Trade-In Remanufacturing, Customer Purchasing Behavior, and Government Policy

Home , Customer

This article was downloaded by: [128.252.111.87] On: 03 March 2020, At: 08:09 Publisher: Institute for Operations Research and the Management Sciences (INFORMS) INFORMS is located in Maryland, USA

Manufacturing & Service Operations Management

Publication details, including instructions for authors and subscription information: http://pubsonline.informs.org Trade-in Remanufacturing, Customer Purchasing Behavior, and Government Policy

Fuqiang Zhang, Renyu Zhang

To cite this article: Fuqiang Zhang, Renyu Zhang (2018) Trade-in Remanufacturing, Customer Purchasing Behavior, and Government Policy. Manufacturing & Service Operations Management 20(4):601-616. https://doi.org/10.1287/msom.2017.0696

Full terms and conditions of use: https://pubsonline.informs.org/Publications/Librarians-Portal/PubsOnLine-Terms-and- Conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher approval, unless otherwise noted. For more information, contact [email protected]. The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or support of claims made of that product, publication, or service. Copyright © 2018, INFORMS

Please scroll down for article—it is on subsequent pages

With 12,500 members from nearly 90 countries, INFORMS is the largest international association of operations research (O.R.) and analytics professionals and students. INFORMS provides unique networking and learning opportunities for individual professionals, and organizations of all types and sizes, to better understand and use O.R. and analytics tools and methods to transform strategic visions and achieve better outcomes. For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org MANUFACTURING & SERVICE OPERATIONS MANAGEMENT Vol. 20, No. 4, Fall 2018, pp. 601–616 http://pubsonline.informs.org/journal/msom/ ISSN 1523-4614 (print), ISSN 1526-5498 (online)

Trade-in Remanufacturing, Customer Purchasing Behavior, and Government Policy

a b, Fuqiang Zhang, Renyu Zhang ∗ a Olin Business School, Washington University in St. Louis, St. Louis, Missouri 63130; b New York University Shanghai, Shanghai 200122, China

∗ Corresponding author Contact: [email protected], http://orcid.org/0000-0003-2918-2613 (FZ); [email protected], http://orcid.org/0000-0003-0284-164X (RZ)

Received: May 31, 2016 Abstract. Trade-in remanufacturing is a commonly adopted business practice under Revised: May 27, 2017; October 7, 2017 which firms collect used products for remanufacturing by allowing repeat customers to Accepted: October 12, 2017 trade in used products for upgraded ones at a discount price. This paper studies how Published Online in Articles in Advance: customer purchasing behavior and remanufacturing efficiency affect the economic and May 7, 2018 environmental values of such a business practice. We demonstrate a new benefit of trade- https://doi.org/10.1287/msom.2017.0696 in remanufacturing: it helps exploit the forward-looking behavior of strategic customers, which could be much more significant than the widely recognized revenue-generating Copyright: © 2018 INFORMS and environmental benefits of remanufacturing. High remanufacturing efficiency does not necessarily benefit a firm. With overly high remanufacturing efficiency, product durability is so high that repeat customers are reluctant to trade in and upgrade their used products. When customers are highly strategic, trade-in remanufacturing creates a tension between profitability and sustainability: on one hand, by exploiting the intensive forward-looking customer behavior, trade-in remanufacturing is quite valuable to the firm; on the other hand, with highly strategic customers, trade-in remanufacturing has a substantial negative impact on the environment and social welfare, since it may induce significantly higher production quantities without improving customer surplus. With nearly myopic customers,

however, trade-in remanufacturing benefits both the firm and the environment. There- fore, understanding the interactions between customer purchasing behavior and trade-in remanufacturing is important to both firms and policy makers. Finally, to resolve the above tension, we study how a social planner (e.g., government) should design a public policy to maximize social welfare. The social optimum can be achieved by using a simple linear subsidy/tax scheme for both new production and remanufacturing. The proposed policy can also induce the firm to set the socially optimal remanufacturing efficiency.

Funding: Research of the ﬁrst author is supported by the National Natural Science Foundation of China [Grants 71628203 and 71520107002]. Research of the second author is supported by the Shanghai Pujiang Talent Program [Grant 17PJC074]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2017.0696.

Keywords: customer behavior • trade-in rebates • remanufacturing • environment • government policy

1. Introduction can efficiently disassemble iPhones and sort the high- Remanufacturing is the rebuilding of a product to quality components that can be recycled for manufac- specifications of the original manufactured product turing new devices (Apple Inc. 2017a). using a combination of reused, repaired, and new parts An important issue in remanufacturing is the reman- (Johnson and McCarthy 2014). The initial purpose of ufacturability or remanufacturing efficiency of the remanufacturing was to recover the residual value of product, which depends on product design, dura- the components and materials from used products bility, and other factors (Debo et al. 2006). A prod- (see, e.g., Guide and Van Wassenhove 2009, Debo et al. uct with high remanufacturing efficiency normally 2005). More recently, with growing awareness of sus- requires more durable components and materials tainability, the environmental advantages of reman- (Geyer et al. 2007). Thus, such product would be more ufacturing have also been widely recognized. As a cost effective and environmentally friendly to reman- result, remanufacturing has been increasingly adopted ufacture. At the same time, a product of high reman- in practice to enhance a firm’s competitive edge and ufacturability is also more durable, so customers can public image on the market. For example, to facilitate derive higher residual values from used products. For the recycling and remanufacturing of used products, instance, to ensure the environmental sustainability of Apple recently invented Liam, a line of robots that its business, Apple has adopted high-grade materials

(e.g., aluminum) in its electronic devices so that the trade-in opportunity grants price discounts to repeat devices have longer lifetimes (i.e., are more durable) customers who return their used products, thus and can be easily recycled (i.e., high remanufacturing enabling the firm to price discriminate new and efficiency; see Apple Inc. 2017a). As we will demon- repeat customers (Van Ackere and Reyniers 1995). strate below, the relationship between remanufactur- Under trade-in remanufacturing, strategic customers ing efficiency and product durability will play a sig- will anticipate a potential future price discount in nificant role in how remanufacturing affects firms, the the form of a trade-in rebate, which is ignored by environment, and the society. myopic customers. As a consequence, different cus- According to the United States International Trade tomer purchasing behaviors may lead to drastically dif- Commission (2013), an integral component of closed- ferent market outcomes under trade-in remanufactur- loop supply chains for remanufacturing is core ing. Although both strategic and nonstrategic customer collection, that is, the process of obtaining used prod- behaviors have been widely acknowledged in the liter- ucts from customers. A common practice for core col- ature, it is not clear what role they will play under the lection is to provide trade-in rebates that encourage adoption of trade-in remanufacturing. customers to return their used products. For exam- Recent years have witnessed an increasing number ple, Apple offers both in-store and online trade-in pro- of government interventions in markets based on envi- grams, which encourage customers to exchange their ronmental issues. For instance, starting in 2011, the used iPhones, iPads, and Macs for credits to purchase Chinese Ministry of Finance has maintained a fund new Apple products (Apple Inc. 2017b). Analogously, for the treatment of waste electrical and electronics Amazon allows Kindle owners to trade in their old equipment (WEEE); this fund is used to subsidize the products for newer versions at discount prices (Kellogg recycling and remanufacturing of used electrical and 2011). More examples of using trade-in rebates for core electronic products (see Xie and Bai 2010, Chinese Min- collection have been reported in industries such as fur- istry of Finance 2012). Similarly, the Department for niture, carpets, power tools, etc. (see Ray et al. 2005). Business, Innovation, and Skills in the United King- It is quite common in practice that a customer needs dom established a fund of 775,000 pounds to encour- to decide whether to make an immediate purchase or age the reuse of whole appliances, increase the tonnage to wait for better future opportunities (e.g., a price of separately collected domestic WEEE for recycling, markdown or a new technology). A customer is called and improve the recycling rate of collected equipment strategic or forward looking if she strategizes the pur- (Department for Business, Innovation and Skills 2015). chasing decision to maximize her long-run utilities. In As trade-in remanufacturing gains increasing popular- contrast, a myopic customer does not consider future ity, it is both interesting and relevant to study how the opportunities and bases her purchasing decision on government should design a public policy to regulate the immediate utilities. Customer purchasing behav- the market and enhance social welfare. ior can be quite complex in the real world. It has been The primary goal of this paper is to deepen our empirically verified that customers exhibit a mixture understanding of the trade-in remanufacturing prac- of strategic and nonstrategic purchasing behaviors in tice. Specifically, we analyze the impact of customer various markets (see, e.g., Li et al. 2014, Hendel and purchasing behavior and remanufacturing efficiency Nevo 2013, Chevalier and Goolsbee 2009). So the actual on the value of trade-in remanufacturing to differ- customer purchasing behavior in a market should be ent stakeholders. For this purpose, we develop a two- somewhere between the two extremes of fully strategic period model in which a profit-maximizing firm sells and fully myopic. Despite its complexity, understand- two generations of a product in a market. We use ing the customer purchasing behavior in a market is the customer discount factor to model the intensity important, since whether customers are strategic or not of their forward-looking behavior. If this discount fac- has significant impact on a firm’s operations strategy. tor is large, the customers are highly strategic, and For example, whereas responsive pricing could effec- make their purchasing decisions with serious consid- tively exploit customer segmentation with myopic cus- erations of anticipated future utilities. Otherwise, the tomers, it has a potential adverse impact with strate- customer discount factor is low, so customers care lit- gic customers (Aviv and Pazgal 2008). As another tle about future utilities and are more myopic. In the example, in the durable goods and event ticketing first period, the firm sells the first-generation prod- markets, resale in the secondary market damages the uct in the market. In the second period, the firm sells profit when customers are myopic, but the profit will the second-generation product to new customers (who increase in the presence of the secondary/resale mar- have not purchased in the first period); meanwhile, the ket if customers are forward looking (see Chevalier and firm offers trade-in rebates that allow repeat customers Goolsbee 2009, Su 2010, Cui et al. 2014). (who have purchased in the first period) to exchange Trade-in remanufacturing and customer purchas- used products for new second-generation ones at a dis- ing behavior naturally interact with each other. The count price. The returned first-generation products are Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 603 remanufactured by the firm to reduce the production trade-in remanufacturing on the environment is most costs of new second-generation products. We explicitly significant when remanufacturing efficiency is moder- model two benefits from remanufacturing and recy- ate (i.e., either a low or high remanufacturing efficiency cling used products: First, it generates economic value would help improve the environment). Our results call (i.e., revenue) for the firm. Second, it helps reduce the for caution when adopting the trade-in remanufac- product’s negative impact on the environment. It is turing strategy. In particular, understanding customer worth noting that both benefits depend on the remanu- purchasing behavior and remanufacturing efficiency is facturing efficiency of the product. As discussed above, essential in evaluating this strategy, both for the firm high remanufacturing efficiency normally corresponds and for the environment. to high product durability, that is, high residual value 1.1.2. Government Policy Design. It follows from the of used products for customers (Geyer et al. 2007). above findings that trade-in remanufacturing may cre- 1.1. Main Contributions ate a tension between profitability and sustainability; 1.1.1. Value of Trade-in Remanufacturing. A key mes- that is, it can greatly improve firm profit but mean- sage of our study is that both customer purchasing while hurt the environment, especially under strategic behavior and remanufacturing efficiency have impor- customers. This motivates us to study how government tant implications for the value of trade-in remanufac- intervention can resolve this tension and achieve the turing. Under trade-in remanufacturing, strategic cus- socially optimal outcome for a market where trade-in tomer behavior acts like a double-edged sword. On one remanufacturing is commonly adopted and strategic hand, as recognized in the literature, strategic wait- customer behavior prevails (e.g., the electronics mar- ing of customers keeps them from purchasing early ket). The government is modeled as a central plan- and is thus detrimental to firm profit. On the other ner who aims to maximize the social welfare, that is, hand, trade-in remanufacturing ensures a high sur- the sum of firm profit and customer surplus less envi- plus for repeat customers, which encourages strategic ronmental impact. We consider a linear subsidy/tax customers to purchase early, thus leading to a higher scheme that applies to the production of all prod- profit for the firm when customers are more strate- uct versions. We show that such a simple subsidy/tax gic. We find that the profit improvement from trade- scheme, if designed properly, can induce the social optimum regardless of the customer purchasing behav- in remanufacturing with strategic customers can be much more significant than with myopic customers. ior. Interestingly, the proposed subsidy/tax policy can Thus, our results indicate that a major benefit of trade- also induce the firm to set the socially optimal remanu- in remanufacturing for the firm is to exploit strategic facturing efficiency. In short, our proposed government customer behavior, which has not been identified in policy helps resolve the tension between profitability the previous literature. Moreover, we demonstrate that and sustainability caused by trade-in remanufacturing. high remanufacturing efficiency does not necessarily The rest of this paper is organized as follows. Sec- improve the firm’s profit: When remanufacturing is tion2 positions our work in the relevant literature. The overly efficient, the product durability is so high that model and equilibrium analysis are presented in Sec- repeat customers will be discouraged from trading in tion3. In Sections4 and5, we analyze the value of and upgrading their used products. Therefore, the firm trade-in remanufacturing for the firm and the environ- benefits the most when the remanufacturing efficiency ment, respectively. Section6 characterizes the socially is at a moderate level. optimal government policy. This paper concludes with From the environmental and social perspectives, the Section7. All proofs are given in the online appendix. impact of trade-in remanufacturing also depends critically on customer behavior and remanufacturing effi- 2. Literature Review ciency. We find that while trade-in remanufacturing The impact of customer purchasing behavior on a generally benefits the environment and social wel- firm’s operations decisions has received extensive fare with myopic customers, the opposite is true with attention in the literature. Bensako and Winston (1990) strategic customers. With strategic customers, trade-in show that the presence of strategic customers drives a remanufacturing offers customers a strong incentive monopolist firm to charge a lower price at the initial to purchase early, which prompts the firm to increase stage of the sales season and to mark down less aggres- production quantities. This may outweigh the envi- sively afterward. In a revenue management frame- ronmental advantage of remanufacturing under gen- work, Aviv and Pazgal (2008) demonstrate that the eral circumstances. Moreover, trade-in remanufactur- responsive pricing strategy could effectively improve ing allows the firm to exploit the strategic customers, a monopolist’s revenue with myopic customers, but which may reduce the customer surplus and hence this strategy could lead to significant revenue losses hurt the overall social welfare. Higher remanufacturing with strategic customers. In a follow-up work, Aviv efficiency does not always help improve the environ- et al. (2015) show that the benefits of responsive pric- ment and social welfare. In fact, the negative impact of ing and demand learning depend crucially on the Zhang and Zhang: Trade-in Remanufacturing 604 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS nature of customer purchasing behavior: their values products returned through trade-ins for remanufac- tend to worsen when customers are strategic. Under a turing. Our research reveals a new benefit of trade- newsvendor framework, Cachon and Swinney (2009) in remanufacturing: to exploit the forward-looking show that quick response could deliver a significantly behavior of strategic customers, which is most signifi- higher value to a retailer in the presence of strategic cant when remanufacturing efficiency is moderate. In customers than without them. Caldentey et al. (2017) addition, we deliver a new insight that, depending on consider a robust formulation of a monopolist’s pricing the customer purchasing behavior and remanufactur- problem and characterize the different pricing policies ing efficiency, trade-in remanufacturing may either cre- without knowing the customer valuation and arrival ate a tension between firm profit and environmental timing under different customer purchasing behaviors sustainability or simultaneously benefit both the firm (i.e., fully strategic and fully myopic customers). The and the environment. bottom line of this strand of research is that the effec- Government regulations on remanufacturing and tiveness of an operations strategy is very sensitive to other environmentally relevant operations issues have customer purchasing behavior. In addition to the above also been studied in the literature, but for different modeling works, several papers also empirically exam- problem settings. For instance, Calcott and Walls (2000) ine the customer purchasing behaviors in, for instance, show that in a supply chain, a downstream disposal the airline (Li et al. 2014), soft drink (Hendel and Nevo fee charged by the government may not ensure the 2013), and textbook (Chevalier and Goolsbee 2009) social optimum unless supplemented with upstream industries. The empirical findings suggest that the mar- instruments. Ma et al. (2013) study the impact of a ket is likely to have a mixture of strategic and non- government consumption-subsidy program on a dual- strategic customers, and the overall customer purchas- channel closed-loop supply chain. Cohen et al. (2016) ing behavior is complex. We contribute to this stream characterize the impact of demand uncertainty on gov- of research by investigating how the value of trade- ernment subsidies for green technology adoption. in remanufacturing to the firm and the environment There are a few papers that investigate trade-in depends on different customer purchasing behaviors. rebates or multiple product introductions in the pres- There is a rapidly growing stream of literature on ence of forward-looking customers. Fudenberg and remanufacturing and closed-loop supply chain man- Tirole (1998) study the monopoly pricing of overlap- agement. A comprehensive review of this literature is ping generations of a durable good with and without a given by Guide and Van Wassenhove (2009). Savaskan secondhand market. In an infinite-horizon model set- et al. (2004) study the optimal reverse channel struc- ting, Rao et al. (2009) demonstrate that trade-in rebates ture for the collection of used products from cus- can alleviate the inefficiencies arising from the lemon tomers. Ferguson and Toktay (2005) analyze the com- problem. Liang et al. (2014) analyze the optimal prod- petition between new and remanufactured products uct rollover strategies in the presence of strategic cus- (i.e., the cannibalization effect) and characterize the tomers. Lobel et al. (2016) study a new product launch optimal recovery strategy. In a case where remanu- strategy and show that a technology release precom- facturability is an endogenous decision, Debo et al. mitment can lead to significant profit improvement (2005) investigate a joint pricing and production tech- under forward-looking customer behavior. In this lit- nology selection problem of a manufacturer who erature, the work of Van Ackere and Reyniers (1995) is sells a remanufacturable product to heterogeneous probably the closest to our work. They also consider the customers. Geyer et al. (2007) demonstrate that to pricing problem in the presence of trade-ins and com- maximize the economic value of remanufacturing, pro- pare the market outcomes with strategic and myopic duction cost structure, product life cycle, and compo- customers. The key difference between our work and nent durability need to be carefully coordinated. Atasu theirs is that we highlight the critical role of remanufac- et al. (2008) show that remanufacturing could serve as a turing, whose efficiency would substantially influence marketing strategy to target customers in the green seg- the value of trade-in remanufacturing to the firm and ment and, hence, enhance the profitability of the orig- the environment. Moreover, we analyze how govern- inal equipment manufacturer (OEM). Galbreth et al. ment should design regulatory policies to maximize (2013) study how the rate of product innovation affects social welfare, which is absent in their work. the firm’s reuse and remanufacturing decisions. Gu Finally, our paper is also related to the literature et al. (2015) investigate the quality design and envi- on secondary markets, since the secondary market ronmental consequences of green consumerism with of a durable good also substantially influences cus- remanufacturing. The impact of trade-in rebates has tomer purchases (Hendel and Lizzeri 1999). While this also received some attention in the remanufacturing literature focuses on the benefits (e.g., Hendel and literature. For example, Ray et al. (2005) examine the Lizzeri 1999) and harms (Chen et al. 2013) of secondary value of price discrimination for new and repeat cus- markets, our work highlights the impact of customer tomers with differentiated ages (and qualities) of the purchasing behavior and remanufacturing efficiency Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 605 on both the economic and environmental values of 1 + α V of the upgraded second-generation product, ( ) trade-in remanufacturing. We also study the govern- where α 0 is exogenously given and captures the ≥ ment policy that helps induce the social optimum. The innovation level (e.g., the improved features) of the focus and insights of our paper are, therefore, quite upgraded product. Accordingly, let the production cost different from those in the secondary market literature. of the second-generation product be c2. To model product depreciation, we take the approach of Van Ackere 3. Model and Analysis and Reyniers (1995): If a type V customer has already 3.1. Model Setup bought the product in period 1, her valuation from We consider a monopoly firm who sells a product to consuming the used product in period 2 is 1 k V, ( − ) customers in a two-period sales horizon. In the first where k 0, 1 refers to the depreciation factor. Specif- ∈ [ ) period, the firm produces the first-generation product ically, if k is small, the product is highly durable; if k at a cost c1. The potential market size X is ex ante uncer- approaches 1, the product is almost useless in period 2 tain and continuously distributed with a distribution (either the product is worn out or the technology is function F and a density function f F0 . We obsolete). It can also be computed that the willingness assume that(·) all customers arrive at the(·) beginning(·) of to pay of a type V customer in period 2 is 1 + α V ( ) period 1, but our results and insights continue to hold if if she did not purchase the product in period 1 (i.e., a there are new customers arriving in period 2. The cus- new customer), and is 1 + α V 1 k V k + α V if ( ) − ( − ) ( ) tomers are infinitesimal, each requesting at most one she purchased the product in period 1 (i.e., a repeat cus- unit of the product in any period. Although demand tomer). We do not explicitly model the secondary mar- uncertainty is prevalent with new product introduc- ket of the used first-generation products, but studying tion, the firm can obtain more accurate demand infor- its impact in the presence of trade-in remanufacturing mation as the market matures. Hence, in period 2, the would be an interesting direction for future research. market uncertainty is resolved so the realized market As widely recognized in the literature, the firm can size X becomes known to the firm. generate revenues by extracting materials and compo- Let V denote a customer’s valuation of the first- nents from used products (see, e.g., Savaskan et al. generation product over the two-period horizon, which 2004, Ray et al. 2005). We now model the revenue- is an independent draw from a continuous distribution generating effect of remanufacturing. Customers who G on a support v, v¯ (0 v < v¯). We call the cus- (·) [¯ ] ≤ ¯ bought the product in period 1 can trade in the used tomer with product valuation V the type V customer. product for a new second-generation one at a discount At the beginning of the sales horizon, each customer price in period 2. Customers may incur an inconve- G knows only the distribution of her own valuation , nience cost to trade in used products. We do not explic- but not the realization V. This assumption captures(·) itly model this cost, but it can be absorbed into the customers’ uncertainties about product performance, trade-in price for repeat customers without affecting and fits our setting where the product is brand new at model analysis. The unit net revenue (i.e., cost saving) the beginning. In period 2, all customers observe their of remanufacturing is r , where r 0, c . Following own type V. Customers who purchased the product 2 2 ∈ [ 2] Savaskan et al. (2004), we assume all remanufactured in period 1 learn their type V by consuming the prod- products are upgraded to the quality standards of new uct. Customers who did not purchase the product in ones so customers cannot distinguish them from newly period 1 learn its quality and fit (thus, their type V) through social learning platforms (e.g., the Amazon made products. So our model best fits the setting where customer review system). Hence, the customers are used products were recycled and decomposed so that homogeneous ex ante (i.e., in period 1) but heteroge- its materials and components are reused in new prod- neous ex post (i.e., in period 2). This is a common set- uct manufacturing. Including a valuation gap between ting in models concerning customer purchasing behav- the new and remanufactured products will not change ior (see, e.g., Xie and Shugan 2001, Su 2009, Swinney the qualitative insights as long as the gap is sufficiently 2011). We assume that the valuation distribution G small. In our model, remanufacturing is performed by has an increasing failure rate, that is, h v : g v G¯ (·)v the firm itself (or by a third-party remanufacturer with ( ) ( )/ ( ) is increasing in v, where g G0 is the density a fixed cost). A potential avenue for future research is function and G¯ 1 G .(·) This is a(·) mild assumption to incorporate a strategic third-party remanufacturer and can be satisfied(·) − by(·) most commonly used distri- with pricing power into the current model setting. butions. Let µ : Ɛ V > c1, that is, in expectation, a The environmental impact of the product is the customer’s valuation( exceeds) the production cost. aggregate lifetime impact of the product on the envi- The firm offers an upgraded version of the product ronment. The total environmental impact is the pro- in period 2. This is a customary practice for product duction quantity multiplied by the per-unit impact categories like consumer electronics, home appliances, (see, e.g., Thomas 2011, Agrawal et al. 2012). Let and furniture. A type V customer has a valuation κi > 0 (i 1, 2) denote the unit environmental impact Zhang and Zhang: Trade-in Remanufacturing 606 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS of the first- and second-generation products, respec- the second-generation product, whereas all repeat cus- tively. Such impact may refer to the use of natural tomers decide whether to trade in their used prod- resources, emission of harmful gases, and generation ucts for new second-generation ones. Finally, the firm of solid wastes. Moreover, the values of κi can be esti- produces the second-generation products, recycles and mated by the conventional life-cycle analysis (see, e.g., remanufactures the used products from repeat cus- Agrawal et al. 2012). Let ι 0, κ be the unit environ- tomers, and collects the second-period revenue. 2 ∈ [ 2] mental benefit of remanufacturing the recycled first- To conclude this subsection, we remark that al- generation products in period 2. Note that ι2 mea- though this paper focuses on a two-period model, all sures the reductions in both the environmental impact results and insights can be generalized to an infinite- of producing the second-generation product and the horizon setting where each customer stays in the mar- end-of-use/end-of-life disposals of the first-generation ket for two periods. For conciseness, we define x y product, by recycling and reusing the materials and min x, y and x+ : max 0, x . A summary of the∧ nota- components. tions( is given) in the online( appendix.) A salient feature of our model is that the depreciation factor k (equivalently, the product durability) is 3.2. Customer Purchasing Behavior and correlated with other model primitives c1, r2, and ι2. Equilibrium Analysis As discussed above, a small k implies high product We use δ 0, 1 to denote the risk-free discount fac- ∈ ( ] durability (e.g., the firm uses better and more durable tor of the market, which is also the discount factor components and materials); as a result, the remanufac- of the firm. To study the impact of customer pur- turing efficiency is also higher in this case (Geyer et al. chasing behavior, we denote by δ 0, δ the dis- c ∈ [ ] 2007, Apple Inc. 2017a). To model this effect, we assume count factor of customers, which measures the inten- that the unit revenue generated by remanufacturing, sity of their forward-looking behavior. A large (resp., r2, and the unit environmental benefit of remanufac- small) δc implies customers care a lot (resp., little) turing, ι2, are both concavely decreasing in the depreci- about future utilities, and they are more (resp., less) ation factor k. More durable components and materials strategic. We call δc the customer discount factor and would also incur a higher production cost (Debo et al. the forward-looking behavior intensity interchangeably 2006, Geyer et al. 2007). Thus, we assume c1 is con- hereafter. If δc δ, customers are fully strategic and vexly decreasing in k. In our model, we take product maximize their long-run utilities; if δc 0, customers durability as a long-term strategic choice constrained are fully myopic and maximize their immediate util- by the technology. Hence, the depreciation factor k is ities. Using customer discount factor to capture their exogenously given before the firm makes the pricing and forward-looking behavior is a common approach in production decisions. However, along with the devel- the literature (e.g., Levin et al. 2009, Chevalier and opment of our results, we will briefly discuss how Goolsbee 2009). An alternative modeling approach is endogenizing the product durability/remanufacturing to assume there are two customer segments (strategic efficiency decision would affect our key insights. and myopic) in the market (e.g., Su 2007, Li et al. 2014). The sequence of events unfolds as follows. At the Under this approach, the intensity of forward-looking beginning of period 1, the firm announces the price behavior is measured by the proportion of strategic p1 and decides the production quantity Q1. Each cus- customers. Both approaches generate the same quali- tomer observes p1, but not Q1, and decides whether to tative insights, so we will focus on the former one for make a purchase or to wait until period 2. The first- ease of analysis and exposition. period demand X1 X is then realized, the firm col- We adopt the rational expectation (RE) equilibrium lects its first-period≤ revenue, and all customers stay framework to characterize the market outcome. Under in the market. Note that X1 is determined by the col- the RE equilibrium, each player makes decisions based lective effect of all customers’ purchasing decisions. If on individual beliefs, which are rationally formed X Q , any customer who requests a product can and consistent with actual outcomes. By backward 1 ≤ 1 get one in period 1. Otherwise, X1 > Q1, then the induction, we start with the subgame in period 2. Q products are randomly allocated to the demand, There are Xn X X Q new customers and 1 2 − ( 1 ∧ 1) and X Q customers have to wait because of the Xr X Q repeat customers in the market. Since 1 1 2 1 ∧ 1 limited− availability. At the end of period 1, the firm period 2 is the final period in our model, customers sells the leftover inventory at a (discounted) salvage with different intensities of forward-looking behavior price s r , c . At the beginning of period 2, the adopt the same purchasing strategy therein. Hence, ∈ ( 2 1) firm learns the realized total market size X, and each regardless of the customer discount factor δc, the firm individual customer learns her type V. The firm then should adopt the same pricing strategy in period 2 n n r n n r n n r announces the price p2 for new customers as well as well. Given X2 , X2 , let p2 X2 , X2 and Q2 X2 , X2 r n n r ( ) ( ) ( ) as the trade-in price p2 p2 (p2 p2 is the trade-in be the equilibrium price and production quantity for rebate). All new customers≤ decide whether− to purchase new customers in period 2. Analogously, we define Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 607

r n r r n r p2 X2 , X2 and Q2 X2 , X2 as the equilibrium trade-in customer discount factor δc. Based on Lemma1, the price( and) production( quantity) for repeat customers, following theorem characterizes the RE equilibrium n r and π2 X2 , X2 as the equilibrium second-period profit market outcome. of the firm.( ) Theorem 1. For any customer discount factor δc, there n r n n r n Lemma 1. (a) For any X2 , X2 , p2 X2 , X2 p2 ∗, and exists a unique RE equilibrium with (a) the price p1∗ δc r n r r (r n ) ( ) ≡ + ( ) p2 X2 , X2 p2∗, where p2∗ < p2 ∗. µ δc σr∗ σn∗ , (b) the production quantity Q1∗ δc ( ) ≡ n r n n r ¯ n + n ¯ 1 ( − ) ( ) (b) For any X2 , X2 , Q2 X2 , X2 G p2 ∗ 1 α X2 , F− c1 s m1∗ δc s , and (c) the expected total profit r n r ( ¯ r) +( r ) ( /( )) (( − Π)/( ( ) − )) and Q2 X2 , X2 G p2∗ k α X2. of the firm ∗f δc m1∗ δc s Ɛ X Q1∗ δc c1 s ( ) n r( /( n ))r n + r ( ) ( ( ) − ) ( ∧ ( )) − ( − )· (c) For all X , X , π2 X , X β∗ X β∗ X for some Q∗ δ + δβ∗ Ɛ X . ( 2 2) ( 2 2) n 2 r 2 1( c) n ( ) positive constants β∗ and β∗ . n r Theorem1 shows that the equilibrium first-period Lemma1 implies that both the equilibrium price for price is the expected valuation of the first-generation new customers and the equilibrium trade-in price are product µ plus the (discounted) expected surplus dif- independent of the realized market size Xn , Xr . In ference between a repeat customer and a new one ( 2 2) particular, the firm offers positive trade-in rebates to in period 2, δc σr∗ σn∗ . The equilibrium first-period r n ( − ) repeat customers (i.e., p2∗ < p2 ∗). The equilibrium profit production quantity, on the other hand, can be deter- n r of the firm in period 2, π2 X2 , X2 , is linearly separable mined by the solution of a corresponding newsvendor n r ( ) in X2 and X2, with the coefficients β∗n and β∗r captur- problem. ing the expected per-unit profit from new and repeat customers, respectively. 3.3. Benchmark Model Without Trade-in We now analyze the equilibrium market outcome in Remanufacturing period 1, starting with customers’ purchasing behavior. In the next two sections, we analyze the impact of cus- Each customer rationally anticipates the second-period tomer purchasing behavior and remanufacturing effi- n price for new customers, p2 ∗, and the second-period ciency on the value of trade-in remanufacturing, both r trade-in price, p2∗. The expected utility of a customer from the firm’s perspective and from an environmental + from purchasing the product in period 1 is Up µ perspective. To facilitate our comparison, we introduce + r + δc σr∗, where σr∗ Ɛ k α V p2∗ . The expected util- a benchmark model where the firm does not adopt [( ) − ] ity of a customer from waiting is Uw δc σn∗ , where trade-in remanufacturing. As a consequence, the firm + n + σn∗ Ɛ 1 α V p2 ∗ . Note that σn∗ (resp., σr∗) is the charges the same price for all customers and recycles expected[( surplus) − of a] new (resp., repeat) customer in no used products for remanufacturing in period 2. period 2. Therefore, a customer would opt to make a We call this the no trade-in remanufacturing (NTR) purchase in period 1 if and only if Up p1 Uw, that is, model, which is denoted by the superscript “u” here- − ≥ u n r p ξ δ : U U µ + δ σ∗ σ∗ . Following the after. We use p X , X to denote the equilibrium 1 r c p w c r n 2 ( 2 2) standard≤ ( approach) − in the literature( − (Xie) and Shugan second-period pricing strategy of the firm in the NTR 2001, Su and Zhang 2008, Cachon and Swinney 2011), model, which does not depend on customer purchas- we assume that all customers will make a purchase ing behavior. The characterization of pu , is given in 2 (· ·) in period 1 if p equals their reservation price ξ δ . Lemma 2 in the online appendix. As in the base model, 1 r ( c) Thus, with customer discount factor δc, the first-period customers form beliefs about the second-period prices X X X 1 and time their purchases. The firm, on the other hand, demand, 1, is given by 1 p1 ξr δc . Next, we consider the firm’s· problem{ ≤ ( )} in period 1. forms a belief about customers’ willingness to pay and To maximize its total expected profit, the firm sets the bases its (first-period) price and production decisions

first-period price p1 δc equal to the customer reser- on this belief. One can show that a unique RE equilib- vation price ξ δ ,( which) is the highest price cus- rium exists with any customer forward-looking behav- r ( c) tomers are willing to pay in period 1. Thus, the firm ior intensity δc 0, δ in the NTR model (Theorem 11 ∈ [ ] u u believes that the first-period demand is X1 X, the in the online appendix). Let p1 ∗ δc , Q1 ∗ δc denote n ( ( ) ( )) second-period market size of new customers is X2 the equilibrium first-period price and production deci- X Q +, and that of repeat customers is Xr X Q . sions of the firm in the NTR model. Accordingly, the ( − 1) 2 ∧ 1 Given the customer discount factor δc, the firm sets the associated equilibrium total firm profit is denoted by Πu first-period production quantity Q1 to maximize the f ∗ δc , which depends on the customer discount fac- ( ) total expected profit Π f Q1 δc Ɛ p1 δc X Q1 tor δc. + + + ( n | r ) { ( )( ∧ +) − c1Q1 s Q1 X δπ2 X2 , X2 , where p1 δc µ δc We define a few notations that will prove use- ( n− ) +( )}r ( ) σr∗ σn∗ , X2 X Q1 , and X2 X Q1. ful throughout our analysis. Given the first-period ( − ) ( − ) ∧ u + To characterize the RE equilibrium, we define an production quantity Q1, let σr Q1 : Ɛ k α V + + u n r + u (+ ) ((u n ) r −+ auxiliary variable m1∗ δc : µ δ β∗r β∗n δc σr∗ σn∗ . p2 X2 , X2 and σn Q1 : Ɛ 1 α V p2 X2 , X2 ( ) ( − ) ( − ) (n )) + r( ) (( ) − ( )) As will be clear in our subsequent analysis, m1∗ δc (X2 X Q1 , X2 X Q1) denote the expected is the first-period effective marginal revenue with( ) second-period( − ) surpluses for∧ repeat and new customers Zhang and Zhang: Trade-in Remanufacturing 608 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS in the NTR model. Clearly, σu and σu are the profit in some retail and airline settings (e.g., Su 2007, n (·) r (·) counterparts of σn∗ and σr∗ in the NTR model. If Li et al. 2014). u Q1∗ δc 0 or Q1 ∗ δc 0, the problem is reduced to We emphasize that both the trade-in option and an( uninteresting) one( ) with no repeat customer on the the remanufacturing process with proper efficiency are market in period 2. In this case, neither customer pur- essential for the firm to benefit from strategic customer chasing behavior nor the adoption of trade-in remanu- behavior: The former offers early purchase rewards to facturing matters. Thus, without loss of generality, we repeat customers, which can be well anticipated if cus- u assume Q1∗ > 0 and Q1 ∗ > 0 for the rest of our tomers are strategic, whereas the latter guarantees, in paper. (·) (·) expectation, a repeat customer enjoys a higher surplus than a new customer in period 2. If the firm does not adopt trade-in remanufacturing or the remanu- 4. Value of Trade-in Remanufacturing for facturing efficiency is too high or too low, however, the Firm the expected repeat customer surplus will be lower This section investigates the value of trade-in reman- u than the expected new customer surplus (i.e., σr Q1 < ufacturing from the firm’s perspective. To begin with, u ¯ ( ) σn Q1 for all Q1, and σr∗ < σn∗ if k < k or k > k). In these we perform a sensitivity analysis with respect to the cases,( ) the more strategic the customers,¯ the more reluc- customer discount factor δc, so as to unveil insights on tant they are to make immediate purchases in period 1, the role of customer purchasing behavior. and, as a consequence, the lower the firm profit. Theorem 2. (a) Under trade-in remanufacturing, we have From Theorem2, we can derive some actionable insights for practitioners. In a market with frequent the following: (i) p∗ δc , Q∗ δc , and Π∗ δc are strictly in- 1( ) 1( ) f ( ) new product introductions and intensive strategic cus- creasing (resp., decreasing) in δc if σr∗ > σn∗ (resp., σr∗ < σn∗ ), ¯ tomer behavior (e.g., the electronics market; see Song and (ii) there exist two thresholds, k and k, such that σr∗ > σn∗ ¯ ¯ and Chintagunta 2003, Plambeck and Wang 2009), the (resp., σr∗ < σn∗ ) if and only if k k, k (resp., k < k or k > k¯). ∈ (¯ ) ¯ firm should neither completely abandon remanufac- (b) Under no trade-in remanufacturing (i.e., the NTR turing nor haphazardly improve remanufacturing effi- u ciency. Instead, keeping remanufacturing moderately model), we have the following: (i) p ∗ δ is strictly decreas- 1 c ¯ ( )u Πu efficient (i.e., k k, k ) enables the firm to leverage ing in δ on 0, δ for some δ > 0, (ii) Q ∗ δ and ∗ δ c [ 0] 0 1 ( c) f ( c) the intensive forward-looking∈ (¯ ) customer behavior via are strictly decreasing in δ , and (iii) σu Q < σu Q for c r 1 n 1 its trade-in remanufacturing program. On the other all Q 0. ( ) ( ) 1 ≥ hand, if the firm does adopt trade-in remanufacturing Interestingly, Theorem2 demonstrates that the firm with moderate remanufacturing efficiency, it would be can earn a higher profit with more strategic cus- a good idea to induce more intensive strategic customers if (1) trade-in remanufacturing is adopted and tomer behavior by extensively advertising the trade-in (2) remanufacturing efficiency is moderate, that is, k opportunities in the market. k, k¯ . It is worth noting that increasing remanufac-∈ Our next result characterizes how strategic cus- turing(¯ ) efficiency has two effects on the firm: First, tomer behavior intensity and remanufacturing effi- it enhances the revenue from remanufacturing (i.e., ciency impact the value of trade-in remanufacturing to r2 increases); second, it improves the durability of the firm. the product (i.e., k decreases), which discourages cus- u u Theorem 3. (a) p∗ δ > p ∗ δ and Π∗ δ > Π ∗ δ for tomers from trading in and upgrading their first- 1( c) 1 ( c) f ( c) f ( c) all δ 0, δ . (b) There exists a threshold K such that Π∗ δ generation products. The former effect is to the ben- c ∈ [ ] f ( c) is increasing (resp., decreasing) in k when k K (resp., efit of the firm, whereas the latter is to its detriment. ≤ k K). (c) ∂p∗ δ ∂k is increasing in δ . If remanufacturing efficiency is moderate, the above ≥ | 1( c)/ | c two effects are well balanced so that remanufacturing As one may expect, the firm earns a higher profit could generate a high revenue without overly discour- under trade-in remanufacturing, since it has higher aging repeat customers from upgrading their products. pricing and production flexibilities. A particularly In this case, repeat customers enjoy a higher expected interesting implication of Theorem3 is that the firm surplus than new customers in period 2 (i.e., σr∗ > σn∗ ), may not benefit from higher remanufacturing effi- thus driving the firm to charge a higher price, pro- ciency (note that Π∗ δ is increasing in k for k K). f ( c) ≤ duce more, and, consequently, earn a higher profit If remanufacturing is overly efficient (k K), further ≤ with more strategic customers (i.e., with a higher δc). improving remanufacturing efficiency would make the Therefore, with moderate remanufacturing efficiency, first-generation product so durable that repeat cus- trade-in remanufacturing allows the firm to exploit tomers are reluctant to trade in and upgrade their used and benefit from strategic customer behavior. This products, thus hurting the firm’s profit. The thresh- insight complements the findings in the literature that old K can be interpreted as the optimal product dura- strategic customer behavior may improve a seller’s bility/remanufacturing efficiency for the firm. Our Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 609

Table 1. Summary of Numerical Setup Figure 1. (Color online) Box Plot of γ δ ( c ) Value of trade-in remanufacturing to the firm δ: 0.95 δ : 0, 0.25δ, 0.5δ, 0.75δ, δ 0, 0.2375, 0.475, 0.7125, 0.95 c { } { } X: Gamma distribution with mean 100 and standard deviation 60 50, 60, 70, 80, 90 { } V: Uniformly distributed on 0, 1 [ ] 50 k: 0.3, 0.4, 0.6, 0.6, 0.7 { } c1: 0.05, 0.1, 0.15, 0.2, 0.25 { } 40 κ1: 1 s: 0

c2: 0.25, 0.2625, 0.275, 0.2875, 0.3 30 { } κ2: 0.75 r2: 0 α: 0, 0.05, 0.1, 0.15, 0.2 20 { } Increase of profit (%) ι2: 0.3 10 analysis also reveals that the operational impact of 0 remanufacturing efficiency is strengthened by strategic customer behavior. This is because strategic customers 0 0.2357 0.4750 0.7125 0.9500 are more sensitive to anticipated future adjustments of Customer discount factor trade-in price caused by changes in remanufacturing efficiency. of interesting observations from Table1 and Figure1. There are three beneficial effects of trade-in reman- First, the value of trade-in remanufacturing generally ufacturing: (1) the revenue-generating effect of reman- improves as the customers become more strategic. Sec- ufacturing, that is, remanufacturing can recover the ond, and more importantly, compared to the case with residual value of used products; (2) the price- myopic customers (δc 0), the value of trade-in reman- discrimination effect of trade-in rebates, that is, the dif- ufacturing could be much higher when customers are ferentiated prices for new and repeat customers help strategic (e.g., comparing the results of δc 0 and δc the firm exploit customer segmentation; and (3) the 0.95). This suggests that the benefits of exploiting strate- early purchase inducing effect of trade-in rebates, that gic customer behavior could be much more significant is, the price discount for repeat customers attracts than the benefits of revenue generating and price dis- strategic customers to purchase early. Note that the crimination (the latter two benefits can be captured first two effects benefit the firm regardless of customer already with myopic customers). Thus, our numerical behavior, whereas the third effect improves its profit results deliver an important message that a major driver only if customers are strategic. Moreover, while the first of using trade-in remanufacturing is to exploit strategic effect will be enhanced if remanufacturing efficiency customer behavior, which has not been identified in the improves (i.e., k decreases), the second and third effects previous literature. In addition, such a benefit will be are not monotonic in k. The overall effect of the above magnified if customers are more strategic. three driving forces is that the firm profit Π∗ δ is not f ( c) It is worthwhile differentiating our results from the monotonic in remanufacturing efficiency either. durable goods and secondary market literature that We now use extensive numerical experiments to studies the value of trade-ins/resales in the presence derive additional insights into the value of trade-in of forward-looking customers (e.g., Van Ackere and remanufacturing for the firm under different customer Reyniers 1995, Fudenberg and Tirole 1998, Waldman Π behaviors. The metric of interest is γ δc ∗f δc 2003, Hendel and Lizzeri 1999). The main finding u u ( ) ( ( ) − Π ∗ δ Π ∗ δ 100%, which measures the relative of this literature is that when customers are for- f ( c))/ f ( c) × profit improvement of trade-in remanufacturing with ward looking, trade-ins and resales in the secondary customer discount factor δc. We evaluate γ δc under a market could be beneficial to the firm, because cus- set of 3,125 parameter combinations that cover( ) a wide tomers can anticipate their values when making ini- range of reasonable problem scenarios. The parameter tial purchases. Our results complement this insight in values used in our numerical experiments are sum- the context of remanufacturing. Specifically, if com- marized in Table1. Our focus is on examining the plemented with a moderately efficient remanufactur- impact of customer discount factor δc, which corre- ing process, trade-ins could serve as a lever for the sponds to different customer purchasing behaviors, firm to exploit strategic customer behavior (Theo- on γ δc , the (relative) profit improvement of trade-in rem2, Table1, and Figure1). While remanufactur- remanufacturing.( ) ing generates revenue, the firm does not benefit from

We present the summary statistics of γ δc in Table2 overly eﬃcient remanufacturing under trade-ins (The- and the box plot of γ δ in Figure1. There( are) a couple orem3). Both results highlight that, with moderate ( c) Zhang and Zhang: Trade-in Remanufacturing 610 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS

Table 2. Summary Statistics of γ δ (%) ( c ) Min 25th percentile Median 75th percentile Max Mean Standard deviation δc 0 0.008 1.0 2.5 4.6 11.7 3.1 2.5 δc 0.2375 3.2 7.4 10.5 13.6 18.2 10.4 3.7 δc 0.475 6.6 11.4 16.6 22.5 30.5 17.1 6.1 δc 0.7125 7.1 16.7 23.2 31.8 47.6 24.5 9.4 δc 0.95 5.8 19.9 28.3 40.1 61.6 30.2 13.1 remanufacturing efficiency (thus moderate product behaviors and remanufacturing efficiencies. To begin durability), the value of trade-ins to the firm could with, we study how strategic customer behavior be further elevated. Overly efficient remanufactur- intensity impacts the environmental value of trade- ing, however, coincides with high product durability in remanufacturing. In equilibrium, the total envi- and production cost, and therefore discourages repeat ronmental impact should be the difference between customers from upgrading their used products. This the total environmental impact of production/disposal mechanism, which is in line with the idea of planned and the total environmental benefit of remanufac- obsolescence (e.g., Waldman 2003), dilutes the value turing. Hence, the equilibrium environmental impact + of trade-in remanufacturing to the firm under high with trade-in remanufacturing is Ie∗ δc Ɛ κ1Q1∗ δc n n r + r n r ( ) {r n ( r ) remanufacturing efficiency. δκ2 Q2 X2 ∗, X2∗ Q2 X2 ∗, X2∗ δι2Q2 X2 ∗, X2∗ , ( ( n ) ( + ))r − ( )} To conclude this section, we briefly discuss the set- where X ∗ X Q∗ δ and X ∗ X Q∗ δ . In this 2 ( − 1( c)) 2 ∧ 1( c) ting with endogenized remanufacturing efficiency. In section, we make an additional assumption that κ1 the literature, the design of remanufacturability is usu- ¯ n + ≥ κ2G2 p2 ∗ 1 α . This assumption is not restrictive in ally studied without taking into account customers’ practice,( /( and can)) be satisfied when the environmental forward-looking behavior (e.g., Debo et al. 2005), impact of the first-generation product is not too low. which is a key element of our model. An immediate In particular, it applies to the case where the newer implication of Theorem3 is that if the firm can adjust generation product dominates the older generation in its remanufacturing efficiency at no additional cost, terms of environmental sustainability, that is, κ1 κ2 > then the optimal level of remanufacturing efficiency ¯ n + ≥ δκ2G2 p2 ∗ 1 α . K will be chosen. In practice, however, higher reman- How( does/( customer)) purchasing behavior affect the ufacturing efficiency requires higher investments in, value of trade-in remanufacturing to the environment? for instance, recycling and remanufacturing technol- The answer is given in the next theorem, where we ogy (e.g., the robot Liam invented by Apple for reman- u define Ie ∗ δc as the equilibrium total environmental ufacturing; see Apple Inc. 2017a). Thus, one can model impact in( the) NTR model. the upfront fixed cost of installing remanufacturing ¯ Theorem 4. (a) There exists a threshold δq < δ such that technology ct k as a convexly decreasing function of u ( ) if δ > δ¯ , I∗ δ > I ∗ δ for δ > δ¯ unless ι is too large the depreciation factor k. The equilibrium depreciation c q e ( c) e ( c) c q 2 level k∗ should be the one that maximizes the profit (formally specified in the online appendix). with endogenized remanufacturing efficiency, that is, (b) Assume that c1 is sufficiently small (formally speci- Π Π fied in the online appendix). There exists a threshold δ > 0 k∗ arg maxk 0, 1 ∗f δc ct k . Since ∗f δc ct k is q ∈[ ]{ ( )− ( )} ( )− ( ) u ¯ k such that Ie ∗ δc > Ie∗ δc for δc < δq unless ι2 is too small not necessarily quasi-concave in , the analysis under ( ) ( ) ¯ endogenized remanufacturing efficiency is quite chal- (formally specified in the online appendix). lenging. However, we can show that k∗ > K > 0, so the Theorem4 reveals a dichotomy on the environmen- insight that overly efficient remanufacturing could be tal value of trade-in remanufacturing. With highly detrimental to the firm will still hold. Furthermore, our strategic customers, trade-in remanufacturing is likely numerical results suggest that how the value of trade- to be detrimental to the environment, whereas with in remanufacturing depends on the customer discount mostly myopic customers, trade-in remanufacturing factor δc is robust over different values of k. Thus, even may benefit both the firm and the environment. Our under endogenized remanufacturing efficiency, trade- next goal is to understand why the contrasting effects in remanufacturing could help exploit strategic cus- of trade-in remanufacturing with different customer tomer behavior, and delivers higher value for the firm behaviors would occur. Recall that trade-in reman- if customers are more strategic. ufacturing offers strategic customers early purchase rewards, so they purchase more in the first place and 5. Value of Trade-in Remanufacturing for recycle used products more frequently. As a conse- the Environment quence, trade-in remanufacturing induces larger pro- We proceed to examine the environmental value of duction quantities and accelerates product rollover, trade-in remanufacturing under different customer which places higher pressure on the environment. Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 611

If customers are myopic, however, trade-in remanu- Figure 2. (Color online) Box Plot of η δ ( c ) facturing could not motivate them to purchase early. Value of trade-in remanufacturing to the environment Instead, its price-discrimination effect enables the firm 180 to earn a higher (per-customer) profit from new customers. Hence, the firm will lower the first-period 160 production quantity to enlarge the (potential) second- 140 period market size of new customers and to deceler- 120 ate product rollover. The following theorem formalizes the above intuition and characterizes the contrasting 100 effects of trade-in remanufacturing on the production 80 decision under different customer behaviors. 60 u Theorem 5. (a) Q∗ δ > Q ∗ δ for all δ δ¯ , δ . 1( c) 1 ( c) c ∈ ( q ] 40 (b) Assume that c1 is sufficiently small. Then Q1∗ δc < u ( ) 20 Q1 ∗ δc for all δc 0, δq . In particular, if σr∗ > σn∗ , we have ( ) ∈ [ ¯ ) u ¯ ¯ Increase of environmental impact (%) δq δq, that is, Q1∗ δc > Q1 ∗ δc for all δc > δq δq, and 0 ¯ u ( ) ¯( ) ¯ Q1∗ δc < Q1 ∗ δc for all δc < δq δq. ( ) ( ) ¯ 0 0.2375 0.4750 0.7125 0.9500 Theorems5(a) and4(a) echo the findings of a few Customer discount factor related works in the literature (e.g., Debo et al. 2005, Galbreth et al. 2013, Gu et al. 2015) that remanu- with more intensive strategic customer behavior (η δ facturing may increase the production quantity and ( c) thus worsen the environment. Our analysis strength- is significantly higher with a larger δc). Though bene- ens this insight and demonstrates that the environmen- ficial to the firm (see Table1 and Figure1), the early tal value of trade-in remanufacturing depends criti- purchase inducing effect of trade-in remanufacturing cally on customer purchasing behavior. If customers gives rise to much higher production quantities under are mostly myopic, trade-in remanufacturing actually more intensive forward-looking customer behavior, leads to lower production quantities and, thus, a bet- thus dominating the recycling effect of remanufactur- ter environment (Theorems5(b) and4(b)). This result ing and leading to a much worse outcome from an highlights the necessity of understanding customer environmental perspective. purchasing behavior in the market when evaluating The above results suggest that customer purchasing the environmental value of trade-in remanufacturing. behavior has opposing effects on the value of trade- We now numerically illustrate the environmental in remanufacturing: Intensive forward-looking behav- values of trade-in remanufacturing under different ior of customers makes this strategy attractive to the customer purchasing behaviors. We employ the same firm, but not desirable for the environment. In par- numerical setup as in Section4 (see Table1). We ticular, trade-in remanufacturing may create a tension u between firm profitability and environmental sustain- are interested in the metric η δc : Ie∗ δc Ie ∗ δc u ( ) ( ( ) − ( ))/ ability when customers are highly strategic, but benefits I ∗ δ 100%, referring to the relative change in envi- e c both the firm and the environment with myopic cus- ronmental( ) × impact after adopting trade-in remanufac- tomers. When customers are highly strategic (δ δ), turing. If η δ > 0, trade-in remanufacturing increases c c the early purchase inducing effect dominates the envi-≈ the total negative( ) impact and is, thus, detrimental ronmental benefit of remanufacturing. In this case, the to the environment. Otherwise, if η δ 0, trade-in c firm benefits from trade-in remanufacturing signifi- remanufacturing benefits the environment.( ) ≤ cantly, but the environment suffers a lot. Thanks to In our experiments, we evaluate η δ under the c the economic and environmental benefits of remanu- 3,125 parameter combinations detailed( in) Table1 and facturing and the price-discrimination effect of trade- obtain the following findings: (1) under each param- in rebates, both the firm and the environment, how- eter combination, η δ is significantly higher as δ c c ever, will benefit from trade-in remanufacturing when increases (i.e., customers( ) become more strategic), and customers do not exhibit strong strategic purchasing (2) it exhibits the clear pattern that as δ increases, the c behaviors (i.e., δ 0). proportion of problem instances with a positive η δ c c Our results reveal≈ a tension between profitabil- also becomes larger. For different values of δ , the sum-( ) c ity and sustainability under the adoption of trade- mary statistics are presented in Table3, whereas the in remanufacturing. The following theorem further box plot is depicted in Figure2. shows that this tension is most prominent under mod- Table3 and Figure2 confirm that the environmental erate remanufacturing efficiency. value of trade-in remanufacturing is highly sensitive to customer purchasing behavior. Trade-in remanufactur- Theorem 6. (a) Ie∗ δc is strictly increasing (resp., decreasing leads to much higher total environmental impact ing) in δ if k (k, k¯) (resp., k < k or k > k¯). (b) There c ∈ (¯ ) ¯ Zhang and Zhang: Trade-in Remanufacturing 612 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS

Table 3. Summary Statistics of η δ (%) ( c ) Min 25th percentile Median 75th percentile Max Mean Standard deviation

δ 0 10.2 6.9 5.5 3.8 4.5 5.0 2.7 c − − − − − δ 0.2375 5.6 3.6 1.8 7.9 39.9 2.6 8.7 c − − − δ 0.475 3.9 0.8 3.0 26.8 88.7 13.1 18.6 c − − δ 0.7125 2.3 4.3 15.8 58.2 129.1 30.9 30.5 c − δ 0.95 1.2 10.0 37.8 87.1 172.0 49.2 41.4 c −

exists a threshold K such that I∗ δ is increasing (resp., pricing and production strategy. Interestingly, Theo- e e ( c) decreasing) in k if k K (reps. k K ). rem7 contrasts with our intuition that higher produc- ≤ e ≥ e By comparing Theorem6 with Theorems2 and3, tion quantities (see Theorem5(a)) will lead to higher one can see that trade-in remanufacturing has oppo- consumptions and thus a higher customer surplus. site impacts on the environment and firm profit. In the However, our analysis suggests that, under intensive scenario where the impact on firm profit is most sig- strategic customer behavior, trade-in remanufacturing nificant (i.e., remanufacturing efficiency is moderate), increases production quantities (thus hurting the envi- the associated environmental impact is also the high- ronment) without improving customer surplus. Fur- est. This observation further highlights the aforemen- thermore, our numerical study indicates that the social tioned profitability–sustainability tension under trade- welfare (i.e., firm profit plus customer surplus less in remanufacturing and demonstrates that this tension environmental impact) may decrease under trade-in is most significant when remanufacturing efficiency is remanufacturing as well. moderate. To summarize, customer purchasing behavior and Although an increased production quantity means remanufacturing efficiency play important roles in more pressure on the environment, it also increases the economic and environmental values of trade-in the consumption level of the product. We next explore remanufacturing. In a market with not-so-strategic cus- the impact of trade-in remanufacturing on total customers, trade-in remanufacturing benefits both the

tomer surplus under different customer purchasing firm and the environment. With highly strategic cus- behaviors. When customers are highly strategic (i.e., tomers, however, trade-in remanufacturing is even the profitability–sustainability tension is most inten- more beneficial to the firm, while it seriously hurts sive), there are two opposing effects of trade-in reman- the environment, decreases customer surplus, and pos- ufacturing on customer surplus. First, the first-period sibly lowers social welfare. In this case, the value u production quantity is larger (Q1∗ δc > Q1 ∗ δc for δc > of trade-in remanufacturing is mainly about help- ¯ ( ) ( ) δq), so that customers can earn a higher total sur- ing the firm to exploit strategic customer behavior, plus in period 1. Second, the second-period price for which is much more significant than the widely recog- u n r n new customers is lower (p X , X < p ∗), so that 2 ( 2 2) 2 nized revenue-generating and environmental benefits the customer surplus will suffer in period 2 (σn∗ < u u of remanufacturing. The tension between profitability σ Q ∗ δ ). If the first (resp., second) effect plays a n ( 1 ( c)) and sustainability, in addition, is strengthened by mod- dominating role, trade-in remanufacturing enhances erate remanufacturing efficiency. In this case, reman- (resp., reduces) customer surplus. Let S stand for total c ufacturing generates revenue without overly discour- customer surplus. The following theorem shows the aging repeat customers from upgrading used products latter effect actually dominates, and thus customer sur- through the trade-in program, thus prompting the plus is diminished by trade-in remanufacturing if cus- firm to produce more and accelerate product rollover. tomers are highly strategic. In short, when making decisions related to trade-in ¯ Theorem 7. There exists a threshold δs such that Sc∗ δc < remanufacturing, firms and policy makers should keep u ¯ ( ) Sc ∗ δc for δc > δs . in mind the customer purchasing behavior and reman- ( ) Theorem7 compares the total customer surpluses ufacturing efficiency in the focal market. in the base model and the NTR model. In particular, we show that when customers are sufficiently strate- ¯ 6. Social Optimum and Government gic (i.e., δc > δs ), they are worse off with the adoption u of trade-in remanufacturing (Sc∗ δc < Sc ∗ δc ). Strategic Intervention customers well perceive the potential( ) price( ) discounts With increasing societal awareness of sustainability, of trade-in rebates, and thus are more willing to pur- the question of how to regulate a market with envi- chase in period 1. The firm, on the other hand, extracts ronmental concerns has attracted increasing atten- its second-period surpluses by complementing trade- tion from the government (e.g., Xie and Bai 2010, in remanufacturing with a wisely designed first-period Department for Business, Innovation and Skills 2015). Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 613

As shown in Sections4 and5, adopting trade-in reman- Theorem 8. (a) With customer discount factor δc, we have s + s s s ¯ 1 + ufacturing may create a tension between firm prof- p1∗ δc µ δc σr∗ σn∗ and Q1∗ δc F− c1 κ1 s (s ) ( − ) ( ) (( − )/ itability and environmental sustainability under inten- m1∗ s , and the equilibrium expected social welfare is ( − )) s s + s + sive strategic customer behavior. In this section, we Ws∗ δc m1∗ s Ɛ X Q1∗ δc c1 κ1 s Q1∗ δc s( ) ( − ) ( ∧ ( )) − ( − ) ( ) analyze how a policy maker (e.g., government) can δσn∗Ɛ X . design the public policy to resolve this tension and [ s] (b) p1∗ δc is strictly increasing (resp., decreasing) in δc maximize the social welfare. ( ) s s s s if and only if σr∗ > σn∗ (resp., σr∗ < σn∗). There exist two We first characterize the socially optimal market out- ¯ ¯ s s thresholds, ks and ks (ks < ks ), such that σr∗ > σn∗ (resp., come by assuming that the government can set the s s ¯ ¯ ¯ ¯ σr∗ < σn∗) if and only if k ks , ks (resp., k < ks or k > ks ). prices and production levels in both periods, with an s ∈ (¯ ) ¯ In addition, Q1∗ δc and Ws∗ δc are independent of δc. objective to maximize the social welfare. Let Ws denote ( ) ( ) (c) There exists a threshold Ks such that W ∗ δc is in- the social welfare, which is defined by the expected s ( ) creasing (resp., decreasing) in k for k Ks (resp., k Ks ). profit of the firm Π f , plus the expected customer sur- ≤ ≥ plus Sc, net the expected environmental impact Ie , Since the social planner balances firm profit, cus- that is, tomer surplus, and environmental impact, whereas the W Π + S I . firm maximizes its own profit only, the social welfare– s f c − e Note that the government revenues and costs to reg- maximizing equilibrium outcome may be quite dif- ulate the market (i.e., taxes and subsidies) are not ferent from the profit-maximizing one, as shown by included in the computation of social welfare. This comparing Theorem8 with Theorem1. Observe that is because the taxes and subsidies are transactions the equilibrium social welfare–maximizing first-period s between different stakeholders in a society, and there- production quantity Q ∗ δc is independent of customer 1 ( ) fore would not affect social welfare directly. All results discount factor δc; so is the equilibrium social welfare and qualitative insights derived in this section are W ∗ δ . This is in sharp contrast to the equilibrium s ( c) robust and will continue to hold if the government has outcome in the base model, which depends critically a budget constraint so that the total expected cost to on customer purchasing behavior (Theorem2). Similar implement the government policy cannot exceed the to the profit-maximizing equilibrium, the equilibrium s budget limit. social welfare–maximizing first-period price p1∗ δc As in the base model, we start with the second- depends on the forward-looking customer behavior( ) period pricing and production problem. For any intensity δc. As remanufacturing becomes more effi- n r given realized market size X2 , X2 in period 2, we cient (i.e., k decreases), the optimal social welfare, n n r r n r ( ) use ps X2 , X2 , ps X2 , X2 to denote the equilibrium W ( ( ) ( )) s∗ δc , first increases and then decreases. Therefore, second-period pricing strategy that maximizes the higher( ) remanufacturing efficiency does not necessar- social welfare, and denote by w Xn , Xr the equi- 2( 2 2) ily improve the social welfare, even if the government librium second-period social welfare. As shown by tries to maximize the social welfare by directly exer- Lemma 4 (in the online appendix), under the socially cising price and production decisions. Overly efficient optimal second-period pricing strategy, the prices for remanufacturing is associated with an overly durable new and repeat customers are equal to the respective product, thus discouraging customers from trading in net total unit production and environmental cost (i.e., n n r n + r n r r their used products. Such a mechanism hurts the over- ps X2 , X2 ps ∗ c2 κ2 and ps X2 , X2 ps∗ c2 +( ) ≡ ( ) ≡ − all social welfare. This is similar to the impact of reman- r2 κ2 ι2). Moreover, the equilibrium social welfare − n r ufacturing efficiency in the base model, where the firm is linear in the realized market size X2 , X2 ; that is, n r s n + s r ( ) seeks to maximize its own profit (Theorem3(b)). w2 X2 , X2 σn∗X2 σr∗X2, where the linear coefficient s ( ) s We proceed to study how the government, whose σn∗ (resp., σr∗) is the equilibrium expected surplus of a new (resp., repeat) customer, which is also the equilib- objective is to maximize the expected social welfare rium unit social welfare of selling to new (resp., repeat) Ws , could induce the firm, whose objective is to max- Π customers in period 2. imize its expected profit f , to set the socially opti- In period 1, the government and customers base mal prices and production quantities. We consider a s s general class of policies under which government sub- their decisions on rational beliefs. Let p1∗ δc , Q1∗ δc denote the equilibrium first-period price( and( ) produc-( )) sidies (resp., taxes) are provided (resp., charged) for the sales of both product generations and for the recy- tion quantity with customer discount factor δc. Analo- gous to the base model, we introduce the first-period cling/remanufacturing of used products. Specifically, s s s effective marginal welfare, m ∗ : µ + δ σ ∗ σ ∗ , which let sg : s1, s2, sr denote the subsidy/tax scheme. The 1 ( r − n ) ( ) measures the marginal social welfare to produce in government offers the firm a per-unit subsidy s1 for the period 1. The following theorem characterizes the production of the first-generation product, a per-unit social welfare–maximizing equilibrium outcome and subsidy s2 for the production of the second-generation analyzes the impact of customer purchasing behavior product, and a per-unit subsidy sr for remanufactur- and remanufacturing efficiency. ing/recycling. If si < 0 (i 1, 2, r), the government taxes Zhang and Zhang: Trade-in Remanufacturing 614 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS the production of respective product version or the actually intensifies the aforementioned profitability– recycling and remanufacturing of used products. This sustainability tension: it further increases the firm general subsidy/tax scheme is motivated by the cur- profit and environmental impact simultaneously. This rent common practice in the electronics industry that is because subsidizing remanufactured products not the government establishes a fund for the treatment only promotes the adoption of remanufacturing, but of WEEEs. OEMs contribute to this fund in the form also increases the production levels of the first- of taxes, whereas firms that recycle and remanufac- generation product. The environment thus suffers from ture used products get subsidies from this fund. See the increased production levels under the subsidization Chinese Ministry of Finance (2012) and Department for remanufacturing alone. Therefore, the government for Business, Innovation and Skills (2015) for exam- should be careful about designing the subsidization ples in China and the United Kingdom, respectively. policy, because haphazard subsidization for trade-in In our model, the firm both manufactures new prod- remanufacturing may result in an undesired outcome. ucts and recycles used ones, so it may get both taxed Finally, we examine the firm’s profit under the sub- and subsidized. The proposed government subsidiza- sidy/tax scheme that induces the social optimum. tion/taxation policy is quite general. Quite a few spe- Specifically, we are interested in how customer pur- cial forms of the proposed general policy have been chasing behavior and remanufacturing efficiency will discussed in the literature; see, for example, Calcott impact the firm’s profit under the socially optimal gov- and Walls (2000), Webster and Mitra (2007), Ma et al. ernment policy. (2013), and Wang et al. (2014). Theorem 10. (a) The profit of the firm under the sub- s s Theorem 9. For each customer discount factor δc 0, δ , sidy/tax scheme, Π ∗ δ , is independent of δ . (b) Π ∗ δ is ∈ [ ] f c c f c there exists a unique linear subsidy/tax scheme s∗ δc ( ) ( ) g( ) increasing (resp., decreasing) in k if k Ks (resp., k Ks ). s∗ δc , s∗ δc , sr∗ δc under which the social welfare– ≤ ≥ ( 1( ) 2( ) ( )) Thus, under the subsidy/tax scheme s∗g δc , if the firm has the maximizing RE equilibrium outcome is achieved. flexibility to set remanufacturing efficiency( ) k, it will select

For any customer discount factor δc, the government the socially optimal one, Ks . can use a simple linear subsidy/tax scheme, s∗ δ , to g( c) In contrast to the setting where government reg- induce the social optimum. The linear subsidy/tax pol- ulation is not explicitly considered (Theorem2(a)), icy sg has great flexibility in controlling the margin the firm profit under the socially optimal government of the firm and the willingness to pay of customers. policy is independent of customer purchasing behav- Hence, the government can use this incentive scheme, ior. This property further highlights the capability of if well designed, to regulate the market and ensure that the proposed subsidy/tax scheme to counter strate- the firm sets the socially optimal prices and produc- gic customer behavior. Another salient feature of the tion quantities, and the customers make the socially socially optimal subsidy/tax scheme is that if the firm optimal purchasing decisions accordingly. More specif- can adjust remanufacturing efficiency without addi- ically, the government should provide a combined sub- tional costs, this policy can also induce the firm to sidy/tax scheme for the production of both product set the socially optimal remanufacturing efficiency, K . generations and the recycling of used products. The s Therefore, the proposed subsidy/tax scheme also components of s∗ δ may have different signs, so it is g c incentivizes the firm to make the socially optimal possible that the government( ) will tax the firm on one remanufacturing technology choice by setting reman- product generation and subsidize it for another (e.g., ufacturing efficiency at the socially optimal level K . the Chinese government charges the OEM for the pro- s In summary, to alleviate the tension between prof- duction of electrical and electronic products and sub- itability and sustainability and achieve the social sidizes those who recycle and remanufacture e-waste). optimum, it suffices for the government to use an This phenomenon results from the government’s goal easy-to-implement linear subsidy/tax scheme for the of balancing the trade-off between firm profit, cus- production of both product generations and reman- tomer surplus, and environmental impact. We remark ufacturing. The proposed subsidy/tax scheme could that, expectedly, the optimal subsidy/tax rate for the induce the firm to both adopt the operational (i.e., first-generation product s1∗ δc is sensitive to changes in customer purchasing behavior( ) δ , whereas the reman- production and pricing) strategies that maximize the c social welfare and to set the (socially) optimal remanu- ufacturing efficiency k will affect s∗ δ and the optimal 1( c) facturing efficiency level. subsidy/tax rate for remanufacturing sr∗ δc . To promote remanufacturing and leverage( ) its environmental benefit, the government sometimes adopts 7. Conclusion the policy to subsidize for remanufacturing alone In this paper, we develop an analytical model to study (e.g., Chen 2015, Recycler 2015). This intuitive policy how customer purchasing behavior and remanufac- is a special case of our general subsidy/tax scheme turing efficiency influence the economic and environ- with s1 s2 0 and sr > 0. However, this policy mental values of trade-in remanufacturing. From the Zhang and Zhang: Trade-in Remanufacturing Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS 615

firm’s perspective, we identify a new benefit of trade-in Acknowledgments remanufacturing, that is, it helps exploit the forward- The authors would like to thank Hau Lee, the anonymous looking behavior of strategic customers. A trade-in associate editor, and two referees for their constructive com- rebate essentially offers an early purchase reward and ments that helped improve the paper. thus can deliver higher additional value when customers are highly strategic. In particular, under the References adoption of trade-in remanufacturing with moder- Agrawal V, Ferguson M, Toktay LB, Thomas V (2012) Is leasing greener than selling. Management Sci. 58(3):523–533. ate remanufacturing efficiency, more intensive strate- Apple Inc. (2017a) Apple environmental responsibility report. gic customer behavior can increase the firm profit. Accessed October 1, 2017, https://images.apple.com/ environ However, more efficient remanufacturing may not nec- -ment/pdf/Apple_Environmental_Responsibility_Report_2017.pdf. essarily enhance the firm profit, because high reman- Apple Inc. (2017b) Recycling an Apple product is as easy as it is ufacturing efficiency is accompanied by high product good for the planet. Accessed October 1, 2017, https://www .apple.com/recycling/. durability, which discourages repeat customers from Atasu A, Sarvary M, Van Wassenhove L (2008) Remanufacturing as trading in and upgrading their used products. a marketing strategy. Management Sci. 54(10):1731–1746. From an environmental perspective, the value of Aviv Y, Pazgal A (2008) Optimal pricing of seasonal products in the trade-in remanufacturing depends on the intensity presence of forward-looking consumers. Manufacturing Service Oper. Management 10(3):339–359. of strategic customer behavior. With highly strategic Aviv Y, Wei M, Zhang F (2015) Responsive pricing of fashionable customers, this business practice decreases the unit products: The effects of demand learning and strategic con- environmental impact, but increases the production sumer behavior. Working paper, Tel Aviv University, Tel Aviv, quantities through the early purchase inducing effect. Israel. Bensako D, Winston W (1990) Optimal price skimming by a mono- Overall, trade-in remanufacturing may have a signif- polist facing rational consumers. Management Sci. 36(5):555–567. icant negative impact on the environment. Such neg- Cachon G, Swinney R (2009) Purchasing, pricing, and quick re- ative impact is intensified by moderate remanufactur- sponse in the presence of strategic consumers. Management Sci. ing efficiency, with which the firm benefits most from 55(3):497–511. Cachon G, Swinney R (2011) The value of fast fashion: Quick trade-in remanufacturing. Moreover, under intensive response, enhanced design, and strategic consumer behavior. strategic customer behavior, adopting trade-in reman- Management Sci. 57(4):778–795. ufacturing may decrease the customer surplus and Calcott P, Walls M (2000) Can downstream waste disposal policies social welfare. Hence, for a market with highly strategic encourage upstream “design for environment”? Amer. Econom. Rev. 90(2):233–237. customers, caution is needed when adopting trade-in Caldentey R, Liu Y, Lobel I (2017) Intertemporal pricing under mini- remanufacturing, because it could be detrimental to max regret. Oper. Res. 65(1):104–129. the environment and the society. With not-so-strategic Chen J (2015) Remanufacturing auto parts to receive 10 per- customers, however, trade-in remanufacturing leads to cent subsidy. China Auto Rev. (January 26), https://www a lower first-period production quantity and, thus, gen- .chinaautoreview.com/pub/CARArticle.aspx?ID=11672. Chen J, Esteban S, Shum M (2013) Why do secondary markets harm erally improves the environment. Our results indicate firms? Amer. Econom. Rev. 103(7):2911–2934. that customer purchasing behavior and remanufactur- Chevalier J, Goolsbee A (2009) Are durable goods forward-looking? ing efficiency play important roles in evaluating the Evidence from college textbooks. Quart. J. Econom. 124(4): impact of trade-in remanufacturing on the environ- 1853–1884. Chinese Ministry of Finance (2012) Measures for the collection and ment. Specifically, under intensive strategic customer administration of the funds for the recovery and disposal of behavior and moderate remanufacturing efficiency, waste electronic and electrical products [in Chinese]. Accessed trade-in remanufacturing creates a tension between May 5, 2017, http://faolex.fao.org/docs/pdf/chn149808.pdf. firm profitability and environmental sustainability; Cohen M, Lobel R, Perakis G (2016) The impact of demand uncertainty on consumer subsidies for green technology adoption. however, with (nearly) myopic customers, it generally Management Sci. 62(5):1235–1258. benefits both the firm and the environment. Cui Y, Duenyas I, Sahin O (2014) Should event organizers prevent To resolve the tension caused by trade-in remanufac- resale of tickets. Management Sci. 60(9):2160–2179. turing, we also study the government policy that bal- Debo L, Toktay LB, Van Wassenhove L (2006) Joint life-cycle dynamics of new and remanufactured products. Production Oper. Man- ances firm profit, customer surplus, and environmen- agement 15(4):498–513. tal impact. To achieve the socially optimal outcome, it Debo L, Toktay LB, Van Wassenhove L (2005) Market segmentation suffices for the government to employ a simple linear and product technology selection for remanufacturable prod- incentive scheme, which imposes either subsidies or ucts. Management Sci. 51(8):1193–1205. Department for Business, Innovation and Skills (2015) WEEE fund taxes on the production of both product generations as open to local authorities. Accessed May 20, 2016, http://www well as remanufacturing. Such policy counters strate- .wastecare.co.uk/weee-fund-open-to-local-authorities/. gic customer behavior and induces a social optimum Ferguson M, Toktay LB (2005) The effect of competition on recovery that does not depend on customer purchasing behav- strategies. Production Oper. Management 15(3):239–252. Fudenberg D, Tirole J (1998) Upgrades, tradeins, and buybacks. ior. In addition, the proposed subsidy/tax scheme can RAND J. Econom. 29(2):235–258. induce the firm to set the socially optimal remanufac- Galbreth M, Boyaci T, Verter V (2013) Product reuse in innovative turing efficiency. industries. Production Oper. Management 22(4):1011–1033. Zhang and Zhang: Trade-in Remanufacturing 616 Manufacturing & Service Operations Management, 2018, vol. 20, no. 4, pp. 601–616, © 2018 INFORMS

Geyer R, Van Wassenhove L, Atasu A (2007) The economics of reman- Recycler (2015) Scotland calls for remanufacturing subsidies. (Jan- ufacturing under limited component durability and finite prod- uary 19), https://www.therecycler.com/posts/scotland-calls uct life cycles. Management Sci. 53(1):88–100. -for-remanufacturing-subsidies/. Gu W, Chhajed D, Petruzzi N, Yalabik B (2015) Quality design and Savaskan R, Bhattacharya S, Van Wassenhove L (2004) Closed-loop environmental implications of green consumerism in remanu- supply chain models with product remanufacturing. Manage- facturing. Internat. J. Production Econom. 162(April):55–69. ment Sci. 50(2):239–252. Guide D, Van Wassenhove L (2009) The evolution of closed-loop Song I, Chintagunta P (2003) A micromodel of new product adoption supply chain research. Oper. Res. 57(1):10–18. with heterogeneous and forward-looking consumers: Applica- Hendel I, Lizzeri A (1999) Interfering with secondary markets. tion to the digital camera category. Quant. Marketing Econom. RAND J. Econom. 30(1):1–21. 1:371–407. Su X (2007) Inter-temporal pricing with strategic customer behavior. Hendel I, Nevo A (2013) Intertemporal price discrimination in Management Sci. 53(5):726–741. storable goods markets. Amer. Econom. Rev. 103(7):2722–2751. Su X (2009) Consumer returns policies and supply chain perfor- Johnson M, McCarthy I (2014) Product recovery decisions within mance. Manufacturing Service Oper. Management 11(4):595–612. the context of extended producer responsibility. J. Engrg. Tech. Su X (2010) Optimal pricing with speculators and strategic con- Management 34(October–December):9–28. sumers. Management Sci. 56(1):25–40. Kellogg C (2011) Amazon offers trade-in program for old Kindles. Los Su X, Zhang F (2008) Strategic customer behavior, commitment, and Angeles Times (October 21), http://latimesblogs.latimes.com/ supply chain performance. Management Sci. 54(10):1759–1773. jacketcopy/2011/10/amazon-offers-trade-in-program-for-old Swinney R (2011) Selling to strategic customers when product value -kindles.html. is uncertain: The value of matching supply and demand. Man- Levin Y, McGill J, Nediak M (2009) Dynamic pricing in the presence agement Sci. 57(10):1737–1751. of strategic consumers and oligopolistic competition. Manage- Thomas V (2011) The environmental potential of reuse. Sustainability ment Sci. 55(1):32–46. Sci. 6(1):109–118. Li J, Granados N, Netessine S (2014) Are consumers strategic? Struc- United States International Trade Commission (2013) The Year in tural estimation from the air-travel industry. Management Sci. Trade 2012: Operation of the Trade Agreements Program, 64th Report 60(9):2114–2137. (U.S. International Trade Commission, Washington, DC). Liang C, Çakanyildirim M, Sethi S (2014) Analysis of product rollover Van Ackere A, Reyniers D (1995) Trade-ins and introductory offers strategies in the presence of strategic customers. Management Sci. in a monopoly. RAND J. Econom. 26(1):58–74. 60(4):1033–1056. Waldman M (2003) Durable goods theory for real-world markets. Lobel I, Patel J, Vulcano G, Zhang J (2016) Optimizing product J. Econom. Perspect. 17(1):161–171. Wang Y, Chang X, Chen Z, Zhong Y, Fan T (2014) Impact of sub- launches in the presence of strategic consumers. Management Sci. sidy policies on recycling and remanufacturing using system 62(6):1778–1799. dynamics methodology: A case of auto parts in China. J. Cleaner Ma W, Zhao Z, Ke H (2013) Dual-channel closed-loop supply Production 74(July):161–171. chain with government consumption-subsidy. Euro. J. Oper. Res. Webster S, Mitra S (2007) Competitive strategy in remanufacturing 226(2):221–227.

and the impact of take-back laws. J. Oper. Management 25(6): Plambeck E, Wang Q (2009) Eﬀects of e-waste regulation on new 1123–1140. product introduction. Management Sci. 55(3):333–347. Xie J, Shugan S (2001) Electronic tickets, smart cards, and online Rao R, Narasimhan O, John G (2009) Understanding the role of trade- prepayments: When and how to advance sell. Marketing Sci. ins in durable goods markets: Theory and evidence. Marketing 20(3):219–243. Sci. 28(5):950–967. Xie S, Bai C (2010) New rules on disposal of waste electrical and Ray S, Boyaci T, Aras N (2005) Optimal prices and trade-in rebates electronic products. Lexology (November 30), http://www for durable, remanufacturable products. Manufacturing Service .lexology.com/library/detail.aspx?g=6e6da831-c189-4c81-82d4 Oper. Management 7(3):208–228. -3f6b45b3105b.