63.
[35] Rust J. Stationary equilibrium in a market for durable assets. Econometrica
1985;53(4):783–805.
[36] Salema MIG, Barbosa-Povoa APFD, Novais AQ. Simultaneous design and
planning of supply chains with reverse flows: a generic modelling framework.
European Journal of Operational Research 2010;203:336–49.
[37] Silver M. Disparities in the prices of new and old models when they coexist in
the market: some econometric estimates. Omega 1999;27(1):13–24.
[38] Network design for reverse logistics, Omega 36:535–48.
[39] Steinback K. Maintaining a vibrant presence: independents continue to evolve
in IT leasing. Monitor 2005;32(7):1–2.
[40] Su X. Consumer returns policies and supply chain performance. Manufactur-
ing and Service Operations Management 2009;11:595–612.
[41] Ta HP, Gn HL. Singapore leasing industry: an empirical study. Omega 1987;15
(4):301–11.
[42] Vadde S, Zeid A, Kamarthi SV. Pricing decisions in a multi-criteria setting for
product recovery facilities. Omega 2011;39(2):186–93.
[43] van Ackere A, Reyniers DJ. Trade-ins and introductory offers in a monopoly.
Rand Journal of Economics 1995;26(1):58–74.
[44] Vorasayan J, Ryan SM. Optimal price and quantity of refurbished products.
Production and Operations Management 2006;15(3):369–83.
[45] Waldman M. Planned obsolescence and the R&D decision. The RAND Journal
of Economics 1996;27(3):583–95.
[46] Wang F, Lai X, Shi N. A multi-objective optimization for green supply chain
network design. Decision Support Systems 2011;51:262–9.
[47] Wu C. OEM product design in a price competition with remanufactured
products. Omega 2013;41:287–98.