ELSEVIER

Contents lists available at ScienceDirect

Operations Research Perspectives

journal homepage: www.elsevier.com/locate/orp

Pricing decision model for new and remanufactured short-life cycle products with time-dependent demand

Shu San Gana,c, I. Nyoman Pujawana'*, Suparnoa, Basuki Widodob

a Department of Industrial Engineering - Faculty of Industrial Technology, Sepuluh Nopember Institute of Technology, Surabaya 60111, Indonesia b Department of Mathematics - Faculty of Mathematics and Natural Science Sepuluh Nopember Institute of Technology, Surabaya 60111, Indonesia c Department of Mechanical Engineering - Faculty of Industrial Technology, Petra Christian University, Surabaya 60236, Indonesia

article info

abstract

Article history: Received 15 July 2014 Received in revised form 6 November 2014 Accepted 19 November 2014 Available online 3 December 2014

Keywords:

Short life cycle product Remanufacturing Closed loop supply chain Pricing Optimization

In this study we develop a model that optimizes the price for new and remanufactured short life-cycle products where demands are time-dependent and price sensitive. While there has been very few published works that attempt to model remanufacturing decisions for products with short life cycle, we believe that there are many situations where remanufacturing short life cycle products is rewarding economically as well as environmentally. The system that we model consists of a retailer, a manufacturer, and a collector of used product from the end customers. Two different scenarios are evaluated for the system. The first is the independent situation where each party attempts to maximize his/her own total profit and the second is the joint profit model where we optimize the combined total profit for all three members of the supply chain. Manufacturer acts as the Stackelberg leader in the independently optimized scenario, while in the other the intermediate prices are determined by coordinated pricing policy. The results suggest that (i) reducing the price of new products during the decline phase does not give better profit for the whole system, (ii) the total profit obtained from optimizing each player is lower than the total profit of the integrated model, and (iii) speed of change in demand influences the robustness of the prices as well as the total profit gained.

© 2014 The Authors. Published by Elsevier Ltd.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Technology-based product has shorter life cycle due to rapid innovation and development in science and technology, as well as customer behavior in pursuing latest innovation and style. Le-breton and Tuma [ 1 ] pointed out that technology based commodities such as mobile phones and computers have shorter innovation cycle so that the previous generation becomes obsolete faster, either functionally and psychologically. Similarly, Hsueh [2] also argued that product life cycle in electronic industry is shorter than before, due to technology advances, and as a result, an outdated product could reach its end-of-use even it is still in a good condition. Shorter life-cycle has negative contribution toward sustain-ability, since there is an increase in product disposal. Customers want newer products and discard the old ones, and these preferences would exhaust landfill space in shorter time. In addition,

* Corresponding author.

E-mail addresses: gshusan@peter.petra.ac.id (S.S. Gan), pujawan@ie.its.ac.id (I.N. Pujawan), suparno@ie.its.ac.id (Suparno), b_widodo@matematika.its.ac.id

(B. Widodo).

there are more natural resources and energy used to create new products than actually needed, due to unnecessary increased obsolescence. To make it worse, electronic products are prominent as the ones with shorter and shorter life cycle, while the wastes are toxic and not environmentally friendly. There are many attempts made in developed countries to control electronic wastes such as Waste of Electric and Electronic Equipment (WEEE) directives, implemented in most European countries since 2003, RoHS in United States, 2003, and Extended Producer Responsibility (EPR) issued by OECD in 1984. However, these regulations pose as burdens to the industries when implemented only for conformity, because there are additional costs for handling e-wastes and increased material cost for avoiding or minimizing toxic materials.

Several strategies have been introduced to mitigate products disposal and wastes, such as life cycle approach, regulation and society approach. One aspect of life cycle approach is dealing with products at their end-of-use. According to de Brito & Dekker [3], there are situations where customer has the opportunity to return a product at a certain life stage, which can be referred to leasing cases and returnable containers, and is called end-of-use return. Hsueh [2] considered a different kind of return, where a product may be returned because it has become outdated, and the customer

http://dx.doi.org/10.1016/j.orp.2014.11.001

2214-7160/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4. 0/).

wants to buy a new product. Herold [4] proposed alternatives to end-of-use products which are reprocessing, collect-and-sell, and collect-and-dispose. Remanufacturing is one option to manage products at their end-of-use which offers opportunity for complying with regulation while maintaining profitability [5-7].

Remanufacturing is a process of transforming used product into "like-new" condition, so there is a process of recapturing the value added to the material during manufacturing stage [8,9]. The idea of remanufacturing used products has gained much attention recently for both economic and environmental reasons. As suggested by Gray and Charter [9], remanufacturing can reduce production cost, the use of energy and materials.

There are numerous studies on remanufacturing. However, most of the published works on remanufacturing have considered durable or semi-durable products. Very little attempt has been made to study how remanufacturing maybe applied to products with short life cycle. In some developing countries like Indonesia, there is a large segment of society that could become potential market for remanufactured short-life cycle products like mobile phones, computers and digital cameras.

In remanufacturing practice, there are three main activities, namely product return management, remanufacturing operations, and market development for remanufactured product [10]. In terms of marketing strategy, there are general concerns that re-manufactured product would cannibalize the sales of new product. However, Atasu et al. [11] concluded that remanufacturing does not always cannibalize the sales of new products. He proposed that managers who understand the composition of their markets, and use the proper pricing strategy should be able to create additional profit. Therefore, pricing decision is an important task in an effort to gain economic benefit from remanufacturing practices.

There are several studies that focused on pricing of remanufac-tured products, but many of them have not considered the whole supply chain, and also only a very few concern about obsolescence of short life cycle products. Our study will be focused on pricing decisions in a closed loop supply chain involving manufacturer, retailer and collector of used products (cores), where customers have the option to purchase new or remanufactured short life cycle products in the same market channel. We consider a monopolist of a single item with no constraint on the quantity of remanufac-turable cores throughout the selling horizon.

2. Literature review

Remanufacturing of mobile phones and electronic products has been recognized as an important practice in the United States, and as a potential in China and India. Helo [12] claimed that product life cycle has significantly shortened by rapid technological advancement, and coupled with fashionable design that attracts frequent purchases of new products, has generated pressure on and opportunities for reverse logistics. Franke et al. [13] suggested that remanufacturing of durable high-value products such as automobile engine, aircraft equipment, and machine tools, has been extended to a large number of consumer goods with short life cycle and relatively low values, like mobile phones and computers. He also quoted market studies by Marcussen [14] and Directive 2002/96/EC which revealed that there is a significant potential for mobile phone remanufacturing due to the large supply market of the used mobile phones in Europe and the high market demand in Asia and Latin America.

Neto and Bloemhof-Ruwaard [15] found that remanufacturing significantly reduces the amount of energy used in the product life cycle, even though the effectiveness of remanufacturing is very sensitive to the life span of the second life of the product. They also proposed that the period of the life cycle in which the product is returned to recovery, the quality of the product (high-end versus low-end), the easiness to remanufacture and the recovery costs

can affect whether or not remanufacturing is more eco-efficient than manufacturing. Rathore et al. [16] studied the case of reman-ufacturing mobile handsets in India. They found that used phone market is very important, even though with a lack of government regulation for e-wastes. It is also observed that there is a negative user-perception of second hand goods and that the process of re-manufacturing has not been able to capture much required attention from its stakeholders. Wang et al. [17] showed that the mobile phone market in China is growing rapidly. The number of mobile accounts was 565.22 million in February 2008 according to a report from Ministry of Information Industry of the People's Republic of China. The above mentioned studies have affirmed our intuitive proposition that there is a high potential for remanufacturing short life cycle products.

Motives for deploying reverse chain can be for profitability (or cost minimization) or for sustainability (environmental impact mitigation), which either could be driven by regulation and/ or morale. In our research, the underlying motive considered would be focused on profitability, which seems to be the suitable motive applied to industries in a situation with the absence of environment protection regulation, like in most of the developing countries. There are numerous studies that investigated the factors that influence decision to remanufacture as well as the factors for successful remanufacturing. We categorized the factors into four aspects, namely product characteristics, demand-related factors, process-related factors, and supply-related factors.

The first aspect, product characteristics of short life cycle products, consist of (1) innovation rate (fast vs. slow) as an extension to technology factor [8,18-20]; (2) residence time [21]; (3) product residual value [22]; (4) qualitative obsolesce, as an extension to product characteristics [3,23].

Second, demand-related factors, consist of (1) market size or existence of the demand, [18,19,24]; (2) market channel, which is about selling remanufactured products using the same channel as the new product, or differentiated [8,18-21,25-27]; (3) pricing of new and remanufactured products, with demand as a function of price [28,32,33,30,18,19,26]; (4) existence of green segment, [23,31].

Supply-related factors can be described by (1) acquisition price and (2) source of return, whether it is limited and then pose as a constraint, or unlimited. These factors were studied in [8,18,19,23, 24,32].

The last factors, which are process-related, consist of (1) remanufacturing technology availability [8,18,32]; (2) remanufacturing cost, [8,18,19,21,22,32]; (3) reverse flow structure readiness [8,20, 24,32-34].

There are several studies that discuss pricing strategies involving remanufactured products, obsolescence, and nonlinear demand function. However, none has considered the situation that we address in this paper. Table 1 shows the review result and where our proposed model stands.

3. Problem description

A closed-loop supply chain consists of three members, which are manufacturer, retailer, and collector, as depicted in Fig. 1. The closed-loop is initiated by production of new product, which is sold at a wholesale price Pnw to the retailer. The new product is then released to the market at a retail price Pn1 for the period when product life-cycle is within introduction-growth-maturity (IMG) phases, or during increasing and stable phases. When the new product has reached its decline phase, retailer starts to apply different pricing, Pn2. In the model development, the price is differentiated between IMG phases and decline phase to study the impact of this differentiated pricing, as Kotler & Armstrong [52] suggest that reduced price during decline phase could increase the quantity of demanded goods. An example of short life-cycle product where the new product reaches its decline phase in a short time is, Samsung Galaxy Tab 10.1 that was released on second quarter

Table 1

Literatures on pricing models.

Supply Chain members involved Differentiating New & Reman Planning Horizon Demand function Decision variables Objective Considering obsolescence Remark

Guide et al. (2003) [35] Remanufacturer Only reman product Single period Dr known Price of *Reman *Core Max profit No Consider several quality classes of cores

Bakal and Akcali (2006) [36] Remanu-facturer Only reman product Single period Linear in price Price of *Reman *Core Max profit No Consider effect of recovery yield

Ferrer and Swaminathan (2006)[37] Manufacturer No (Pn = Pr) •Infinite •Two period •Multi period Linear in price *Price *Quantity Max profit No Consider monopoly & duopoly

Vadde et al. (2006) [38] Product recovery facility Only reman product Selling horizon Function of price and obsolescence *Price Max profit Yes Consider 2 types of obsolescence *Gradual *Sudden

Mitra (2007) [39] Retailer Reman & refurbish products Selling horizon Two cases: •Linear in price •Non-linear Price of *Reman *Refurbish Max revenue No Consider the availability of product

Atasu et al. (2008) [31] Manufacturer Yes (Pn = Pr) Two period Linear in price *Price *Quantity Max profit No Consider green segment, market diffusion, competition with other OEM

Qiaolun et al. (2008) [29] 'Manufacturer •Retailer 'Collector No (Pn = Pr) Selling horizon Linear in price Price of *Retail *Wholesale *Collecting Max profit No Manufacturer is the Stackelberg leader

Li et al. (2009) [40] Remanufacturer Only reman product Single period Stochastic, function of price Price of *Reman *Core Max utilization No Consider random yield and random demand

Liang et al. (2009) [41] Remanufacturer Only reman product Single period None Price of core High return on investment No Consider selling price follows GMB, and core price follows option principles

Ferrer & Swaminathan (2010)[42] Manufacturer Yes (Pn = Pr) •Infinite •Two period •Multi period Linear in price *Price *Quantity Max profit No Consider monopoly & duopoly

Ovchinnikov (2011)[28] Manufacturer Yes (Pn = Pr) Pn fixed Selling horizon Dn known & constant Dr function of price *Price *Quantity of reman Max profit No Also study customers' switching behavior a (Pr) s [0, 1]

Shi et al. (2011) [43] Manufacturer No (Pn = Pr) Single period Stochastic, linear in price *Price *Quantity of new & reman Max profit No Consider understocking & overstocking risks

Vadde et al. (2011) [44] Product recovery facility No new products Single period Deterministic Prices Max revenue Min cost No Consider several types of used products

Wei & Zhao (2011) [45] 'Manufacturer •Retailer No (Pn = Pr) Single period Linear in price Price of *Retail *Wholesale *Collecting Max profit No Consider two competing retailers

Pokharel & Liang (2012)[46] Consolidation center Only cores Single period Dr is known *Core price *Quantity of cores Min cost No Consider stochastic return quantity and quality

Wu (2012a) [47] •OEM •Remanufacturer Yes (Pn = Pr) Two period Linear in price Prices *New *Reman Max profit No Consider level of interchangeability

(continued on next page)

Table 1 (continued)

Supply Chain members involved Differentiating New & Reman Planning Horizon Demand function Decision variables Objective Considering obsolescence Remark

Wu (2012b) [48] *OEM *Remanufacturer Yes (Pn = Pr) *Two period *Multi period Linear in price Prices *New *Reman Max profit No Consider degree of disassemblability

Chen & Chang (2013)[49] Manufacturer Yes (Pn = Pr) •Static *2-period *Multi periods over life-cycle •Linear in price, with substi- tutable coefficient •Dynamic (over time) Price of *New *Reman for each period Max profit No •Static unconstrained •Dynamic pricing - constrained •Consider system of manufacturing only & hybrid settings

Jena & Sarmah (2013)[50] *Remanufacturer •Retailer Only reman product Single period Random Cores price Max profit No Consider 3 schemes of collection: direct, indirect, coordinated

Xionget al. (2013) [51] Manufacturer Only reman product Finite & infinite horizon Random Cores price Min cost No Consider lost sales and uncertain quality of used products

Proposed model *Manufacturer *Retailer *Collector Yes (Pn=Pr) Selling horizon Function of time and price Price of *Retail *Wholesale *Collecting Max profit Yes Consider the impact of speed of change in demand

Note: Pn = price of new product, Pr = price of remanufactured product, Dn = demand of new product, Dr = demand of remanufactured product.

Fig. 1. Framework of the closed-loop pricing model.

2011. According to trial documents of Apple Inc. vs. Samsung Electronics Co., Ltd [53], the sales of that product is entering decline phase starting on first quarter 2012.

After a certain period of time, some products reach their end-of-use and become the objects of used products collection. The used product would be acquired by collector under certain acquisition prices, Pc1 and Pc2, for product originated from IMG phases and decline phase, respectively. The collected product is then transferred to manufacturer at price Pf, as the input for remanufacturing process. The remanufactured product is sold to retailer at wholesale price Prw and released to the market at retail price Pr.

The product considered in this model is single item, short life-cycle, with obsolescence effect after a certain period. Demand functions are time-dependent functions which represent the short life-cycle pattern along the entire phases of product life-cycle, both for new and remanufactured products; and linear in price.

There are four periods considered in this model, as depicted in Fig. 2. In the first period [0, t1], only new product is offered to the market, while in second [t1, /] and third period [/, t3] both new and remanufactured products are offered. The difference between second and third period is on the segments of life-cycle phases for both types. During second period, both new and remanufac-tured products are at the IMG phases. In the third period, the new product has entered the decline phase while remanufactured product has not. In the fourth period [t3, T], manufacturer has stopped producing new product and only offers remanufactured product which is assumed to be on the decline phase.

The market demand capacity is adopted from Wang and Tung [54] and extended to cover the obsolescence period, where

ti № t3 T

Fig. 2. Demand pattern of a product with gradual obsolescence, over time.

demand decreases significantly. The demand patterns are constructed for both new and remanufactured product and the governing functions are formulated as follows:

Dn(t ) =

\Dni(t) = U/ (1 + ke-kUt) ; Dn2(t) = U/ (kU(t - m) + 5) ;

0 < t < m M < t < t3

k = U /D0 - 1

5 = 1 + ke-kUM

D (t) = ¡Dr1(t) = V/ (1 + he-nV(t-t1)) ; t1 < t < t3

r() \Dr2(t) = V/(nV(t - t3) + e) ; t3 < t < T

h = V/Dro - 1 e = 1 + he-nV (t3-t1)

where Dn(t) and Dr (t) are demand pattern for new and remanufactured products, respectively, as seen in Fig. 2. U is a parameter representing the maximum possible demand for new product, / is the time when the demand reaches its peak, i.e. at U level. D0 is the demand at the beginning of the life-cycle (when t = 0), and X is the speed of change in the demand as a function of time. A parallel definition is applicable for V, t3, Dr0, and n respectively for the remanufactured products. It is obvious that Dn (t) and Dr (t) are continuous at / and t3, respectively, as shown in Appendix A.

The new products are sold at retail price Pn1 during [0, / ], and Pn2 during [/, t3]. Since demand function is also linear in price, there is a maximum price Pm (known and fixed) as the upper limit,

Fig. 3. Demand of new product during [0, m]; (a) time-view (b) price-view.

at which demand would be zero. Remanufactured products are sold at retail price Pr during [t1; T], and the maximum price is Pn2, since customer would choose to buy new product rather than remanufactured one when the remanufactured product price is as high as Pn2. Therefore,

Demand of new product during

[0,M] = Dn1(t)(1 - Pm/Pm) (3.3)

Demand of new product during

[M, t3]= Dn2(t)(1 - Pn2/Pm) (3.4)

Demand of reman product during

[t1, t3]= Dr 1(t)(1 - Pr/Pn2) (3.5)

Demand of reman product during

[t3, T] = Dr2(t)(1 - Pr/Pn2). (3.6)

Fig. 3 illustrates the demand of new product for the period of [0, m].

The demand function information is shared to all members of the supply chain.

Manufacturer decides the wholesale prices for new product (Pnw) and remanufactured product (Prw), retailer determines the retail prices (Pn1, Pn2, Pr), while collector determines collecting prices Pc1 and Pc 2 for cores collected from end-of-use product within the periods of [0, m] and [m, t3] respectively. Since the product has short life-cycle, remanufacturing process is only applied to cores originated from new products.

Return rate (t) is an increasing function of the collecting price. We use the return rate function proposed by Qiaolun et al. [29], which was a result of their survey that employs a power function. The return rate t is defined on [t1, T] and depends on Pc as follows

t = y p°c

where y are positive constant coefficients, and 0 s [0, 1] are exponents of the power functions, which determine curve's steepness. It is assumed that collector only accepts cores with a certain quality grade, and all collected cores will be remanufactured.

Since our research is focusing on pricing decision, we do not make an attempt to show detailed derivation of production and operational costs, and instead treat those costs as given parameters, which consist of unit raw material cost for new product (crw), unit manufacturing cost (cm), unit remanufacturing cost (cr), and unit collecting cost (c). The objective of the proposed model is to find the optimal prices that maximize profit, and we investigate two scenarios, (1) maximize profit independently, and (2) maximize joint profit along the supply chain.

4. Optimization

4.1. Independently optimized profit

4.1.1. Retailer's optimization

In this scenario, manufacturer makes the first move by releasing initial wholesale prices (Pnw, Prw). Retailer then optimizes the

retail prices Pn1, Pn2, and Pr. The profit function can be formulated as follows:

U ( Pn1 ,

__ , _ |(pn, - pn,)

+ I -I 1 - — I (Pn2 - Pnw) dt

+ ' kU(t - t) + 5 \ Pm/ v n2 nw)

"'3 V ( Pr ,

+ h T+hF^ i1 - £,1(P' - •) dt

V / P' , Vt 1 - P- (P' - Prw) dt

t3 JV (t - t3) + £ \ Pn2/

= d^ 1 - P1} (Pnl - Pnw) + djl - Pn2 ) (Pn2 - Pnw)

+ (d3 + dA)[ 1 - ) (P' - P'w) ' n2 /

where 1

d1 = - ln

d3 = — ln

(1 + k) e-kUJ

d2 = - ln

1 (kU(t3 - j) + 5

d4 = - ln

1 fnV(T - t3) + £

n \(1 + h)e-nV (t3-t1) / n

The objective function is to maximize profit (4.1), and consequently it needs to satisfy the first derivative conditions dnR/dPn1 = 0, dnR/dPn2 = 0, and dnR/dPr = 0. The profit function (4.1) is not always concave along the considered interval, because Pn2 took a hyperbolic form as a result of being the upper bound of Pr, so we need to establish the interval on which profit function is concave.

Property 1. The objective function (1) is concave when

3/ (d3 + d4) PmPrl

Proof. See Appendix B.

The above result implies that the demand of remanufactured product (d3 and d4) influences the price of new product during the decline stage. Demand capacity during decline stage which is affected by the length of that period has also contributed in shifting the interval of the concave function.

The optimal retail prices Pn1*, Pn2*, and Pr* are obtained by solving equations from first derivatives conditions:

P*1 = (Pm + Pnw)/2 P* = (Pn2 + P'w)/2

- 2 d2 P2)3 + (*(! + ^ +

d3 + d4

(d3 + d4) (P'w)2

Following Property 1, Eq. (4.2) becomes the lower bound of Pn2. It is expected that P*2 is lower than P^ to increase demand rate at the decline stage, however the model allows P*2 to attain higher value than P*1, which in turns is not attractive for customers. Our investigation showed that Pn2 has a tendency to attain higher value than Pn1, which is also consistent with Ferrer and Swaminathan's finding [37]. However, this is not an expected result since higher price during decline stage is not attractive for customers and might not be able to improve the demand rate. Therefore, we set a common price for new product, Pn1 = Pn2. Retailer's optimization model becomes

Max n = (d1 + d2)( 1 - (Pn - Pnw)

Pn,Pr \ ImJ

+ (d3 + d4)[ 1 - Pf ) (Pr - Prw)

Decision variables:

Pn: price of new product

Pr : price of remanufactured product

Parameters:

Pnw : wholesale price of new product Prw : wholesale price of remanufactured product Pm: maximum price for new product d1 : total demand for new product within [0, m] d2: total demand for new product within [m, t3] d3: total demand for remanufactured product within [t1, t3] d4: total demand for remanufactured product within [t3, T]

The existence of optimal prices Pn, Pr is shown in Property 2, and the condition for obtaining prices that maximize retailer's profit (4.6) is given in Property 3. The optimal retail prices are given in Proposition 1.

Property 2. There exists global extrema for profit maximization problem (4.6) in {(Pn, Pr)\Pnw < Pn < Pm; Prw < Pr < Pm; Pn 6 R, Pr 6 R,}.

Proof. See Appendix C.

Property 3. The objective function (4.6) is concave when

— - Pr + M 2 r V d3 + d4 P,

di + d2 P3

Proof. See Appendix D.

Proposition 1. The optimal prices for optimization model (4.6) are P* and P* where

-2(d1 +d2) ^ + ( (pm _ pnw) + , p:

d3 + d4 2

P2 = 0

p* +Pr,

Proof. See Appendix E.

From Proposition 1, we can observe that the optimal prices are not only determined by wholesale prices given by the manufacturer, but also by the demand pattern imposed in d1 , d2, d3, and d4, which confirms the influence of demand patterns to the optimal retail prices.

4.1.2. Collector's optimization

After retailer decides the optimal retail prices, collector then uses the resulting demand rates as the parameters in the profit op-

(4.10)

timization model, and the objective function is Max ne = Y Pc (d1 + d2)( 1 - Pn) (Pf - Pc - c ).

pc \ ' m )

Decision variables:

Pc : acquisition price for used product

Parameters:

Pn, Pm, d1, d2 as mentioned earlier Pf : transfer price of remanufacturable core from collector to remanufacturer Y : constant coefficient of the return rate Q : exponent of the return rate power functions c: unit collecting cost

Property 4 shows the existence of optimal price Pc for collector's profit function (4.10). First derivative condition is applied to obtain the optimal collecting price as shown in Proposition 2.

Property 4. There exists a global extrema for collector's profit function (4.10) in {Pc |0 < Pc < Pf ; Pc g

Proof. See Appendix F.

Proposition 2. The collector's profit function (4.10) attains its maximum in {Pc |0 < Pc < Pf ; Pc g R} and the optimal collecting price is

0 (pf - c

(в +1)

(4.11)

Proof. See Appendix G.

We assume balanced quantity throughout the supply chain, which is supported by Guide's work [26]. Collector should only collect as much as the demand of the remanufactured product, which consequently determines transfer price.

Proposition 3. The optimal transfer price is

Pf* = c +

в + 1 / (d3 + d4) (1 - Pr /Pn) в \y(d1 + d2)(1 - Pn/Pm)

(4.12)

Proof. See Appendix H.

4.1.3. Manufacturer's optimization

After observing retailer's and collector's prices, manufacturer determines the wholesale prices for both new (Pnw) and remanufactured products (Prw) in order to maximize her profit which is expressed in the following objective function:

Max nM = (d1 + d2)( 1 - Pn ) (P*w - cTV1 - cm)

Pnw,Prw \ Pm /

+ (d3 + d4) ( 1 - -

X I Prw cr c

0 (d3 + d4) (1 - Pr /Pn)

0 \y(d1 + d2)(1 - Pn/Pm) r

(4.13)

subject to the optimal prices of the retailer and collector. We apply Lagrange multipliers method, where we define Lagrangian function associated with (4.13) as given in (4.14)

, . -2(d1 + d2) 3 LiPnw, Prw, Pn , Pr = nM + £ --Pi

(4.14)

d1 + d2 d3 + d4 d3 + d4 ^ -V-2 (Pm - Pnw) + IP2 - P2

+ f(2Pr - Pn - Prw) where £ -ф are the multipliers.

Decision variables:

Pnw : wholesale price of new product

P'w : wholesale price of remanufactured product

Parameters:

Pn, P', Pm, d1, d2, d3, d4, Pf, c, y, Q as mentioned earlier

cra : unit raw material cost for producing new product cm: unit manufacturing cost for producing new product c' : unit remanufacturing cost for producing remanufactured product

The first order conditions of the Lagrangian are regarded as the necessary conditions for the constrained optimization, and yield a nonlinear system. We treat Pn and P' as intermediary decision variables in this optimization problem since Pn and P' from retailer's optimum pricing decisions are not expressed as explicit functions in Pnw and P'w, hence the relations are expressed in the constraint functions.

Following Lagrange multiplier theorem [54], if there exist optimal wholesale prices P*w, P*w, P*, P*, then they are the solutions

of first order conditions for (4.14), which are ^ = 0, ^ = 0, inK = 0, = 0, ^ = 0, and dfM = 0, that yields (4.15) to

dPnw dP'w df 'J \ I

(4.20) respectively.

1 ( Pr

— (d1 + d2) (Pnw - c'w - cm) - (d3 + d4) I 1 - —

X (Prw - cr - c) + /

ds + d4

Y (dl + d2) (1 - Pn/Pm)

(0 + 1) (ds + d4)[ 1 - I)

Pn20 2 (1 - Pn /Pm)

x (0 - 1) 1 - - \Pr - (0 - s)

PnPr , 0Pr

X--1--

Pn2 + 2Pr

+ 6P2J

d1 + d2

- 2Pnl

d1 + d2

(Pm Pnw ) +

ds + d4

- f = 0

ds + d4

Pr w cr c

+ (ds + d4) ( 1 - P

0 +1C (ds + d4) (1 - Pr /Pn) 0 Vy (d1 + d2)(1 - Pn/Pm) 0 + 1

(ds + d4) (1 - Pr /Pn)

Y (d1 + d2) (1 - Pn/Pm)

- 2f = 0

Pm - Pn - £ P n = 0

1 - ^ - - f = 0

-2(d1 + d2)_Ps + ( (Pm - Pnw) + P„2

ds + d4 2

—-4 P 2 = 0

(4.15)

(4.16)

(4.17)

(4.18)

(4.19)

the function in a close neighborhood ofthe solution and maximize profit numerically under an optimization search procedure.

4.2. Joint pvofit optimization

Under the joint profit scenario, all parties aim at maximum total profit along the supply chain. The joint profit function is summation of retailer's profit, collector's profit, and manufacturer's profit. Balanced quantity is also imposed in this model, and for remanufactured product, the quantity of demand is equal to the quantity of returns, which also means collector only collects as much as the demand for remanufactured products. The optimization problem is given by the following expressions:

Max nj = nR + nc + nM = (d1 + d2)( 1 - M (Pn - Pnw)

+ (ds + d4) ( 1 - ^ (Pr - Prw)

+ YP0 (d1 + d2) ( 1 - Pn) (Pf - Pc - c)

+ (d1 + d2) ( 1 - — ) (Pnw - crw - cm) + (ds + d4)

(1 - I) (Prw - Pf - cr)

s.t. (ds + d4) 1 " = YP0 (d1 + d2) 1 ^.

(4.21)

(4.22)

The joint profit function is then simplified to a function of Pn, P', and Pc as presented in Eq. (4.22). Considering balanced quantity throughout the supply chain, then the optimization model for joint profit function becomes (4.23)

nj = (d1 + d2) i 1 - p*) (Pn - c'w - cm) + (d3 + d4)

(4.2s)

X ( 1 - f I (Pr - cr - Pc - c)

nj = (d1 + d2)( 1 - y J (Pn - crw - cm) + (ds + d4) I 1 - P I (Pr - cr - c) -

(ds + d4)

x " - k

Y (d1 + d2){ 1 - Pn

(4.24)

In finding the optimal prices, we assign first derivatives to zero

dnj „ , , (Pm - 2P n + crw + cm

- = (d1 + d2)

+ (ds + d4)

(ds + d4 )[ 1 - -

2 P r - Pn - Pr w = 0 .

(4.20)

The second order conditions for the Lagrangian function, which reflects the sufficient condition for a maxima, are not practical to be expressed analytically. Therefore, to ensure that the solution is a maxima we apply a numerical procedure to check the values of

Y (d1 + d2) ( 1 - Pn

0Pn (Pn - Pr) 0 (Pm - Pn)

(4.25)

Pr - cr - c

(di + d2) (1 - Pm) (Pnw - Crw - cm) + (d3 + dA) (l - P£) (Prw - Pf - Cr) - П

nR + Пм + Пс

y Pi (i - f) (Pf - p**- c) - Пс

nR + Пм + Пс

(di + d2)( 1 - Pm) (P* * - Pnw) + (d3 + d4)( 1 - Pw) (p; * - Prw)

П + Пм + Пс

(4.28)

(4.29)

(4.30)

Box I.

^ = " (d3 + d4)

Pr - cr - c -

в + 1

(d3 + d4)( 1 -

_ Y (d1 + d2) ^ - Pt)_

+ (d3 + d4) ( 1 - у ) = 0 or

Pn - 2Pr + cr + c +

(d3 + d4)( 1 -

_ Y (d1 + d2) (1 - Pt)_

(4.26)

The optimal values of Pn and Pr that maximize joint profit are obtained by solving Eqs. (4.24) and (4.25). Since we assume balanced quantity (consistent with Guide's work [26]), the optimization model is reduced to a problem with two decision variables, Pn and Pr. These results only determine optimal retail prices, and leave the wholesale prices (Pnw and Prw) and transfer price (Pf) to be determined under coordinated decision policy, which incorporate equal relative profit difference between independent and joint profit scenario between manufacturer and retailer. The integrated scenario accommodates coordinated pricing policy that ensures higher (or might be lower) profit for each party, and makes this approach interesting for all members of the supply chain. The equal relative profit difference can be expressed as follows:

ЛПм

ЛПс

(4.27) nR nM nc

where AnR = nRJ - nr; AnM = nMj - nM; Anc = ncj - nc.

Therefore, we get Eqs. (4.28)-(4.30) (see Box I) where S = nj - (nR + nM + nc) = AnR + AnM + Anc.

This system of Eqs. (4.28)-(4.30) is solved for Pnw, Prw, and Pf to obtain the wholesale prices and transfer price under joint profit scenario.

5. Numerical example and discussions

In this numerical example, the parameters in demand function are using data from numerical example in [54], because it represents demand pattern of product with gradual obsolescence. However, that study does not consider used product's return, therefore parameters in return function is taken from numerical example in [29]. As for the cost parameters, we developed the data based on case studies in a report for Ellen MacArthur Foundation, Towards The Circular Economy [55].

New product's demand capacity parameters are U = 1000, D0 = 90, X = [0.01, 0.05, 0.1, 0.2], and remanufactured product's demand capacity parameters are V = 500, Dr0 = 50, n = [0.01, 0.05, 0.1, 0.2]. Selling horizon is divided into four time periods where t1 = 1, / = 2, t3 = 3, and T = 4. The unit raw material cost for new product crw = 1500, unit manufacturing

cost cm = 1000, unit remanufacturing cost cr = 800, and unit collecting cost c = 100. Maximum price is Pm = 12 000. Return rate parameters are y = 0.01, and в = 0.7. The decision variables are Pn, Pr, Pnw, Prw, Pc, Pf, which represent price of new product, price of remanufactured product, wholesale price of new product, wholesale price of remanufactured product, collection price and transfer price, respectively. Table 2 presents the results.

From the results above, we have shown that joint profit scenario gives a higher total profit rather than optimized individually. It is also interesting to note that the joint profit model accommodates coordinated pricing policy that ensures higher profit for each party, and making this approach interesting for all members of the supply chain. Demand rate in the joint profit scenario is much higher than in the independent one, even though it came from the same demand parameters. In the independent model, with the lack of integrated decision among the three players, the retail prices were set substantially higher than the true optimums. We also observed that Collector profit is much lower than Retailer's and Manufacturer's, because Collector only gains from remanufactured product. This result is consistent with Qiaolun's [29].

The optimization models also show that transfer price can be found by balancing the return rate with the demand of remanufactured product. Under this approach we can determine transfer price that could benefit both manufacturer and collector, and it puts collector at better position rather than the presumed condition that transfer price is negotiated between manufacturer and collector. Since manufacturer is the Stackelberg leader, it is possible that collector would have been in lower bargaining position. Even though this approach might not be interested for the manufacturer, as it puts limitation to manufacturer's power, but it actually creates sustainability for the overall closed-loop supply chain. The betterment in collector's position would be a good motivation to continue collecting used products for remanufacturing, and support environment protection.

Different speeds of change in the demand of new and reman-ufactured product obviously result in different pricing decisions. However, faster penetration to the market, which is shown by the higher speed of change in demands, does not simply generate higher total profits. It can be seen from the demand function and optimization models that speed of change in demand will influence the sales volumes in each period and subsequently has impacted the optimum pricing decision. As stated in Propositions 1 and 3, the optimal prices depend on the demand volumes in respective periods. We can also find an interval of speed of change in demand where the total profit reaches its highest value. This could lead to a marketing decision where the players should control market penetration such that the speed of change is within the desirable interval.

The impact of demand's speed of change to the optimal prices

In this paper, demand of short life-cycle product is time-dependent and demand pattern over time is influenced by several parameters, such as speed of change in demand as shown in (3.1) and (3.2). The speed of n change in the demands is determined by parameter X and n, for new and remanufactured product, respectively. The higher X and n, the faster demand increases

Table 2

Comparison between independent and joint profit scenarios.

0.01 0.05 0.10 0.20

Independent scenario

Pn 9,889.78 9,896.80 9,896.98 9,896.86

Pr 8,318.83 8,346.32 8,347.02 8,346.56

Pnw 7,018.45 6,997.86 6,997.33 6,997.68

Prw 6,747.80 6,795.85 6,797.06 6,796.26

Pc 422.04 468.28 469.50 468.70

Pf 1,124.96 1,237.25 1,240.21 1,238.27

nm 2,391,233.07 2,453,199.12 2,443,651.42 2,434,076.99

nR 1,246,142.45 1,279,206.64 1,274,245.91 1,269,241.66

nc 145,869.03 176,221.10 176,253.30 175,094.89

Total profit 3,783,244.55 3,908,626.86 3,894,150.63 3,878,413.54

Joint profit scenario

Pn 7,816.53 7,837.87 7,838.40 7,838.06

Pr 4,720.08 4,758.91 4,759.89 4,759.25

Pnw 5,756.12 6,110.48 6,115.01 6,579.60

Prw 4,133.21 3,566.64 3,560.95 2,742.55

Pc 297.96 321.15 321.74 321.36

Pf 913.79 1,004.01 1,006.38 1,004.83

nmj 3,178,657.26 3,248,364.50 3,235,368.72 3,222,922.56

Hrj 1,656,492.54 1,693,841.07 1,687,088.16 1,680,582.67

ncj 193,903.16 233,340.36 233,357.51 231,840.36

Total profit 5,029,052.96 5,175,545.93 5,155,814.39 5,135,345.59

Table 3

Optimal prices under various speeds of change in demand.

X, n Pn Pr Pnw Prw Hm nR nc

0.01 9890 8319 7018 6748 1,246,142.45 2,391,233.07 145,869.03

0.05 9897 8346 6998 6796 1,279,206.64 2,453,199.12 176,221.10

0.1 9897 8347 6997 6797 1,274,245.91 2,443,651.42 176,253.30

0.2 9897 8347 6998 6796 1,269,241.66 2,434,076.99 175,094.89

0.3 9897 8346 6998 6796 1,266,865.91 2,429,539.87 174,376.30

during IMG phases, and decreases during decline phase. Therefore, the accumulation of demand over time would influence the optimal prices, since we consider the whole product life cycle. We vary X and n to study their impacts on decision variables and the objective functions. Table 3 and Fig. 4 show the results.

As shown in Fig. 4, optimal prices are sensitive to parameters that reflect the speed of change in demand for lower market penetration, but they are robust for higher speed of change in demand. This can be explained by the specific nature of demand pattern. For a short life cycle product with demand pattern as given in (3.1) and (3.2), higher speed of change in demand leads to a condition that is closer to constant demand. Therefore, optimal prices do not change significantly compared to the ones with lower speed of change in demand. This situation is applicable to a product with demand pattern such that in the beginning of product's introduction phase, the increase in demand tends to be very steep, such as launching a new model of smartphone. After a certain time period, the demand decreases quite rapidly. We argue that the optimal prices are more robust in higher speed, because the effect of speed of change becomes less significant. This means, when speed of change in demand is high such that the sales reach maximum demand in a very short time, then the price setting does not need to be adjusted when there is a small change in the speed. However, when the speed is low, the sales climb up rather slowly to reach maximum demand, so it is important to adjust price setting when there is a change in the speed, to avoid sub-optimal prices.

The managerial insight for this matter can be explained as follows. Under lower speed of change in demand, decision makers need to carefully update the prices to avoid sub-optimality. But when it moves fast to the peak, price updates become less urgent, since the demand has reached its uniform pattern.

6. Conclusion and future research agenda

In this study we have developed pricing decision models for remanufacturing of short life-cycle product. The study fills the gap of remanufacturing literature which to date has been mostly dominated by durable products. For some short life cycle products, remanufacturing is a sensible activity to do, but the speed of collecting and remanufacturing the used products should be quick as the demand for the product is diminishing fast. Here are some conclusions that we obtain from this study:

• Reducing the price of new products during the decline phase does not give better profit for the whole system,

• The total profit obtained from optimizing each player independently is lower than the total profit of the integrated model where we optimize the joint profit for three members in the supply chain, namely manufacturer, retailer and collector. None of the player is worse off by moving from the independent model to the joint profit model, under coordinated pricing policy,

• The total demand is significantly higher under the integrated model. This is understandable because the retail prices are lower for both the new and remanufactured products. The lack of coordination in making the pricing decision has led the independent models to set high retail prices and hence the demand potential is not well exploited,

• Faster penetration to the market, which is shown by the higher speed of change in demands, does not simply generate higher total profits. There exists an interval of speed of change in demand where the total profit reaches its highest value. This could be a support for marketing decision by controlling market penetration such that the speed of change is within the desirable interval,

Fig. 4. Optimal prices charts under various speeds of change in demand.

• When demand penetration is low, small changes in the demand rate affect price settings substantially. However, when demand penetration is high, price decision is robust against the change in demand rate.

Future research may be directed toward development of models that consider different demand processes, multiple objective functions, and the case when balanced quantity is not the case. It may be possible that the collector is not able to collect at the quantity desired by the manufacturer. It is also possible that the manufacturer has a certain capacity constraint where not all demand can be satisfied. In such as case it is important to take into account the service level.

Appendix A. Dn(t) and Dr (t) are continuous at p and t3, respectively

Similarly,

Dn(t3) = Dr2(t3) =

1 + he-"V (t3-t{) V

(nV(t3 - t3) + e) e 1 + he-nVto-^

X Dn(t3) = Dr2(t3)

lim Dr (t) = lim

t^t- 1 + he-nV (t-t1) 1 + he-nV (t3-t1i)

3 3 V V

lim Dr (t) = lim -=---

() (nV(t - t3) + e) 1 + he-nV(t3-t1)

x lim Dr(t) = lim Dr(t).

Therefore Dr(/) = limt^t3 Dr (t3) = 1+he-nVV(t3-t1)

continuous at t = t3. Appendix B. Proof of Property 1

^ Dr (t) is

Dn1 (p) = Dn2(p) =

1 + ke-XUp U

(XU(p - p) + S) s

X Dn1 (p) = Dn2(p)

lim Dn(t) = lim -— =

t^p- n() t^p- 1 + ke-XUt

1 + ke-XUp

1 + ke-XUp

lim Dn(t) = lim -=-—

(XU(t - p) + S) 1 + ke-XUp

x lim Dn(t) = lim Dn(t).

Therefore D*(p) = limt^p D*(p) = 1+ku->.up ^ Dn(t) is continuous at t = p.

n = d^1 - P^J (Pn1 - Pnw) + d^1 - p*2 ) (Pn2 - Pnw) + (d3 + d4)( 1 - (Pr - Prw)

0 P n1

0 P n1

d Pn2d Pn1 _ 2d1 = ~Pm

d Pn1d Pn2

d2 2 (d3 + d4) Pr (Pr - Prw) P + P 2

Pm Pn2

> 0; since Pr > Prw

|H | =

92nR 92nR 92nR

3Pn2i 9Pni 3Pn2 9Pni 9Pr

92nR 92nR 92nR

3Pn2 9Pni dPn22 9Pn2 9 Pr

92nR 92nR 92nR

dPr dPni 9Pr9Pn2 9P2r

dnR _ ds + dA

~dPT = Pn

( —2Pr + Prw + Pn) = 0

2di (ds + d4)

(ds + d4) P,

Since |Hi| = 92nR/dP2nl < 0; and |H2| = ( 92nR/9P2nl)(92Hr/ 9Pn22)-( d2nR/dPn1 dP22)2 > 0, then for nR to be a concave function, |H31 = |H | should be less than or equal to zero. Therefore

(d3 + d4) Pr2w 4d2

(ds + d4) PmP2w

Appendix C. Proof of Property 2

Since nR is a two-variable rational function with a rational parameter in the coefficients, then it is discontinuous when Pm = 0 and Pn = 0. However, Pm is a nonzero parameter, and Pnw < Pn < Pm with positive Pnw, therefore nR is continuous in {(Pn, Pr) |Pnw < Pn < Pm; Prw < Pr < Pm; Pn G R, Pr G R,}. Since nR is continuous in a closed interval then it attains global extrema there. □

Appendix D. Proof of Property 3

dnR di + d2

d3 + d4 , 2 (Pm + Pnw 2Pn) + ^ (Pr Prw Pr

d2n di + d2

9n^ _ ds + d4

~dP7 = _pT"

ds + d4 - 2^p+rAPr (Pr - Prw) < 0

(-2Pr + Prw + Pn)

d2n ds + d4

ds + d4

Pn Pr Pr Pn 32n 92nR

(2Pr - Prw)

Pn2 Pr2

4 (di + d2)(ds + dA) (ds + d4)2

(2Pr - Prw)2 > 0

4 (di + d2) (ds + d4)

2Pr-Prw < J 4

(2Pr - Prw)2 > 0

di + d2 PI ds + d4 ' Pm

di + d2-Pi > 0. □

- Pr +,,

2 r ds + d4 Pm

Appendix E. Proof of Proposition 1

The critical points for (4.6) are obtained by applying first derivatives condition,

P* ~h Prw

-2Pr + Prw + Pn = 0 ^ P* = n rw

dnR di + d2 ds + d4 , , >

= (Pm + Pnw - 2Pn) + ^ Pr2 - PrwPr) = 0

Pn Pm Pn2

-2(di + d2) s di + d2 2

_v 1 '_p j____I (p _p ) p *2

p 'n + p ( m ' nw,M n

+ (ds + d4) -2(di + d2)

P * 4- P

n + rw

P * 4- P

- Prw = 0

di + d2 ds + d4 Pf + i "V"2 (Pm - Pnw) + - I P*

ds + d4 2

From Property 4, the retailer profit function is concave when condition (4.7) is satisfied, therefore

— - P 4 2 P +

di + d2 ps > 0

ds —d4 Pm

Pn di —d2

— < 4-.

Pm ds —d4

Pn + Prw + /di + d2■ Pi > 0

ds —d4 Pm

Since p1 < i while 2 > i then (A6.i) is satisfied. □

Pm — d3+d4 — '

Appendix F. Proof of Property 4

The collector's profit function (4.i0) can be written as nC = 0 (Pc) ■ W (Pc) where

0 (Pc) = P0, 0 G [0, i]

W (Pc) = y (di + d2) (i - Pn) (P/ - Pc - c)

0 (Pc) is continuous on the closed intervalPc |0 < Pc < P/; Pc g R} except for Pc = 0 when 0 = 0. However, this is a removable discontinuity by letting P0 = i at the point of discontinuity, since limx^0 xx = i and P0 = i for Pc = 0,0 = 0.

W (Pc) is continuous on the closed interval {Pc|0 < Pc < P/; Pc G R} since it is a linear function in Pc and Pm = 0.

Therefore nC is continuous in a closed interval and attains its global extrema there. □

Appendix G. Proof of Proposition 2

The critical point for (4.i0) is obtained by applying first derivative condition,

dnC i pn \ 0 .

+n 1- ¡m )p0 -

x [0 (Pf - c) - (0 + i) Pc] = 0 0 (P/ - c) - (0 + i) Pc = 0 0 (P/ - c)

(0 + i)

-pr = Y (di + i - pi) (0 - i) pc0-2

x 0 P/ - c - (0 —i) Pc

- (0 + 1)y (di + d2) ( i - y ) P0

Since 0 (Pf - c) - (9 + 1) Pc = 0 at the critical point, then

= - (0 + 1)y(d1 + d2) ( 1 - P0-1 < 0.

Therefore P* is a local maximum point. Following Property 5, we check the boundary points and compare them with the local maximum. Since nC (0) = 0, nC (Pf ) < 0, and nC (P*) > 0 then P* is the global maximum within interval {Pc|0 < Pc < Pf ; Pc g R}. □

Appendix H. Proof of Proposition 3

Balanced quantity for demand of remanufactured product and acquired used product is satisfied when

(d3 + d4) 1 --) = YP0 (d1 + d2) 1 - ^ .

Since optimal collecting price is given as (4.11), then the respective transfer price is

(d3 + d4) I 1 - ^ I = Y

0 (Pf - c )

(d1 + d2) ( 1 - Pn

P * = 9+1 ( (d3 + d4) (1 - Pr /Pn) y D f c 0 \Y(d1 + d2)(1 - Pn/Pm)J .

References

[1 [2 [3

[9 [10 [11 [12 [13

[15 [16 [17 [18

Lebreton B, Tuma A. A quantitative approach to assessing the profitability of car and truck tire remanufacturing. Int J Prod Econ 2006;104(2):639-52. Hsueh C. An inventory control model with consideration of remanufacturing and product life cycle. Int J Prod Econ 2011;133(2):645-52. de Brito MP, Dekker R. A framework for reverse logistics. In: Dekker LN, Fleischmann R, Inderfurth M, Van Wassenhove K, editors. Reverse Logistics: Quantitative Models for Closed-Loop Supply Chains. Heidelberg: SpringerVerlag; 2004. p. 3-27.

Herold ME. Managing end-of-use products, Helsinki University of Technology, 2004.

Lee HB, Cho NW, Hong YS. A hierarchical end-of-life decision model for determining the economic levels of remanufacturing and disassembly under environmental regulations. J Clean Prod 2010;18(13):1276-83. Future Global Manufacturing. Are there Environmental Limits and Solutions? Kaebernick H, Manmek S, Anityasari M. The International Manufacturing Leaders Forum (IMLF), Taiwan. 2006. p. 1-11.

Kerr W, Ryan C. Eco-efficiency gains from remanufacturing—A case study of photocopier remanufacturing at Fuji Xerox Australia. J Clean Prod 2001;9(1): 75-81.

Lund RT, HauserWM. Remanufacturing—An American Perspective. In: 5th International Conference on Responsive Manufacturing—Green Manufacturing (ICRM 2010). 2010. p. 1-6.

Gray C, Charter M. Remanufacturing and Product Design. Int J Prod Dev 2008; 6(3-4):375-92.

Guide VDRJ, Van Wassenhove LN. The evolution of closed-loop supply chain research. Oper Res 2009;57(1):10-8.

Atasu A, Guide VDRJ, Van Wassenhove LN. So what if remanufacturing cannibalizes my new product sales? Calif Manage Rev 2010;52(2):56-76. Helo P. Managing agility and productivity in the electronics industry. Ind Manag Data Syst 2004;104(7):567-77.

Franke C, Basdere B, Ciupek M, Seliger S. Remanufacturing of mobile phones— capacity, program and facility adaptation planning. Omega 2006;34(6): 562-70.

Marcussen CH. Mobile Phones, WAP and the Internet—The European Market and Usage Rates in a Global Perspective 2000-2003, In: Centre for Regional and Tourism Research, 2003. [Online]. Available: http://www.crt.dk/uk/staff/ chm/wap.htm.

NetoJQF, Bloemhof-Ruwaard JM. An analysis of the eco-efficiency of personal computers and mobile phones. Prod Oper Manag 2012;21(1):101-14. Rathore P, Kota S, Chakrabarti A. Sustainability through remanufacturing in India: a case study on mobile handsets. J Clean Prod 2011;19(15):1709-22. Wang J, Zhao J, Wang X. Optimum policy in hybrid manufactur-ing/remanufacturing system. Comput Ind Eng 2011;60(3):411-9. Ferguson ME, Souza GC. Closed-loop supply Chains—New Developments to Improve the Sustainability of Business Practices. CRC Press; 2010.

Souza GC. Remanufacturing in closed-loop supply chains. Prod Invent Manag J 2009;45(1):56-66.

Lebreton B. Strategic Closed-Loop SupplyChain Management. Lecture Notes in Economics and Mathematical Systems, vol. 586. Berlin Heidelberg: Springer; 2007. p. 1-158.

Debo LG, Toktay LB, Van Wassenhove LN. Market segmentation and product technology selection for remanufacturable products. Manage. Sci. 2005;51(8): 1193-205.

Gobbi C. The Reverse Supply Chain: Configuration, Integration and Profitability, Technical University of Denmark, 2008.

Subramanian R, Ferguson ME, Toktay LB. Remanufacturing and the component commonality decision. Prod Oper Manag 2011;22(1):36-53. Vasudevan H, Kalamkar V, Terkar R. Remanufacturing for sustainable development: Key challenges, elements, and benefits. Int J Innov Manag Technol 2012;3(1):84-9.

Thierry M, Salomon M, van Nunen JAEE, Van Wassenhove LN. Strategic issues in product recovery management. Calif Manage Rev 1995;37(2):114-35. Guide VDRJ,TeunterRH, Van Wassenhove LN. Matching demand and supply to maximize profits from remanufacturing. Manuf Serv Oper Manag 2003;5(4): 303-16.

Geyer R, Jackson T. Supply loops and their constraints: The industrial ecology of recycling and reuse. Calif Manage Rev 2004;46(2):55-73. Ovchinnikov A. Revenue and cost management for remanufactured products. Prod Oper Manag 2011;20(6):824-40.

Qiaolun G, Jianhua J, Tiegang G. Pricing management for a closed-loop supply chain. J Revenue Pricing Manag 2008;7(1):45-60.

Ferguson ME, Toktay LB. The effect of competition on recovery strategies. Prod Oper Manag 2006;15(3):351-68.

Atasu A, Sarvary M, Van Wassenhove LN. Remanufacturing as a marketing strategy. Manage. Sci. 2008;54(10):1731-46.

Ayres RU, Ferrer G, Van Leynseele T. Eco-efficiency, asset recovery and remanufacturing. Eur Manag J 1997;15(5):557-74.

Guide VDRJ, Van Wassenhove LN. Managing product returns for remanufac-turing. Prod Oper Manag 2001;10(2):142-55.

Zanoni S, Ferretti I, Tang O. Cost performance and bullwhip effect in a hybrid manufacturing and remanufacturing system with different control policies. Int J Prod Res 2006;44(18-19):3847-62.

Guide VDRJ, Jayaraman V, Linton JD. Building contingency planning for closed-loop supply chains with product recovery. J Oper Manag 2003;21(3):259-79. Bakal IS, Akcali E. Effects of random yield in remanufacturing with price-sensitive supply and demand. Prod Oper Manag 2006;15(3):407-20. Ferrer G, Swaminathan JM. Managing New and Remanufactured Products. Manage. Sci. 2006;52(1):15-26.

Vadde S, Kamarthi SV, Gupta SM. Pricing of end-of-life items with obsolescence. In: Proceedings of the 2006 IEEE International Symposium on Electronics and the Environment. 2006. p. 156-60.

Mitra S. Revenue management for remanufactured products. Omega 2007; 35(5):553-62.

Li X, Li Y, Cai X. Collection pricing decision in a remanufacturing system considering random yield and random demand. Syst Eng - Theory Pract 2009; 29(8):19-27.

Liang Y, Pokharel S, Lim GH. Pricing used products for remanufacturing. European J. Oper. Res. 2009;193(2):390-5.

Ferrer G, Swaminathan JM. Managing new and differentiated remanufactured products. European J. Oper. Res. 2010;203(2):370-9.

Shi J, Zhang G, Sha J. Optimal production planning for a multi-product closed loop system with uncertain demand and return. Comput Oper Res 2011;38(3): 641-50.

Vadde S, Zeid A, Kamarthi SV. Pricing decisions in a multi-criteria setting for

product recovery facilities. Omega 2011;39(2):186-93.

Wei J, Zhao J. Pricing decisions with retail competition in a fuzzy closed-loop

supply chain. Expert Syst Appl 2011;38(9):11209-16.

Pokharel S, Liang Y. A model to evaluate acquisition price and quantity of used

products for remanufacturing. Intern J Prod Econ 2012;138(1):170-6.

Wu C-H. Product-design and pricing strategies with remanufacturing.

European J. Oper. Res. 2012;222(2):204-15.

Wu C-H. Price and service competition between new and remanufactured products in a two-echelon supply chain. Int J Prod Econ 2012;140(1):496-507. Chen J-M, Chang C. Dynamic pricing for new and remanufactured products in a closed-loop supply chain. Intern J Prod Econ 2013;146(1):153-60. Jena SK, Sarmah SP. Optimal acquisition price management in a remanufac-turing system. Int J Sustain Eng 2014;7(2):154-70.

Xiong Y, Li G, Zhou Y, Fernandes K, Harrison R, Xiong Z. Dynamic pricing models for used products in remanufacturing with lost-sales and uncertain quality. Int J Prod Econ 2014;147(Part C):678-88.

Kotler P, Armstrong G. Principles of Marketing. 14th ed. New Jersey: Pearson Education, Inc; 2012. p. 1-613.

Hill S. Trial reveals Samsung and Apple U.S. sales figures, www.androidauthority.com, 2012. [Online]. Available:

http://www.androidauthority.com/samsung-apple-us-sales-107027/. [Accessed: 08.10.13].

Wang K-H, Tung C-T. Construction of a model towards EOQ and pricing strategy for gradually obsolescent products. Appl. Math. Comput. 2011; 217(16):6926-33.

McKinsey, Towards The Circular Economy, vol 1, 2013.