Next Article in Journal
Digital Transformation and ESG Performance—Empirical Evidence from Chinese Listed Companies
Next Article in Special Issue
The Analysis of Cultural Convergence and Maritime Trade Between China and Saudi Arabia: Toda–Yamamoto Granger Causality
Previous Article in Journal
Development and Validation of the New Environmental Locus of Control (NE-LOC) Scale: A Novel Measure of Internal, External, and Community Locus of Control for Sustainability
Previous Article in Special Issue
The Structure and Influencing Mechanisms of the Global Palm Oil Trade: A Complex Network Perspective
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Self-Recycling or Outsourcing? Research on the Trade-In Strategy of a Platform Supply Chain

School of Economics and Management, Jiangsu University of Science and Technology, Zhenjiang 212100, China
*
Author to whom correspondence should be addressed.
Sustainability 2025, 17(13), 6158; https://doi.org/10.3390/su17136158
Submission received: 14 May 2025 / Revised: 25 June 2025 / Accepted: 3 July 2025 / Published: 4 July 2025

Abstract

Trade-in programs have become a vital mechanism for promoting sustainable consumption and reducing negative impacts on the environment, gaining substantial support from branders, e-platforms, and consumers in recent years. Concurrently, the emergence of professional recyclers has provided firms with viable alternatives for the outsourcing of recycling processes. To investigate the optimal leadership and recycling model with respect to trade-in operations, this study examines the strategy selection in a platform-based supply chain under a resale model. A two-period game-theoretic framework is developed, encompassing four models: self-recycling and outsourcing models under the leadership of the brander or platform. The main findings are as follows: (1) In markets characterized by a low consumer price sensitivity, both branders and platforms tend to choose the self-recycling model to capture the closed-loop value. In contrast, in highly price-sensitive markets, both parties exhibit a preference for “free-riding” strategies. (2) Once the recycling leader is determined, adopting a self-recycling model can lead to a relative win–win outcome in high price sensitivity contexts. (3) With a short product iteration cycle, both the brander and platform should strategically lower their prices in the first period, sacrificing short-term profits to enhance trade-in incentives and maximize long-term gains. (4) When the brander leads the recycling process, they should consider reusing the resources derived from old products; however, in platform-led models, the brander can only consider reusing the recycled resources in a low price sensitivity market. This study provides strategic insights for the sustainable development of the supply chain through the analysis of a game between a brander and an e-commerce platform, enriching the literature on CLSCs through integrating trade-in leadership selection and the choice to outsource, offering theoretical support for dynamic pricing strategies over multi-period product lifecycles.

1. Introduction

In the aftermath of the COVID-19 pandemic, numerous countries have actively encouraged branders and e-platforms to implement trade-in programs as a means to stimulate consumer demand, revitalize economic growth, and promote the replacement of energy-consuming products. From an economic perspective, trade-in initiatives significantly boost consumer spending. For example, on Tmall the number of users purchasing home appliances and furniture through trade-in programs more than doubled year-on-year, resulting in over a 150% growth in related product sales. The consumption driven by trade-in programs further facilitates the out of high-energy-consuming products. From an environmental standpoint, trade-in improves resource efficiency and reduces the pollution caused by hazardous waste, thereby supporting the green transition of supply chains. A representative case is Apple, which utilizes aluminum recovered from old iPhones to manufacture new MacBook Air shells and reuses cobalt extracted from used batteries in the production of new batteries [1]. Moreover, platforms now integrate trade-in activities with environmental awareness campaigns. Tmall, for instance, links its trade-in system with the Ant Forest program, converting recycling behaviors into “green energy” points, which fosters consumer awareness and aligns well with the dual goals of environmental and economic sustainability. Trade-in initiatives, as a core element of circular economy practices, have gained substantial governmental support and have become an important instrument to boost consumption. For instance, the governments of European countries such as France provide cash incentives to consumers who exchange oil-powered vehicles for electric bicycles [2], with this policy having continued after the COVID-19 pandemic. In 2024, the Chinese government further expanded its trade-in subsidy scheme, covering home appliances, automobiles, and affiliated e-commerce platforms [3]. The introduction of such policies has led to a significant surge in the demand for trade-in services. For example, following the implementation of the 2024 trade-in subsidy program in China, the volume of recycled home appliances rose by 14.83% year-on-year [4]. However, this sharp increase in demand has placed considerable pressure on the capacities of recyclers. As the provision of trade-in services requires an investment in establishing and operating corresponding recycling systems, the escalating demand has pushed firms toward the transition to more intelligent and efficient systems, thereby also increasing their recycling expenditures. Consequently, the urgent need for novel trade-in cooperation models to mitigate operational costs has emerged. By selecting the optimal recycling approach, a closed-loop supply chain can effectively improve the resource utilization efficiency and the performance of the integral supply chain [5], achieving economic, environmental, and social benefits simultaneously [6].
In a trade-in program, consumers can return old products in exchange for rebates, effectively lowering the cost of purchasing new items and enhancing their purchase motivation. The returned products are subsequently either remanufactured or processed in an environmentally responsible manner, thus entering a formal recycling system and, in turn, creating new profit opportunities and enhancing the brand’s reputation [7]. With the rapid development of e-commerce, platforms such as Jingdong, Amazon, and Pinduoduo have actively participated in trade-in initiatives, attracting a broad base of consumers. In 2023, JD.com collected over 15 million used electronic devices. Since March of 2024, the participation in its trade-in program has increased by more than 150% year-on-year [8]. When a supply chain comprises a brander and an e-platform, both parties must weigh the costs and benefits associated with offering a trade-in service in order to determine the service’s leading party and an appropriate recycling approach. In a platform-based supply chain, the e-commerce platform directly connects with consumers, and its scale directly influences the size of the potential demand market. As a result, the platform is well-positioned to take the lead in the trade-in service and exerts a significant influence on the strategic choices available to the brand owner. Therefore, this study focuses on both the brander and the partnering platform, offering a new perspective on trade-in strategies and enhancing the study of platform leadership. In parallel, the surge in consumer demand has catalyzed the growth of professional recycling platforms such as “Aihuishou” and “Huishoubao”, which offer efficient and formal recycling channels to consumers. In practice, both branders and e-platforms have explored diverse recycling models. Some have opted to outsource the recycling process to third-party recyclers (3Ps) to avoid the fixed costs associated with in-house operations. For example, JD.com has established a comprehensive partnership with “Aihuishou”, which is responsible for product collection and disposal, and offers consumer rebates in the form of cash or platform vouchers. Apple has also partnered with “Aihuishou” in mainland China and works with 3Ps such as Alchemy in the U.K. and Australia. Drawing on real-world cases, we incorporate a third-party recycler (3P) into the CLSC, enriching the study of trade-in strategy selection within the CLSC framework.
As most branders tend to upgrade their products incrementally [9], consumers have the option to participate in trade-in programs when the next-generation product is released. For example, Apple launched the iPhone 16 one year after the release of the iPhone 15. Consumers who own an iPhone 15 or an earlier model can then decide whether to return their old device in exchange for a discounted price offered by Apple to purchase the new model. Therefore, we investigate trade-in strategies in a two-period product context under the resale model. The analysis is conducted from the perspectives of the dominant party in recycling and the recycling mode. This study attempts to answer the following questions:
  • Who should take the lead in offering a trade-in recycling service: the brander or the platform?
  • To maximize profits, should the dominant party choose to conduct self-recycling or outsource it to a third-party recycler (3P)?
  • During product iterations, how should dynamic pricing strategies be designed to effectively stimulate purchase demand?
To answer these questions, we build upon the existing literature through the implementation of two main novel features: (1) A two-period game-theoretic model between a brander and an e-commerce platform is constructed, analyzing their respective trade-in strategy choices under the resale model. (2) In period 2, a 3P is introduced as an alternative recycling model, and we examine whether the brander or the platform should outsource their recycling operations. Through deriving equilibrium outcomes across different models, the roles of the market potential, consumer price sensitivity, and salvage acquisition in shaping price and recycling model decisions are explored.
By analyzing strategic decisions within trade-in programs, this study provides actionable insights to align economic efficiency with environmental sustainability goals. The theoretical and practical contributions of this study are threefold: (1) It offers guidance on the optimal allocation of trade-in leadership between a brander and a platform. (2) It provides insights into whether outsourcing recycling to a 3P is preferable, exploring how to optimize resource circulation and waste management through trade-in programs. It also advances the development of green supply chain management. (3) The findings can support firms in implementing dynamic pricing strategies aimed at stimulating consumption and guiding environmentally friendly recycling behaviors, achieving a win–win outcome for both economic and environmental benefits.
The remainder of this paper is organized as follows. Section 2 reviews the relevant literature. Section 3 outlines the research problem and consumer demand analysis. Section 4 presents the formulated profit functions and details the sensitivity analysis. Section 5 compares the equilibrium outcomes across different models. Section 6 discusses the extended analysis. Section 7 concludes with managerial implications and directions for future research. Finally, proofs of all propositions and corollaries are presented in the Appendix A.

2. Literature Review

The research related to this study can be broadly categorized into three streams: (a) the strategy design for trade-in programs; (b) reverse channel selection with third-party (3P) participation; and (c) pricing decisions in a closed-loop supply chain (CLSC).

2.1. Research on Trade-In Strategy

The existing research on trade-in programs in conventional supply chains has primarily focused on the decision making of the brander (usually seen as a manufacturer) under varying influencing factors. Yi et al. [10] have analyzed the trade-in strategy selection in the automotive industry based on consumer preferences, while Bai et al. [2] examined rebate decision making under government subsidy schemes. Meanwhile, increasing attention has been given to the coordination mechanisms and allocation of responsibility among supply chain participants. Xiao [11] systematically compared optimal pricing and remanufacturing decisions under brander- and retailer-led trade-in models, concluding that both parties tend to “free-ride” and shift the responsibility of the trade-in provision to the other. Zhao et al. [12] investigated scenarios in which the retailer and brander lead the trade-in and found that a brander-led program enables a “win–win” situation and provides more benefits than when the program is retailer-led.
With the rise of e-commerce platforms in recent years, scholars have begun to explore the implementation of trade-in strategies within platform supply chains. For example, Wang et al. [13] focused on the roles of different sales models in e-commerce settings and showed that, under the resale model, conflicting interests often prevent a consensus on whether the trade-in leadership should be allocated to branders or retailers; meanwhile, under the agency model, equilibrium outcomes such as win–win cooperation or “free-rider” dynamics may emerge. Yu et al. [14] further investigated how differences in rebate-bearing parties affect supply chain operations and analyzed the comparative dominance of the platform versus the brander in administering rebate programs. Zheng et al. [3] noted that platform-led trade-in programs can reshape brander’s channel strategies and that low production costs may lead branders to favor agency models.
Bonina et al. [15] focused on the significance and impact of digital platforms. Based on the development of digital platforms and internet technologies, some studies have focused on the strategies of e-platforms and platform-related supply chains [16,17]. While these studies provide valuable insights into the economic implications of trade-in strategies, they largely overlook the role of 3Ps. The present research addresses this gap by explicitly incorporating the 3P participation and analyzing the selection of trade-in leadership and recycling models within a platform supply chain context.

2.2. Reverse Channel Selection with 3P Participation

In recent years, some studies have focused on issues relating to CLSC recycling strategies; for example, Pan and Lin considered cross-channel recycling [18], while Li et al. developed a CLSC network with remanufacturing [19]. The roles of 3Ps in CLSCs have also garnered increasing scholarly interest. Yang and Sun considered a CLSC with outsourcing and found that an eco-design may be beneficial for both the OEM and the third-party remanufacturer [20]. Li et al. [21] examined how professional recycling platforms optimize their profits through diversified trade-in approaches, such as cash-, refurbishment-, and replacement-based recycling. However, their analysis did not extend to upstream decision making within the supply chain. Cao et al. [22] studied brander commissioning strategies in the presence of 3P competition, demonstrating that a 3P intervention significantly alters the conditions under which branders opt to delegate trade-in responsibilities to retailers. Nevertheless, their model was restricted to a principal–agent framework between branders and retailers, lacking an examination of collaboration mechanisms involving 3P entities. Based on dual-channel sales, Zheng et al. [23] proposed three reverse channel structures: self-recycling, retailer outsourcing, and 3P recycling. With e-commerce platforms increasingly exerting influences across supply chains, recent studies have begun to explore their cooperation strategies with third-party recyclers. Li & Yi [24] examined trade-in and recycling decisions under different payment and recycling models of a B2C platform integrating recycling, showing the efficacy of the strategy using gift cards.
Although prior research has acknowledged the impact of 3P participation on trade-in decision making, most contributions have focused on static, single-period models and overlook dynamic pricing strategies across product lifecycle stages. This limitation motivated the present study to examine trade-in and pricing decisions under a two-period dynamic game framework.

2.3. Pricing Decisions in CLSC

Two-period game models have been widely applied in CLSC research to examine intertemporal decision making and strategic interactions among supply chain members. Ferrer and Swaminathan [25] demonstrated, through a multi-period remanufacturing framework, that the depreciation of the salvage value of returned products significantly influences branders’ pricing strategies. When the profitability of remanufacturing is sufficiently high, branders are incentivized to lower their initial prices to boost the sales volume, thereby increasing the availability of used products for future remanufacturing, even at the cost of short-term profits. This modeling framework has since been extended to dynamic pricing problems in trade-in contexts. For example, Yin and Tang [26] utilized a two-period model to analyze how advance payment schemes can be employed to stimulate intertemporal consumption under trade-in mechanisms. Quan et al. [27] constructed a two-period game between a brander and a retailer, comparing two trade-in models: direct recycling by the brander and outsourcing to the retailer. They highlighted that the brander generally prefers to retain control over trade-in services, although the resulting supply chain profitability varies across contexts. More recently, Hu and Tang [28] incorporated the behavioral heterogeneity of myopic and strategic consumers, revealing how the emergence of product-sharing markets reshapes the trade-in strategies of branders. However, their work remained confined to interactions between a single brander and end-consumers, without considering reverse channels from other actors.
Despite the rich literature on dynamic pricing in CLSCs, few studies have addressed the ownership and control of trade-in services in multi-agent platform supply chains. To address this gap, the present study introduces a dynamic game framework to compare brander- and platform-led trade-in models, explicitly examining pricing decisions and the allocation of trade-in leadership under a two-period structure.

2.4. Research Gap and Contributions

Building on the above review, Table 1 presents a comparative summary of the related literature and clarifies the research gaps addressed in this study. The key distinctions and contributions of this paper are detailed in the following paragraphs.
Existing studies have predominantly focused on traditional retail supply chains, where branders or retailers are the core decision-makers [10,11,12,22,23,26,27,28]. In platform supply chains, e-commerce platforms act simultaneously as sales agents and potential recycling operators [3,14,24]. The dual roles of platforms and their strategic implications for trade-in leadership remain underexplored. This study addresses this gap by modeling the interactions between branders and platforms, allowing for an analysis of the conditions under which each should assume trade-in leadership.
Additionally, within the resale-based platform supply chain framework, this study introduces a 3P into the trade-in decision-making process. In contrast to prior studies [23,24], this study explores the choice of the trade-in leader (either the brander or the platform) and whether recycling operations should be outsourced to an external 3P.
Moreover, most existing studies have been confined to single-period models and derived demand from static consumer utility functions. These approaches fail to capture the intertemporal relationship between the initial-period demand and second-period revenue. While some dynamic trade-in models exist [26,27,28], they are limited to brander-only settings. This study not only considers a platform-based supply chain structure but also incorporates a 3P and employs a two-period dynamic game model to provide a more accurate representation of the trade-in demand. Moreover, it is the first study to integrate the trade-in leadership selection between the platform and the brander, as well as the choice to outsource to a 3P within a unified analytical framework, constituting a novel contribution to the literature on multi-agent CLSC models.

3. Problem Description and Assumptions

3.1. Problem Description

Consider a platform-based supply chain consisting of an e-commerce platform (i.e., a platform) and one of its affiliated brand owners (i.e., a brander). The brander supplies products to the platform at a wholesale price w , after which the platform resells these products to consumers at price p under a reselling model. In this supply chain, the brander produces a specific type of durable product in period 1 and introduces a new-generation product in period 2. To promote the sale of new products, the supply chain implements a trade-in program in period 2. Under this program, consumers who return first-period products (hereafter referred to as “old products”) receive a rebate when purchasing second-period products (hereafter referred to as “new products”). Consequently, the market consists of two distinct consumer groups: new customers who have not previously purchased the product and existing customers who already own an old product. As the platform remains responsible for selling new products under the reselling model, it is assumed that all consumers participating in the trade-in program conduct their transactions through the platform. From the perspective of an existing customer, three options are available in period 2: (1) retain the old product and do not purchase the new product; (2) purchase the new product without participating in the trade-in program; or (3) return the old product and use the received rebate toward the purchase of the new product via the platform.
Based on the decision sequence, the game process considered in this study is divided into two stages. In the first stage, the brander first sets the wholesale price, w 1 , and the platform subsequently determines the resale price p 1 based on the wholesale price. In period 2, decision variables are adjusted according to the selected recycling model. The party leading the recycling process (the brander or the platform) also decides whether to outsource recycling to the 3P to reduce recycling costs. As the wholesale and resale pricing decisions in the first stage are independent of the recycling model, we examine four trade-in scenarios based on the recycling leader and whether recycling is outsourced in the second stage:
  • Model BB: The brander manages the recycling process by itself, encompassing both the handling of old products and the issuance of rebates to consumers, such as IKEA’s “Buy Back & Resell” program. In period 2, the decision variables for the brander are the wholesale price w 2 B B and the rebate amount r B B , while the platform is solely responsible for setting the selling price p 2 B B . The corresponding decision-making sequence is depicted in Figure 1a.
  • Model BT: The brander outsources the recycling to a 3P, similarly to Apple’s trade-in service in cooperation with Aihuishou, Alchemy, and so on. In this model, it is assumed that the brander pays a unit commission k B per item. The decision variables in period 2 are extended to include the wholesale price w 2 B T and the outsourcing commission k B . Subsequently, the 3P is responsible for recycling and processing the old products and providing the rebate r B T to consumers. Meanwhile, the platform remains solely responsible for determining the selling price p 2 B T . This decision sequence is illustrated in Figure 1b.
  • Model PP: The platform takes full responsibility for the recycling of old products and directly offers rebates to consumers, just as Amazon receives old products and resells them after processing them itself. As a result, in period 2, the brander only determines the wholesale price w 2 P P , whereas the platform simultaneously determines both the selling price p 2 P P and the rebate amount r P P . The decision-making sequence is illustrated in Figure 1c.
  • Model PT: Following the trade-in service of Jingdong, where the recycling process is outsourced to Aihuishou, the platform outsources the recycling of the old product to a 3P and pays a unit commission for each returned item. In period 2, the brander only sets the wholesale price w 2 P T , whereas the platform’s decision variables include both the selling price p 2 P T and unit commission k P . The 3P assumes responsibility for processing old products and providing rebates r P T to consumers. The decision sequence is illustrated in Figure 1d.
In the two-period model, δ denotes the discount rate and c i ( i = 1 , 2 ) represents the manufacturing cost of the product in each period. To simplify the analysis of the model results, we fix some minor parameters in order to focus on the key features of the strategies. Referring to existing studies [12,29], we assume that the discount rate is δ = 1 and standardize the values of the c i to zero. Additionally, the recycling costs associated with the trade-in service are classified into two categories: (1) Fixed costs incurred by establishing and operating the recycling system [30,31]. (2) The rebate offered to existing customers for the recycling of old products is considered part of the unit variable cost associated with the recycling process. The salvage value represents the remaining value of an old product after depreciation, which typically decreases with longer usage. As the trade-in occurs at the time of the new product’s release, it is possible to estimate the residual value of a product that has gone through one product cycle. Following the study of Quan et al. [27], we treat the salvage value s as a fixed unit, in order to simplify this model.

3.2. Customer Demand

Following the demand function assumptions reported in the studies of Quan et al. [27] and Ferrer & Swaminathan [25], this study assumes that consumers exhibit price sensitivity in each period and that each consumer demands at most one unit of the product (either new or used). In period i ( i = 1 , 2 ) , a potential consumer decides whether to become a new customer by purchasing a new product from the platform based on the prevailing selling price. As the selling price increases, the consumer demand correspondingly decreases. Accordingly, the demand function for new customers in each period under Model j ( j = B B , B T , P P , P T ) is specified as follows:
q i = α β p i j
where α is the market potential and β denotes the price sensitivity of new customers ( α > 0 , β > 0 ).
When considering a trade-in service, the consumer demand in period 1 serves as the foundation for demand in period 2, as the return of a used product creates the opportunity for consumers to exchange it for a new one. Consumers who purchased products in period 1 decide whether to participate in the trade-in program based on the selling price of the new product and the value of the trade-in rebate. In particular, ( p 2 j r j ) is the price that an existing customer would have to pay to trade-in an old product for a new one. The higher this price, the lower the demand for trade-ins. Moreover, as price sensitivity primarily reflects consumers’ responsiveness to changes in the product’s listed price, although the rebate lowers the payment for existing customers, it does not change their responsiveness to price variations. In the absence of other incentive mechanisms (e.g., gifts, loyalty points, etc.), the price sensitivity of existing customers is assumed to be the same as that of new customers. Therefore, the demand for trade-ins in the second stage is
q t = q 1 β ( p 2 j r j )
The relevant symbols and meanings used in this paper are listed in Table 2.

4. Model Analysis

To investigate the impacts of different recycling models under trade-in programs, we first examine the optimal pricing strategies under brander-led recycling models (i.e., Models BB and BT), followed by an exploration of platform-led recycling models (i.e., Models PP and PT). Subsequently, the effects of the market potential and salvage acquisition on the decision variables across different models are analyzed. All mathematical proofs are provided in Appendix A.

4.1. Brander-Led Models

4.1.1. Model BB

In Model BB, the brander and the platform engage in a two-period Stackelberg game. In period 1, the brander first sets the wholesale price w 1 B B . In period 2, the brander assumes responsibility for the recycling component of the trade-in program, providing a rebate r B B to participating consumers and processing the returned products for salvage value s , while simultaneously wholesaling new products at the wholesale price w 2 B B . In the trade-in process, the platform must satisfy both the demand from new customers (i.e., q 2 ) and the trade-in demand q t .
Accordingly, the brander and platform’s respective profits are as follows:
b B B = ( w 1 B B c 1 ) q 1 + δ [ ( w 2 B B c 2 ) ( q 2 + q t ) + ( s r B B ) q t F ]
p B B = ( p 1 B B w 1 B B ) q 1 + δ ( p 2 B B w 2 B B ) ( q 2 + q t )
The optimal equilibrium solution and profit are detailed in Table 3.

4.1.2. Model BT

Unlike Model BB, Model BT requires the brander to outsource the recycling process to a third-party recycler (3P). Accordingly, in addition to setting the wholesale price, the brander must also determine the unit commission k B paid to the 3P, while the trade-in rebate is provided directly by the recycler. As a result, the brander’s profit derives solely from the wholesale revenue of new products in each period while incurring additional commission expenses. The 3P could receive a commission as well as the salvage value by processing the old products, while the platform’s profit still comes from the resale.
The respective profits of the brander, 3P, and platform are as follows:
b B T = ( w 1 B T c 1 ) q 1 + δ [ ( w 2 B T c 2 ) ( q 2 + q t ) k B q t ]
t B T = δ [ ( k B + s r B T ) q t F ]
p B T = ( p 1 B T w 1 B T ) q 1 + δ ( p 2 B T w 2 B T ) ( q 2 + q t )
The optimal equilibrium solution and profit are detailed in Table 3.

4.2. Platform-Led

4.2.1. Model PP

In Model PP, the recycling process is managed by the platform. Accordingly, the brander’s profit consists solely of the wholesale revenue generated across the two periods. Meanwhile, the platform not only decides the selling price p i P P but also the unit rebate cost r P P when recycling old products, including a fixed processing cost F . The brander and platform’s respective profits are as follows:
b P P = ( w 1 P P c 1 ) q 1 + δ ( w 2 P P c 2 ) ( q 2 + q t )
p P P = ( p 1 P P w 1 P P ) q 1 + δ [ ( p 2 P P w 2 P P ) ( q 2 + q t ) + ( s r P P ) q t F ]
The optimal equilibrium solution and profit are detailed in Table 3.

4.2.2. Model PT

In Model PT, the brander only determines the wholesale price w i P T of the products in both periods. Similarly to the case for Model BT, the platform outsources the recycling process to the 3P. In addition to earning resale profits, the platform must also incur the unit commission cost k p associated with outsourcing. Hence, the respective profit maximization models for the brander, platform, and the 3P are as follows:
b P T = ( w 1 P T c 1 ) q 1 + δ ( w 2 P T c 2 ) ( q 2 + q t )
p P T = ( p 1 P T w 1 P T ) q 1 + δ [ ( p 2 P T w 2 P T ) ( q 2 + q t ) k P q t ]
t P T = δ [ ( k P + s r P T ) q t F ]
The optimal equilibrium solution and profit are detailed in Table 3.
Proposition 1.
To maximize the profits of both the brander and platform, there exist optimal wholesale prices, product selling prices, and rebate and commission strategies. The optimal solutions under Models BB, BT, PP, and PT are detailed in Table 3, respectively.

4.3. Equilibrium Analysis

Lemma 1.
When the brander takes the lead in the recycling process (i.e., Models BB and BT), analyzing the effects of the market potential and salvage value on the decision variables of the supply chain yields the following insights:
(1)
w 1 j α > 0 , w 2 j α > 0 , p 1 j α > 0 , p 2 j α > 0 , r j α > 0 , k B α > 0
(2)
w 1 j s < 0 , w 2 j s = 0 , p 1 j s < 0 , p 2 j s > 0 , r j s > 0 , k B s < 0 ( j = B B , B T )
When the brander leads the recycling process, a higher market potential allows both the wholesale and retail prices to increase, enabling larger commissions and rebates. Due to the strong demand resulting from a large market potential, considering the demand-oriented nature of the selling price, the platform is less pressured to reduce prices to attract consumers and can accept a higher wholesale price. Under outsourcing, raising the commission motivates the third-party recycler (3P) to offer greater rebates, encouraging participation in the trade-in program. Although the brander shares part of the recycling revenue, the growth in the trade-in demand boosts profits through increased new product sales. Therefore, in cooperation with large platforms, the brander can raise the wholesale price and either offer attractive rebates (in the self-recycling model) or increase commission incentives (in the outsourcing model).
When the salvage value increases, the brander lowers the wholesale price in period 1 to stimulate the initial demand, while the platform sets a lower price at first but raises it in period 2. Rebate amounts increase with the salvage value, whereas commission rates tend to decrease under outsourcing. As the salvage value is derived from products sold in period 1, the brander can cultivate a potential customer base for trade-ins by reducing the initial wholesale price and guiding the platform to adopt a market penetration pricing strategy. It is noteworthy that, when the product iteration cycle shortens, the salvage value of old products increases due to the reduced usage time. The resulting higher salvage value directly boosts the recycling revenue, thereby motivating the brander or the 3P to maximize trade-in participation by increasing the rebate value. This finding aligns with real-world practices; for instance, in Apple’s trade-in policy the rebate amount is significantly influenced by the model’s release date. Under otherwise identical conditions, consumers receive a USD 400 rebate when they trade-in with an iPhone 15, which is USD 110 higher than that for an iPhone 14 [32]. Furthermore, under the outsourcing model, the increase in the salvage value provides the 3P with intrinsic motivation to collaborate with branders in executing trade-ins, thereby making them more willing to accept a lower commission rate. Consequently, if the recycling process is outsourced to a 3P, the unit commission rate should be appropriately reduced under a short product iteration cycle.
Lemma 2.
In Model BB, the influence of the market potential on the profits of the brander and platform is as follows:
(1)
If  0 < β < 561 α 7 s , then  b B B α > 0 . Otherwise,if  β > 561 α 7 s , then  b B B α < 0 .
(2)
p B B α > 0 . Thus,the optimal platform profit increases as the market potential increases.
In Model BB, the brander’s profit is influenced by both the consumer price sensitivity and market potential. When the consumer price sensitivity is relatively low, the expansion of the market potential leads to an increase in demand, thus enhancing the brander’s profit. When consumers are highly sensitive to price changes, the brander’s profit decreases with the growth in the consumer demand. The underlying reason for this is that in the low price sensitivity market, consumers attach greater significance to non-price factors, such as the brand value and product quality. With an increasing consumer demand, the scale effect causes the marginal cost of the brander’s recycling system to decline. When the price sensitivity is relatively high, consumers focus more on price factors, and the demand is constrained. It becomes difficult for the brander to increase their profit by raising the wholesale price, and the expansion of the rebate cost under the condition of a large market potential leads to a reduction in profit.
Lemma 2(2) points out that, when the brander conducts self-recycling, the platform’s profit steadily grows as the market potential expands. Given that the platform does not bear the processing costs associated with trade-ins under this model, higher sales volumes are more beneficial to the platform’s revenue as the selling price increases (see Lemma 1). This conclusion also explains why major e-commerce platforms, such as JD.com and Taobao, are inclined to cooperate in implementing trade-in programs with famous branders and adopting dynamic pricing strategies, leveraging their extensive consumer base.
Lemma 3.
In Model BT, the influence of the market potential on the profits of the brander and platform is as follows:
(1)
If  0 < β < 7667 α 129 s , then  b B T α > 0 . Otherwise,if  β > 7667 α 129 s , then  b B T α < 0 .
(2)
p B T α > 0 . Thus,the optimal platform profit increases as the market potential increases.
Similarly to Model BB, in Model BT, when the consumer price sensitivity is relatively low, the brander’s profit is positively correlated with the market potential. Conversely, when the consumer price sensitivity is high, the brander’s profit exhibits a negative correlation with the market potential. Given consumers’ heightened sensitivity to price changes, the brander must increase their outsourcing commissions to incentivize recyclers to offer higher rebates, thereby attracting existing customers and consequently narrowing the brander’s profit margin.
Lemma 3(2) points out that, under the BT model, the platform’s profit increases as the potential consumer market expands. The platform consistently benefits from economies of scale under the BT model, aligning with the conclusions drawn under the BB model.
Based on the analysis of Lemmas 2 and 3, in markets characterized by low price sensitivity (e.g., professional equipment, high-end household appliances, and luxury goods), the scale effect under the brander-led model enhances the profits of both the brander and platform and promotes collaboration between them. Therefore, branders in these industries should partner with large e-commerce platforms to capitalize on their extensive consumer base. Conversely, in markets characterized by high price sensitivity (e.g., mobile phones, daily household appliances), or when dealing with platforms such as Temu that cater to price-sensitive customers, branders should avoid leading the recycling process.
Lemma 4.
When the platform takes the lead in the recycling process (i.e., Models PP and PT), analyzing the effects of the market potential and salvage value on the decision variables of the supply chain yields the following insights:
(1)
w 1 j α > 0 , w 2 j α > 0 , p 1 j α > 0 , p 2 j α > 0 , r j α > 0 , k P α > 0
(2)
w 1 j s > 0 , w 2 j s > 0 , p 1 j s < 0 , p 2 j s > 0 , r j s > 0 , k P s < 0 ( j = P P , P T )
Consistent with the brander-led case, when the platform leads the recycling, the prices set by the brander, platform, and 3P all rise with the market potential. As the brander’s profit now comes solely from wholesaling, considering the additional profit for the platform derived from the trade-in service, adjusting the wholesale price upward is acceptable.
Unlike the brander-led model, the salvage value now positively affects the wholesale price in both periods. This change arises because the brander can leverage its upstream position in the supply chain to moderately raise the wholesale price, thereby enhancing profits. The platform establishes a potential trade-in customer base through price penetration, and increasing the selling price in period 2 helps to offset part of the rebate or commission costs. Furthermore, when outsourcing recycling, the commission can be reduced for reasons consistent with Lemma 1, aligning with the commission strategy observed when the brander outsources.
Based on the preceding analysis, it is evident that (except for the wholesale price of the brander) the impacts of the salvage value and market potential on the pricing strategy do not change due to the shift in the dominant right. Platforms should continue using penetration pricing to attract trade-in participants. In practice, for products with short iteration cycles, such as mobile phones, e-commerce platforms like JD.com frequently utilize discount promotions via coupons during the new product launch phase but often exclude trade-in participants from other discounts, increasing profit margins and aligning with the results observed with our model.
Lemma 5.
In Model PP, the influence of the market potential on the profits of the brander and platform is as follows:
(1)
b P P α > 0 . Thus,the optimal brander profit increases as the market potential increases.
(2)
If  0 < β < μ 1 , then  p P P α > 0 . If  β > μ 1 , then  p P P α < 0 .
When the platform dominates the recycling process, the brander’s profit is consistently and positively correlated with the market potential in a stable manner. In contrast to the models where the brander leads the recycling process, in Model PP the brander is relieved of the rebate cost burden, thereby allowing its profit growth to remain unaffected by consumers’ price sensitivity.
In a market characterized by a low price sensitivity, the platform’s profit grows in tandem with the market potential; however, in one with high price sensitivity, an increase in the market potential results in a decline in the platform’s profit. Similarly to Lemma 2, the scale effect is pronounced in a low price sensitivity market, leading to a reduction in the platform’s marginal cost when recycling old products. Conversely, in a high price sensitivity market, the platform must increase its rebate costs to attract more trade-in demands, thereby causing its profit to decrease.
Lemma 6.
In Model PT, the influence of the market potential on the profits of the brander and platform is as follows:
(1)
b P T α > 0 . Thus,the brander’s optimal profit increases as the market potential increases.
(2)
If  0 < β < μ 2 , then  p P T α > 0 ; If  β > μ 2 , then  p P T α < 0 .
Lemma 6 reveals that in Model PT the brander’s profit grows in tandem with the market potential. Additionally, the effects of the consumer price sensitivity and market potential on the platform profit align with the findings regarding Lemma 5; specifically, in markets characterized by high price sensitivity, the platform raises commissions (for reasons similar to those of the brander with respect to Lemma 3), which leads to diminishing returns.
A combined analysis of Lemmas 5 and 6 shows that when serving consumers with low price sensitivity, the platform can achieve economies of scale and maintain effective collaboration with the brander. Conversely, when targeting consumers with high price sensitivity, both models under platform-led recycling present reduced profits as the market potential increases. In such cases, the platform should avoid taking the lead in recycling to mitigate profit losses.

5. Strategy Analysis

Through the model construction and solution, as presented in the preceding section, equilibrium solutions for two types of trade-in models under brander-led recycling (i.e., Model BB and BT) and platform-led recycling (i.e., Models PP and PT) were derived. In this section, we examine the pricing strategies and profit levels of supply chain members across different models to elucidate the pricing differences under each model, with an emphasis on identifying the optimal strategy when the consumer price sensitivity is high. Furthermore, through profit comparisons, this study reveals the effects of recycling dominance and the optimal model selection for the dominant party, providing an in-depth analysis of the optimal recycling models for different decision-making entities under various scenarios.

5.1. Brander-Led vs. Platform-Led

Corollary 1.
Comparing the optimal price and consumer demands in period 1 and period 2 among different leaders, the findings are as follows:
(i)
In period 1:
(1)
If  β > μ 3 , then  w 1 B B < w 1 P P  and  w 1 B T < w 1 P T .
(2)
The selling price is higher in period 1 under brander-led recycling (i.e., p 1 B B > p 1 P P , p 1 B T > p 1 P T ), whereas the consumer demand in period 1 is greater under platform-led recycling (i.e., q 1 B B < q 1 P P , q 1 B T < q 1 P T ).
(ii)
In period 2:
(1)
If  β > 42 α 61 s , then  w 2 B B < w 2 P P  and  w 2 B T < w 2 P T .
(2)
The selling price is higher in period 2 under platform-led recycling (i.e., p 2 B B < p 2 P P , p 2 B T < p 2 P T ).
(3)
In self-recycling,the rebate provided is greater under the brander-led model(i.e., r B B > r P P ), while in the outsourcing model,it is greater under the platform-led model(i.e., r B T < r P T ).
When consumers are highly price sensitive, the platform-led model results in a higher wholesale price but a consistently lower selling price across both periods, which boosts the consumer demand. This is because the platform can receive value from old products with the stronger cost control under Model PP, relieving the brander from the need to significantly lower the wholesale price.
In period 2, although the wholesale price remains higher under platform leadership, the selling price also increases due to the platform’s need to cover both wholesale and recycling costs, amplifying the double marginalization. Meanwhile, the trade-in rebate depends on the recycling method. According to Corollary 1(i), when the platform chooses self-recycling, the potential trade-in demand is higher, allowing the platform to rely on profits from both new product sales and old product recycling, thereby reducing the rebate level. Under the outsourcing mode, as the rebate is provided by the third-party recycler (3P), the 3P tends to offer a higher rebate under Model PT to attract more existing customers to participate in the trade-in and benefit from economies of scale.
According to the above analysis, for branders who produce products associated with high price sensitivity (i.e., smartphones, daily furniture, etc.), insisting on taking the lead in the recycling process may result in the excessive compression of the wholesale price. However, under a platform-led model, the brander should pay close attention to the platform’s pricing decisions to avoid overly high concessions that could weaken their profit margins. For platforms, regardless of the specific brand partnership, taking the lead in product recycling can stimulate greater consumer participation in trade-in programs through penetration pricing and effective cost control. Based on the findings in Corollary 1 and the optimal profits determined under each model, we derive the following conclusions.
Corollary 2.
The brander and platform’s optimal profits under the self-recycling models (i.e., Model BB and PP) are related as follows:
(1)
If  0 < s < s 1 , when  0 < β < β 1 min , Π b B B > Π b P P . Otherwise,if  0 < s < s 1 , when  β > β 1 min , Π b B B < Π b P P .
(2)
If  0 < s < s 1 , when  0 < β < β 2 min , Π p B B < Π p P P . Otherwise,if  0 < s < s 1 , when  β > β 2 min , Π p B B > Π p P P .
Corollary 2 compares the game players’ profits under different self-recycling models. The results indicate that when the salvage value of the old product is lower than a certain threshold, the brander’s profit under Model BB is greater in a market characterized by low price sensitivity; meanwhile, in one with a high price sensitivity, the brander’s profit under Model PP is greater. Therefore, the platform’s optimal model selection is opposite to that of the brander. To more explicitly elucidate this conclusion, this study visually illustrates the model preferences of the brander through Figure 2a and the platform through Figure 2b. In markets characterized by low price sensitivity (i.e., luxury goods, professional equipment, high-end household appliances, etc.), consumers pay more attention to non-price factors. Therefore, the brander can maintain a relatively high wholesale price (see Corollary 1) and obtain closed-loop profits and control costs under Model BB. Similarly, the platform can enhance its profitability in such markets by engaging in self-recycling (see Lemma 5). However, in markets characterized by high price sensitivity (i.e., smartphones, daily furniture, etc.), consumers are sensitive to price changes. Both the brander and platform need to reduce their prices or increase the rebate value to attract customers. At this time, the salvage value of recycling may not be sufficient to cover the cost. In this case, the brander and the platform are more inclined to let the 3P recycle and become a “free rider.” This is in line with the conclusions of Lemma 2 and Lemma 5, indicating that both of the game players need to avoid taking the lead in recycling.
Corollary 3.
The brander and platform’s optimal profits under the outsourcing recycling models (i.e., Models BB and PP) are related as follows:
(1)
Π b B T > Π b P T , and the Brander prefers Model BT to PT.
(2)
Π p B T < Π p P T , and the Platform prefers Model PT to BT.
Corollary 3 systematically compares the profit levels of the brander and platform under the outsourcing models, revealing that both parties achieve optimal profitability when assuming a leadership role. As illustrated in Figure 3a, the preference for specific models by the brander is clearly delineated, while Figure 3b shows the platform’s preference. This phenomenon can be attributed to the fact that the leading party has the authority to govern critical decisions, including pricing strategies and outsourcing volumes. Commission adjustments enable the alignment of the 3P’s rebate strategies with the leader’s strategic objectives. For instance, increasing commission rates can motivate the 3P to enhance customer rebates, while the associated costs can be effectively transferred via pricing mechanisms to optimize the overall profitability.
The above results from Corollaries 2 and 3 expose the following implications. First, when catering to customers with low price sensitivity, both the brander and the platform are inclined towards self-recycling, resulting in conflicts of interest; meanwhile, when consumers are more sensitive to price variations, both parties prefer to “free-ride”. Additionally, in the outsourcing context both parties achieve higher profits when they assume the role of the dominant party, also resulting in conflicts of interest. This phenomenon accounts for why, during cooperation between enterprises such as Apple and Jingdong, both retain their own dominant recycling channels.
Corollary 4.
The supply chain system’s optimal profit between the brander- and platform-led models satisfies the following relationship:
(1)
If  Π B B < Π P P , the supply chain prefers Model PP over BB.
(2)
If  β > β 3 max , then  Π B T < Π P T  and the supply chain prefers Model PT over BT.
Under the self-recycling model, the total supply chain profit is higher when the platform takes the lead (i.e., Model PP). Similarly, in outsourcing recycling models within highly price-sensitive markets, the supply chain also tends to prefer platform-led recycling (i.e., Model PT). Figure 4a provides a more intuitive illustration of the supply chain’s trade-in leader preferences between Model BB and PP, while Figure 4b points out the optimal trade-in leader for the supply chain in outsourcing models. As detailed in Corollary 1, platform leadership stimulates a greater initial demand, allowing Model PP to achieve economies of scale and reduce the marginal cost of recycling, thereby enhancing the overall supply chain profitability. In the outsourced recycling scenario, the 3P offers a higher rebate under Model PT, which encourages greater consumer participation in trade-in programs, further boosting supply chain profits. For industries such as home furniture, smartphones, and computers, both the brander and platform serve highly price-sensitive consumers. To maximize the overall supply chain profit and mitigate double marginalization effects, platform-led models should be prioritized. However, as revealed in Corollaries 2 and 3, there exists a conflict between the brander and platform regarding the control over the recycling process. Therefore, to align the interests of both parties, coordinated incentive mechanisms (e.g., compensation schemes) should be set.
Despite having separately examined the optimal choice of dominance in both the self-recycling and outsourcing models, decision-making processes cannot force the dominant entity to strictly adhere to either self-recycling or outsourcing recycling. The model selected by the dominant entity may lead to the overall profit of the supply chain falling short of its optimal potential. Consequently, the subsequent analysis delves further into whether the dominant entity should opt for the outsourcing recycling model from a strategic perspective.

5.2. Self-Recycling vs. Outsourcing

Corollary 5.
Comparing the optimal price and consumer demand in period 1 among self- and outsourcing recycling models, the findings are as follows:
(i)
In period 1:
(1)
If  β > μ 4 , then  w 1 B B < w 1 B T  and  w 1 P P < w 1 P T .
(2)
The selling price in period 1 is higher in outsourcing recycling (i.e., p 1 B B < p 1 B T , P 1 P P < P 1 P T ), whereas the consumer demand in period 1 is greater in self-recycling (i.e., q 1 B B > q 1 B T , q 1 P P > q 1 P T ).
(ii)
In period 2:
(1)
If  β > 499 α 951 s , then  w 2 B B = w 2 B T  and  w 2 P P > w 2 P T .
(2)
If  β > 499 α 951 s , then  p 2 B B > p 2 B T  and  p 2 P P > p 2 P T .
(3)
If  β > 62 α 303 s , then  r B B > r B T  and  r P P > r P T .
Comparing the first-period pricing under different recycling approaches with the same dominant party, it becomes evident that in markets characterized by a high price sensitivity, the wholesale price under the outsourcing model (i.e., Models BT and PT) is higher. In contrast, unaffected by consumer price sensitivity, the selling price is consistently higher in the outsourcing model. As a result of the higher selling price, the initial consumer demand is greater under the self-recycling model (i.e., Models BB and PP). The same situation between the wholesale and selling price is due to the fact that, when serving consumers who are sensitive to price changes, the shift in the recycling cost creates more space for both the brand and the platform to raise their prices appropriately.
In terms of the second-period pricing, the findings indicate that when the brander dominates the recycling process, the wholesale price remains unaffected by the recycling approach. However, unlike in period 1, the wholesale price under Model PP is higher in highly price-sensitive markets. Furthermore, both the platform’s retail price and the rebate are greater in the self-recycling model (i.e., Models BB and PP). This difference arises because increasing the rebate is a crucial strategy for encouraging existing customers to participate in trade-in programs during the second period. However, under the outsourcing model, the rebate is provided by a 3P, whose profit is solely derived from processing old products. This limits the rebate level and consequently requires both the brander and the platform to lower their prices to stimulate consumption.
The aforementioned analysis demonstrates that although the outsourcing model helps to reduce recycling costs, it weakens the revenue in the second period. In industries characterized by high consumer price sensitivity, firms should carefully balance recycling costs against profits. Therefore, based on the goal of profit maximization, we further analyzed the optimal recycling approach for members of the supply chain.
Corollary 6.
The brander and platform’s optimal profits under brander-led models (i.e., Models BB and BT) are related as follows:
(1)
When  0 < α < α 0 , if  0 < β < β 4 min  or  β > β 4 max , then  Π b B B > Π b B T ; if  β 4 min < β < β 4 max , then  Π b B B < Π b B T ; otherwise,when  α > α 0 , then  Π b B B > Π b B T .
(2)
If  0 < β < β 5 max , then  Π p B B < Π p B T ; otherwise,if  β > β 5 max , then  Π p B B > Π p B T .
Corollary 6(1) demonstrates that in the brander-led models, when the market potential is relatively small, the brander’s profit can be maximized through self-recycling in markets characterized by either a high or extremely low price sensitivity; otherwise, the outsourcing model should be adopted. However, in large-scale markets, the brander’s decision regarding outsourcing is unaffected by the consumer price sensitivity, and self-recycling consistently achieves the highest profitability. To provide a clearer illustration of these findings, Figure 5a presents the influences of various factors on the model preferences of the brander. Specifically, under a low market potential, self-recycling enables the brander to maximize their closed-loop revenue. Nevertheless, as the consumer sensitivity to rebate prices increases, outsourcing to professional recyclers by the brander can optimize the trade-in demand while maintaining a constant wholesale price (as indicated by Corollary 5), thereby reducing marginal costs. Moreover, when consumers exhibit extreme price sensitivity, self-recycling allows the brander to avoid paying premiums to the 3P, thus offering greater benefits to consumers and enhancing participation in the trade-in program. In scenarios with a high market potential, the brander significantly benefits from economies of scale (see Lemma 1), especially in terms of the wholesaling revenue, leading to reduced marginal costs. Thus, when cooperating with a large platform, famous branders should adopt self-recycling to further capitalize on their recycling profits.
As shown in Figure 5b, in brander-led models, the platform could achieve the maximum profit in Model BT in low price sensitivity markets and in Model BB in high price sensitivity markets. Given that the wholesale price in period 2 remains constant across both brander-led models (see Corollary 5), the initial wholesale price emerges as the pivotal determinant of the platform’s profitability. Under the outsourcing model, the brander lowers the initial wholesale price to enhance their future earnings, thereby promoting growth in the platform’s profit. Nevertheless, when consumers exhibit a heightened sensitivity to price fluctuations, the initial wholesale price under Model BB is comparatively lower (as evidenced by Corollary 5). Consequently, under such circumstances, the platform exhibits a stronger preference for the brander’s self-recycling model.
Based on the comprehensive analysis presented above, it can be concluded that when a high-end brand or a daily necessities brand collaborates with a large e-commerce platform, if the brander selects the recycling model under its leadership, the self-recycling model should be adopted to optimize profitability. Additionally, drawing on the findings from Corollaries 2 and 3, although branders may be reluctant to assume the leading role in high price sensitivity markets, the platform—driven by a “free-rider” mentality—tends to delegate leadership to the brander during the model selection. In such cases, the brander should opt for self-recycling to achieve a relatively win–win outcome. When consumers exhibit moderate price sensitivity, particularly on small- and medium-sized platforms, the brander should choose to outsource in order to pursue profit maximization, despite the potential misalignment with the platform’s expectations. Therefore, for brands with a high user stickiness, even though consumers are attentive to price changes but may still make purchase decisions, branders should adopt the 3P outsourcing model; notably, this is consistent with Apple’s trade-in program observed in practice.
Corollary 7.
The brander and platform’s optimal profits under platform-led models (i.e., Models BB and BT) are related as follows:
(1)
If  0 < β < β 6 max , then  Π b P P < Π b P T ; otherwise,if  β > β 6 max , then  Π b P P > Π b P T .
(2)
When  α < α 1 , if  0 < β < β 7 min  or  β > β 7 max , then  Π p P P > Π p P T ; otherwise,if  β 7 min < β < β 7 max , then  Π p P P > Π p P T .
Corollary 7(1) compares the brander’s profit under platform-led models and reveals that the brander only achieves a higher profit in low price sensitivity markets under the outsourcing model (i.e., Model PT). Otherwise, the brander prefers Model PP, as shown in Figure 6a; this is because, when consumers are more sensitive to the price, the brander can charge a higher wholesale price for new products under Model PP (see Corollary 5), thereby generating greater profit.
Comparing the platform’s own profit under platform-led models, similarly to Corollary 6(1), we can find that when the potential market demand is relatively low, the platform should adopt self-recycling only when the consumer price sensitivity is either extremely low or extremely high; otherwise, the recycling process should be outsourced to a 3P. Similarly to the brander-led case, when the market potential is high, the platform prefers self-recycling. Figure 6b illustrates the model preferences of the platform clearly. This is because, when serving consumers with low price sensitivity, there is no need to increase rebate expenditures, making self-recycling the most advantageous choice due to the additional recycling revenue. In contrast, when consumers are sensitive to price changes, outsourcing helps to reduce the cost burden associated with recycling. However, as the market potential expands, the advantage of the 3P outsourcing weakens. The increased demand lowers the platform’s marginal recycling cost and enhances resale profits (see Lemma 4), making self-recycling the preferred strategy. Furthermore, when consumers are extremely price-sensitive, the platform should also adopt self-recycling for the reasons discussed with respect to Corollary 6(1).
Combining the above results with Corollaries 2 and 3, it can be inferred that when consumers are relatively price-sensitive, the brander tends to delegate the leadership of recycling decisions to the platform. Therefore, in such markets the platform should adopt the outsourcing model. The platform should only adopt a self-recycling model when consumers are extremely price-sensitive or when the potential market size is large, in which case a win–win outcome can be achieved. This also explains why major e-commerce platforms such as JD.com and Suning tend to collaborate with 3Ps.
Corollary 8.
The supply chain system’s optimal profit between self-recycling and outsourcing models satisfies the following relationship:
(1)
Π B B > Π B T , and the supply chain prefers Model BB to BT.
(2)
Π P P > Π P T , and the supply chain prefers Model PP to PT.
When either the brander or the platform serves as the dominant party, the overall supply chain profit is higher when that party adopts self-recycling. To reveal this finding clearly, Figure 7a shows the supply chain’s optimal choice under brander-led models, and Figure 7b shows the choice under platform-led models. As inferred from Corollary 5, compared with the outsourcing model, self-recycling results in a larger number of existing customers and greater potential trade-in demand, thereby generating higher profits derived from the recycling of old products. Therefore, to maximize overall supply chain profits, the dominant party should prioritize self-recycling. Moreover, beyond daily necessities, self-recycling should likewise be adopted when considering products for which consumers are less price-sensitive (i.e., professional equipment, luxury goods, etc.). For instance, when the luxury brand “Gucci” partnered with the e-commerce platform “The RealReal” for resale operations, the platform opted for self-recycling [33]. Synthesizing insights from Corollaries 6 and 7 it is evident that, under certain conditions, the dominant party may prefer the outsourcing model, which can prevent the supply chain from achieving its optimal profit level. Nevertheless, when consumers exhibit high price sensitivity or when there is a significant potential demand for new products, the self-recycling model enables the supply chain to achieve a relative win–win outcome.

5.3. Optimal Model Selection

Corollary 9.
When the profits of the brander and the platform under the four recycling models are compared, the following conclusions can be obtained:
(1)
If  β > max { β 8 min , β 1 min } , then  Π b P P > max { Π b B B , Π b B T , Π b P T } .
(2)
If  β > max { β 9 min , β 2 min } , then  Π b B B > max { Π b B T , Π b P P , Π b P T } .
Corollary 9 indicates that when the consumer price sensitivity is high, both the brander and the platform prefer to act as “free riders,” which is consistent with Corollaries 2 and 3. To further illustrate this result, Figure 8a presents the brander’s model preference, while Figure 8b shows that of the platform—both based on their respective profit maximization objectives. As shown in the figure, in high price sensitivity contexts, the profits of both the brander and the platform are maximized when the counterpart undertakes self-recycling. However, if the counterpart adopts an outsourcing model, neither party is able to achieve maximum profits among the four models.
Furthermore, drawing on Lemmas 2 and 3 and Lemmas 5 and 6, it is evident that becoming the dominant recycler is not conducive to the profit growth of the party itself when facing highly price-sensitive consumers. Accordingly, based on the conclusions of Corollaries 6 and 7, when forced to assume the role of the dominant recycler, the outsourcing model should be considered if the market potential is small. In contrast, when the consumer price sensitivity is high, the dominant recycler should adopt the self-recycling model to avoid incurring additional costs associated with outsourcing commissions. In the case of the large market potential, the dominant recycler should also opt for self-recycling, which aligns with the expectations of the model selector. This pattern explains why major e-commerce platforms such as Tmall tend to maintain self-recycling channels. In addition, this relative win–win model selection situation further clarifies why platforms such as JD.com and major home appliance brands have prioritized partnerships with brands which are capable of dismantling and processing old products when establishing “Renewal Alliances” [8].

6. Extension

6.1. Reusing Salvage Value

Building upon the original platform-based supply chain models, this section further considers the brander’s reuse of the salvage from old products. Specifically, when the brander assumes leadership in the recycling process and selects the outsourcing model, it procures the processed salvage of old products from the 3P for the subsequent reproduction. When the platform leads the recycling process, the brander does not intervene in the platform’s decision regarding whether to outsource or manage recycling internally. Consequently, the brander acquires the salvage from the platform under all circumstances. Throughout this process, the brander retains the decision-making authority over the transfer pricing of the salvage. The decision sequence remains consistent with the original models.
The optimal equilibrium prices and the computations relating to the corollaries for the extended models are provided in Appendix A.
Corollary 10.
When the brander contemplates the acquisition of the salvage, the profits of the brander and the platform under different models can be compared to reveal that
(1)
If  0 < α < α 0  and  β > β 4 max , or  α > α 0 , then  Π b P P * > max { Π b B B * , Π b B T * , Π b P T * } .
(2)
If  β > β 12 max , then  Π b B B * > max { Π b B T * , Π b P P * , Π b P T * } .
Comparing the profits of the brander and the platform under the four different models, it can be observed that when the brander considers acquiring the salvage, Model BB enables both parties to achieve a win–win outcome in high price sensitivity markets. Specifically, in such markets, Model BB leads to profit maximization for both the brander and the platform. As the brander must pay a transfer cost for the salvage to the platform under platform-led models, brander-led models are preferred as they allow for more direct cost control (i.e., Models BB and BT), as illustrated in Figure 9a. Moreover, following the same reasoning as in Corollary 6, the brander should adopt self-recycling in high price sensitivity markets. As for the platform, its model preferences are shown in Figure 9b. When serving highly price-sensitive consumers, the platform still prefers to act as a “free rider” for reasons consistent with Corollary 3. In such a market, the brander may be reluctant to lead the recycling process if there is no acquisition of the salvage (see Corollary 9). However, the situation changes when the platform proposes selling the reusable portion of the salvage to the brander. To maximize its own profit, the brander then chooses to lead the recycling process and adopts a self-recycling strategy. This ultimately results in a win–win outcome for both parties.
Corollary 11.
In contrast to the original model presented above, through analyzing the profits of supply chain members under different modes, the following conclusions can be reached:
(1)
Π b B B * = Π b B B Π p B B * = Π p B B .
(2)
Π b B T * = Π b B T Π p B T * = Π p B T .
(3)
If  0 < β < β 10 max , then  Π b P P * > Π b P P , while  Π p P P * < Π p P P .
(4)
If  0 < s < s 1  and  0 < β < β 11 min , then  Π b P T * > Π b P T , while  Π p P T * < Π p P T .
According to Corollary 11(1) and (2), when the brander leads recycling whether or not it acquires the salvage from a 3P does not affect the profits of either party. Under outsourcing, the transfer price for the salvage offsets the commission cost, leading to equivalent outcomes. In the context of self-recycling, no such transfer occurs. Therefore, when the brander leads the recycling, it should consider utilizing the processed old products in production to promote resource reuse and maximize its profits.
However, when the platform leads recycling the situation changes. In markets characterized by low price sensitivity, acquiring the salvage helps the brander to increase its profit. Figure 10a and Figure 11a illustrate the brander’s preferences regarding the acquisition and reuse of the salvage under platform-led self-recycling and outsourcing models, respectively. However, the transfer price offered by the brander is limited and, as the wholesale price is affected by the transfer price, the platform’s profit margin is compressed. Figure 10b and Figure 11b present the corresponding preferences of the platform, which may be insufficient to offset the platform’s recycling costs. As a result, the platform is generally reluctant to sell recycled resources to the brander. In contrast, in markets characterized by high price sensitivity, branders benefit more by using price reductions to drive sales, without considering the acquisition of the salvage.
These insights explain why, in practice, branders such as Apple outsource recycling but still reuse old product components. When the platform leads acquiring salvage may benefit the brander under specific market conditions but carries the risk of lowering the platform cooperation. It is worth noting that, even though the brander’s acquisition of the salvage under platform-led models is detrimental to the platform’s profitability, Figure 9b shows that in low price sensitivity markets the platform can still maximize its profit through the effective control of costs and revenues. Therefore, when the platform acts as the model selector and chooses to lead the recycling process, the brander may still acquire the salvage for reuse in low price sensitivity markets. This is particularly relevant for high-end or professional product lines, where purchasing salvage from their resale partners can also strengthen the brander’s profit and reputation.

6.2. Revenue-Sharing Mechanism

Considering that under the self-recycling model the brander or the platform can obtain a salvage value revenue from recycling, and there exists a conflict of interest between the two parties, this section further examines a revenue-sharing mechanism for self-recycling to explore coordination under a profit conflict.
In Model BB, the profits of the two players are
b B B m = ( w 1 B B m c 1 ) q 1 + δ [ ( w 2 B B m c 2 ) ( q 2 + q t ) + μ ( s r B B m ) q t F ]
p B B m = ( p 1 B B m w 1 B B m ) q 1 + δ [ ( p 2 B B m w 2 B B m ) ( q 2 + q t ) + ( 1 μ ) ( s r B B m ) q t ]
In Model PP, the profits of the two players are
b P P m = ( w 1 P P m c 1 ) q 1 + δ [ ( w 2 P P m c 2 ) ( q 2 + q t ) + ( 1 μ ) ( s r P P m ) q t ]
p P P m = ( p 1 P P m w 1 P P m ) q 1 + δ [ ( p 2 P P m w 2 P P m ) ( q 2 + q t ) + μ ( s r P P m ) q t F ]
where μ represents the proportion of the recycling revenue allocated to the recycling leader.
Observation.
In contrast to the original model presented above for each player, the following conclusions can be reached:
Figure 12a,b illustrate that, under the brander-led recycling model, implementing recycling revenue sharing can lead to a win–win outcome for both parties. By sharing the recycling revenue, the platform gains a portion of the additional revenue, which increases its willingness to cooperate and mitigates the conflict observed in Corollaries 2 and 9, thereby achieving a Pareto improvement.
In contrast, as shown in Figure 12c,d, when the platform leads the recycling process, revenue sharing does not simultaneously enhance the profits of both the brander and the platform. Specifically, regardless of the proportion of the recycling revenue allocated to the platform, its profit seldom exceeds the level without revenue sharing. In fact, sharing the revenue may require the platform to relinquish part of its earnings, leading to a diminishing marginal profit. However, even if the platform itself does not benefit directly, when it receives a sufficiently large portion of the recycling revenue, the brander’s profit may still improve. According to Corollary 9, the platform demonstrates a preference for the PP model in some conditions. Therefore, by designing a reasonable sharing ratio, the platform can use salvage value sharing to incentivize the brander’s cooperation.

7. Conclusions and Suggestions

7.1. Conclusions

This study constructed models of a supply chain in which a brander and an e-commerce platform engage in a Stackelberg game, allowing for a systematic analysis of their respective strategy choices for leading trade-in operations under the resale model. Based on an analysis of the impacts of the market potential, consumer price sensitivity, and the salvage of old products, the main research conclusions are as follows:
(1)
The consumer price sensitivity is an important factor influencing the choice of the dominant party. Conflicts and “free-riding” behaviors arise between the brander and the platform regarding the recycling model selection in markets characterized by different levels of price sensitivity. In low price sensitivity markets, both branders and platforms tend to choose self-recycling to obtain closed-loop benefits, while in high price sensitivity markets, both parties prefer that the other party take the lead in recycling.
(2)
The consumer price sensitivity and market potential jointly affect the dominant party’s choice of the recycling model. When the consumer price sensitivity is extremely high or extremely low, the dominant party should adopt self-recycling. At moderate sensitivity levels, outsourcing becomes the preferred option. However, when the market potential is high, adopting self-recycling can achieve a relative win–win outcome for both parties in the supply chain.
(3)
During the trade-in process, both the brander and platform can lead in recycling. The trade-in leadership significantly influences the brander’s pricing system. The platform consistently maintains a penetration pricing strategy, lowering product prices in period 1 to forgo part of its short-term profits and enhance future gains during the trade-in phase. When the brander leads recycling, it should cooperate with the platform by lowering the wholesale price in period 1, especially given the accelerating nature of product iteration cycles. Conversely, when the platform leads recycling, the brander can appropriately raise the wholesale price to optimize its own profitability.
(4)
When the brander leads recycling, the acquisition of the salvage does not affect the profits of supply chain members. While self-recycling by the brander could bring a win–win outcome in high price sensitivity markets, in platform-led models, branders can only profit from the acquisition of the salvage in low price sensitivity markets.

7.2. Management Insights

Based on the aforementioned findings, the following managerial insights can be inferred:
(1)
For both the brander and the e-commerce platform, it is essential to consider the influence of the consumer price sensitivity when formulating trade-in strategies. For branders involved in luxury goods, high-end products, or professional equipment, customers typically place a greater emphasis on the quality or brand value rather than the product’s price itself, therefore exhibiting low price sensitivity. Under these circumstances, both the brander and the platform can achieve higher closed-loop profits by adopting a self-recycling model for trade-in operations. In contrast, for branders involved in the daily necessities sector, consumers are more inclined to compare prices across multiple sources and are highly sensitive to price changes. In such cases, the brander and the platform must carefully deliberate the allocation of the leadership in recycling to avoid conflicts and may maximize their overall profits through compensation mechanisms or collaborative approaches.
(2)
Under an economic downturn the dominant party responsible for recycling should comprehensively consider the consumer price sensitivity and market size when selecting the recycling mode. In industries involving daily necessities, where consumers exhibit a high sensitivity to price changes, the dominant party should adopt a self-recycling model. Conversely, for branders with high customer loyalty, where the consumer price sensitivity is relatively low, either the brander or the collaborating platform should outsource recycling activities to save costs. These recommendations are particularly relevant for branders and platforms with limited market potential, for which the achievable economies of scale are constrained. When selecting a trade-in model, they should carefully evaluate the trade-off between the benefits and costs of recycling in light of their market potential. For large e-commerce platforms and well-known branders, a broad customer base, strong brand recognition, and a high product replacement frequency collectively contribute to substantial market potential. These branders and platforms are better positioned to establish self-recycling systems. Therefore, adopting a self-recycling model is recommended in such contexts.
(3)
Pricing strategies should be reasonably adjusted based on the market demand and product iteration under different recycling leadership models to balance trade-in costs and profits, thus optimizing the overall supply chain efficiency. When the brander leads recycling, both the brander and platform should appropriately lower their prices in period 1 (especially under a short product iteration cycle), in order to attract more consumers to participate in the trade-in program. Simultaneously, the brander should actively consider reusing the salvage derived from traded-in products as part of reproduction processes in order to achieve resource reuse, enhance the supply chain’s sustainability, and attract more customers to replace their high energy-consuming products. When the platform leads recycling the brander should consider moderately increasing the wholesale price under a short product iteration cycle, while platforms should maintain a penetration pricing strategy in period 1. Furthermore, when reselling high-end products, luxury brands, or professional equipment, the brander should acquire the salvage of old products from the platform for reuse in production, thus enhancing the brand’s reputation.

7.3. Limitations and Future Studies

The research conclusions of this article provide a theoretical basis for branders and platforms regarding the selection of a trade-in strategy, which is helpful for enterprises aiming to optimize their recycling model and pricing strategy under different market environments, in order to improve the overall efficiency of the supply chain. However, this article still has certain limitations. The consequent research prospects are presented below:
First, we examined the strategic interactions among three key stakeholders: a brander, a platform, and a third-party recycler. However, real-world supply chains often exhibit more complex dynamics, such as multi-tiered recycling networks, government involvement, and competition among multiple branders, which are not captured in the presented model. Second, we assumed that consumers who participate in trade-in programs always choose to purchase a new product. In practice, consumers may choose to buy new products either from the platform or directly from the brander or may only take the rebate without purchasing a new product. Third, this study focused solely on the resale-based platform model, while, in reality, many platforms operate using a combination of resale and agency models. Future research could incorporate hybrid channel strategies and heterogeneous consumer choices to enhance the model’s realism and applicability. Finally, empirical studies or real-world case validations would further strengthen the practical value of this research framework.

Author Contributions

Conceptualization, Y.S. and Z.H.; methodology, L.Z.; writing—review and editing, Y.S. All authors have read and agreed to the published version of the manuscript.

Funding

The Jiangsu Province Social Science Foundation Project (No. 23GLD003) and the Jiangsu University Philosophy and Social Science Research Project (No. 2022SJYB2243).

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Proof of Proposition 1.
When δ = 1 and c 1 = c 2 = 0 , we solve for the optimal equilibrium and profits under different models:
(1) Model BB:
Based on the brander’s decision making, the platform first optimizes the selling price in period 2:
p 2 B B = 2 α + β p 1 B B + r B B + 2 w 2 B B 4 β
Subsequently, the brander determines the optimal wholesale price and rebate in period 2 to maximize the profit. The Hessian matrix with respect to the wholesale price and rebate is first computed:
H = 2 β β β 3 β 2
which is a negative definite; that is, the profit function is strictly concave in w 2 B B and r B B . Hence, the following unique solutions can be found:
w 2 B B = α 2 β ,   r B B = p 1 B B + s 2
Next, considering the profit in period 1, the platform determines the selling price p 1 B B based on the outcomes above. Substituting the expressions derived above into the platform’s profit function, the platform’s optimal p 1 B B is obtained as
p 1 B B = 14 α β s + 16 β w 1 B B 31 β
Then, the brander chooses an optimal value of w 1 B B to maximize the total discounted profit. Using the equilibrium outcomes above, the optimal w 1 B B can be calculated:
w 1 B B = 161 s 896 + 487 α β
Substituting this value into the relevant decision variables and profit functions, and simplifying accordingly, the final equilibrium outcomes and profits under Model BB were derived. The results are summarized in Table 1.
(2) Model BT
First, given the brander’s decisions, the platform optimizes the selling price in period 2:
p 2 B T = 2 α + β p 1 B T + r B T + 2 w 2 B T 4 β
Then, using this outcome, the third-party recycler optimizes the rebate r B T :
r B T = 2 α + 3 β k b + p 1 B T + s + 2 β w 2 B T 6 β
Using r B T , the brander optimizes its second-period profit to determine the wholesale price w 2 B T and unit commission cost k B . The Hessian matrix is
H = 5 β 3 β 2 β 2 3 β 4
which is a negative definite; that is, the profit function is strictly concave in w 2 B T and k B . Hence, the following unique solutions can be found:
w 2 B T = α 2 β ,   k B = p 1 B T s 2
Next, to maximize the overall profit, the platform determines the selling price p 1 B B by substituting the expressions derived above into the platform’s profit function. The platform’s optimal p 1 B B is obtained as
p 1 B T = 182 α 3 β s + 192 β w 1 B T 381 β
Then, the brander chooses an optimal value of w 1 B T to maximize the total discounted profit. Using the equilibrium outcomes above, the optimal w 1 B T can be calculated:
w 1 B T = 24409 α 4227 β s 46464 β
Substituting this value into the relevant decision variables and profit functions, and simplifying accordingly, the final equilibrium outcomes and profits under Model BT were derived. The results are summarized in Table 3.
(3) Model PP
The method for deriving the optimal equilibrium solution is similar to that for Model BB. The results are summarized in Table 3.
(4) Model PT
The method for deriving the optimal equilibrium solution is similar to that for Model BT. The results are summarized in Table 3 and the substitute symbols used are as follows:
Table A1. Substitute symbols in Proposition 1.
Table A1. Substitute symbols in Proposition 1.
Ω 1 7310699 α 2 + 4312110 β α s + 720243 β 2 s 2
Ω 2 19595 α 2 86528 F β 10554 β α s + 14843 β 2 s 2
Ω 3 179700 α 2 88988 β α s + 92261 β 2 s 2
To ensure the existence of the model’s optimal solution, the selling price must be greater than the wholesale price in each period, i.e., p i j > w i j ( i = 1 , 2 ; j = B B , B T , P P , P T ). Accordingly, the following prerequisite condition is established: α > 61 β s 107 , or β < 107 α 61 s . □
Proof of Lemma 1.
(1)
In Model BB: w 1 B B α = 487 896 β > 0 , w 2 B B α = 1 2 β > 0 , p 1 B B α = 41 56 β > 0 , p 2 B B α = 295 448 β > 0 , r B B α = 41 112 β > 0 .
In Model BT: w 1 B T α = 2219 4224 β > 0 , w 2 B T α = 1 2 β > 0 , p 1 B T α = 49 66 β > 0 , p 2 B T α = 233 352 β > 0 , r B T α = 103 264 β > 0 , k B α = 49 132 β > 0 .
(2)
In Model BB: w 1 B B s = 23 128 < 0 , w 2 B B s = 0 , p 1 B B s = 1 8 < 0 , p 2 B B s = 9 64 > 0 , r B B s = 7 16 > 0 .
In Model BT: w 1 B T s = 1409 15488 < 0 , w 2 B T s = 0 , p 1 B T s = 13 242 < 0 , p 2 B T s = 255 3872 > 0 , r B T s = 203 968 > 0 , k B s = 255 484 < 0 . □
Proof of Lemma 2.
(1)
Π b B B α = 561 α 7 β s 896 β .
When 0 < β < 561 α 7 s , 561 α 7 β s > 0 ; i.e., b B B α > 0 .
When β > 561 α 7 s , b B B α < 0 .
(2)
p B B α = 10111 α + 6391 β s 50176 β > 0 . □
Proof of Lemma 3.
(1)
b B T α = 43 s 4224 + 697 α 1152 β .
When 0 < β < 7667 α 129 s ; i.e., b B T α > 0 .
When β > 7667 α 129 s , b B T α < 0 .
(2)
Π p B T α = 664609 α + 196005 β s 3066624 β > 0 . □
Proof of Lemma 4.
(1)
In Model PP: w 1 P P α = 271 832 β > 0 , w 2 P P α = 67 208 β > 0 , p 1 P P α = 37 52 β > 0 , p 2 P P α = 275 416 β > 0 , r P P α = 37 104 β > 0 .
In Model PT: w 1 P T α = 537 1376 β > 0 , w 2 P T α = 65 172 β > 0 , p 1 P T α = 63 86 β > 0 ,
p 2 P T α = 237 344 β > 0 , r P T α = 271 688 β > 0 , k p α = 63 172 β > 0 .
(2)
In Model PP: w 1 P P s = 71 832 > 0 , w 2 P P s = 59 208 > 0 , p 1 P P s = 7 52 < 0 , p 2 P P s = 59 416 > 0 , r P P s = 45 104 > 0 .
In Model PT: w 1 P T s = 197 2752 > 0 , w 2 P T s = 61 344 > 0 , p 1 P T s = 11 172 < 0 ,
p 2 P T s = 61 688 > 0 , r P T s = 339 1376 > 0 , k p s = 183 344 < 0 . □
Proof of Lemma 5.
(1)
b P P α = 329 α + 209 β s 832 β > 0 .
(2)
Π p P P α = 19595 α 5277 β s 43264 β .
When 0 < β < μ 1 , Π p P P α > 0 .
When β > μ 1 , Π p P P α < 0 , in which μ 1 = 19595 α 5277 s . □
Proof of Lemma 6.
(1)
b P T s = 582 α + 199 s β 1376 β > 0 .
(2)
Π p P T α = 89850 α 22247 β s 236672 β .
When 0 < β < μ 2 , Π p P T α > 0 .
When β > μ 2 , Π p P T α < 0 , in which μ 2 = 89850 α 22247 s . □
Proof of Corollary 1.
(i)
In period 1:
(1)
Compare the wholesale price in period 1 under the self-recycling models (Models BB and PP): w 1 B B w 1 P P = 2537 α 3087 β s 11648 β . When 107 α 61 s < β < 2537 α 3087 s , w 1 B B > w 1 P P . When β > 2537 α 3087 s , w 1 B B < w 1 P P . Then, compare the wholesale price in period 1 under the outsourcing models (Models BT and PT): w 1 B T w 1 P T = 269863 α 324783 β s 1997952 β . When 0 < β < μ 3 , w 1 B T > w 1 P T . When β > μ 3 , w 1 B T < w 1 P T , in which μ 3 = 2698863 α 324783 s . As μ 3 > 2537 α 3087 s , if β > 2698863 α 324783 s , then w 1 B B < w 1 P P and w 1 B T < w 1 P T .
(2)
P 1 B B P 1 P P = 15 α + 7 β s 728 β > 0 and p 1 B T p 1 P T = 616 α + 639 β s 62436 β > 0 . Hence, p 1 B B > p 1 P P and p 1 B T > p 1 P T hold universally. In addition, q 1 B B q 1 P P = 15 α + 7 β s 728 < 0 and q 1 B T q 1 P T = 616 α + 639 β s 62436 < 0 . Hence, q 1 B B < q 1 P P and q 1 B T < q 1 P T hold universally.
(ii)
In period 2:
(1)
Compare the wholesale price in period 2 under the self-recycling models (Models BB and PP): w 2 B B w 2 P P = 37 α 59 β s 208 β . When 37 α 59 β s > 0 (i.e., 0 < β < 37 α 59 s ), then w 2 B B > w 2 P P . When β > 37 α 59 s , w 2 B B < w 2 P P . Then, compare the wholesale price in period 2 under the outsourcing models (Models BT and PT): w 2 B T w 2 P T = 42 α 61 β s 344 β . When 42 α 61 β s > 0 (i.e., 0 < β < 42 α 61 s ), w 2 B T > w 2 P T . When β > 42 α 61 s , w 2 B T < w 2 P T . Therefore, if β > 42 α 61 s , then w 2 B B < w 2 P P and w 2 B T < w 2 P T .
(2)
p 2 B B p 2 P P = 15 α + 7 β s 5824 β < 0 ; i.e., p 2 B B < p 2 P P holds universally. In addition, p 2 B T p 2 P T = 4499 α + 3797 β s 166496 β < 0 ; i.e., p 2 B T < p 2 P T holds universally.
(3)
r B B r P P = 15 α + 7 β s 1456 β > 0 ; i.e., r B B > r P P holds universally. In addition, r B T r P T = 17 110 α + 1077 β s 499488 β < 0 ; i.e., r B T < r P T holds universally. □
Proof of Corollary 2.
(1)
Compare the brander’s profit between Models BB and PP:
Π b B B = 385 β 2 s 2 β ( 1792 F + 14 α s ) + 561 α 2 1792 β , b P P = 329 α 2 + 418 β α s + 153 β 2 s 2 1664 β . Therefore, Π b B B Π b P P = 2863 β 2 s 2 β ( 23296 F + 6034 α s ) + 2687 α 2 23296 β . Assuming that f 1 ( β ) = 2863 β 2 s 2 β ( 23296 F + 6034 α s ) + 2687 α 2 , then f 1 ( β ) is a continuous and concave function. Solving f 1 ( β ) = 0 , two distinct zeros are obtained:
β 1 min = 11648 F + 3017 α s 8 182 11648 F 2 + 6034 F α s + 121 α 2 s 2 2863 s 2 > 0 ,
β 1 max = 11648 F + 3017 α s + 8 182 11648 F 2 + 6034 F α s + 121 α 2 s 2 2863 s 2 > 0 .
Thus, when 0 < s < 91378 F 2039 α , under the given condition ( 0 , 107 α 61 s ) , the function has exactly one zero β 1 min . Hence, if 0 < β < β 1 min , then f 1 ( β ) > 0 ; i.e., Π b B B > Π b P P . Otherwise, if β > β 1 min , then Π b B B < Π b P P . This result is consistent with Figure 1a.
(2)
Compare the platform’s profit between Models BB and PP:
Π p B B = 4655 s 2 β 2 + 12782 β α s + 10111 α 2 100352 β , Π p P P = 14843 β 2 s 2 β ( 86528 F + 10554 a s ) + 19595 a 2 86528 β .
Hence, Π p B B Π p P P = 2122533 β 2 s 2 + β ( 16959488 F + 4228742 α s ) 2131861 α 2 16959488 β . Assuming that f 2 ( β ) = 2122533 β 2 s 2 + β ( 16959488 F + 4228742 α s ) 2131861 α 2 , then f 2 ( β ) is a continuous and convex function. Solving f 2 ( β ) = 0 , two distinct zeros are obtained:
β 2 min = 1211392 F + 302053 α s 416 8479744 F 2 + 4228742 F α s 6413 α 2 s 2 303219 s 2 ,
β 2 max = 1211392 F + 302053 α s + 416 8479744 F 2 + 4228742 F α s 6413 α 2 s 2 303219 s 2 .
Thus, when 0 < s < 639646 F 26769 α , under the given condition ( 0 , 107 α 61 s ) , the function has exactly one zero, β 2 min . Hence, if 0 < β < β 2 min , then f 2 ( β ) < 0 ; i.e., Π p B B < Π p P P . Otherwise, if β > β 2 min , then Π p B B > Π p P P .
As 91378 F 2039 α > 639646 F 26769 α , then if 0 < s < s 1 , the conclusions of Corollary 3(1) and Corollary 3(2) both hold with s 1 = 639646 F 26769 α . This result is consistent with Figure 1b. □
Proof of Corollary 3.
(1)
Compare the brander’s profit between Models BT and PT:
b B T = 27657 s 2 β 2 2838 β α s + 84337 α 2 278784 β , Π b P T = 1164 α 2 + 796 β α s + 155 β 2 s 2 5504 β . Thus, Π b B T Π b P T = 1091299 α 2 1855722 β α s + 851661 β 2 s 2 11987712 β > 0 ; i.e., Π b B T > Π b P T holds universally. This result is consistent with Figure 3a.
(2)
Compare the platform’s profit between Models BT and PT:
Π p B T = 7310699 α 2 + 4312110 β α s + 720243 β 2 s 2 67465728 β , Π p P T = 179700 α 2 88988 β α s + 92261 β 2 s 2 946688 β . Thus, Π p B T Π p P T = 10161406849 α 2 19698951162 β α s + 10825410402 β 2 s 2 124744131072 β < 0 ; i.e., Π p B T < Π p P T holds universally. This result is consistent with Figure 3b. □
Proof of Corollary 4.
(1)
Compare the whole supply chain profit between Models BB and PP:
Π B B = Π b B B + Π p B B = F + 41527 α 2 + 11998 β α s + 26215 β 2 s 2 100352 β ,
Π P P = Π b P P + Π p P P = 36703 α 2 86528 F β + 11182 β α s + 22799 β 2 s 2 86528 β .
Thus, Π B B Π P P = 781 15 α + 7 β s 2 16959488 β < 0 holds universally; i.e., Π B B < Π P P . This result is consistent with Figure 4a.
(2)
Compare the whole supply chain profit between Models BT and PT:
Π B T = Π b B T + Π p B T + Π t B T = 27738403 α 2 + 3120414 β α s + 10924587 β 2 s 2 67465728 β F 67465728 β ,
Π P T = Π b P T + Π p P T + Π t P T = 761260 α 2 1893376 F β + 72668 β α s + 330867 β 2 s 2 1893376 β .
Thus, Π B T Π P T = 3198890997 β 2 s 2 + 1963901280 β α s + 2266145354 α 2 249488262144 β . Assuming that f 3 ( β ) = 3198890997 β 2 s 2 + 1963901280 β α s + 2266145354 α 2 , then f 3 ( β ) is a continuous and convex function. Solving f 3 ( β ) = 0 , two distinct zeros are obtained:
β 3 min = 11 29756080 473 33711005298 α 1066296999 s < 0 ,
β 3 max = 11 29756080 + 473 33711005298 α 1066296999 s > 0 .
Thus, under the given condition ( 0 , 107 α 61 s ) , the function has exactly one zero, β 3 max . Hence, if 0 < β < β 3 max , then f 3 ( β ) > 0 ; i.e., Π B T > Π P T . Otherwise, if β > β 3 max , then Π B T < Π P T . This result is consistent with Figure 4b. □
Proof of Corollary 5.
The calculation process is similar to that of Corollary 1.
Proof of Corollary 6.
(1)
b B B b B T = 112833 s 2 β 2 + β ( 2310 α s 975744 F ) + 10285 α 2 975744 β . The remaining calculation process is similar to that of Corollary 2, where
β 4 min = 11 14784 F + 35 α s + 2 7 7805952 F 2 36960 a F s 38015 α 2 s 2 37611 s 2 ,
β 4 max = 11 14784 F 35 α s + 2 7 7805952 F 2 36960 a F s 38015 α 2 s 2 37611 s 2 .
In particular, β 4 min > 0 and β 4 max > 0 .
Thus, when 0 < α < α 0 , if 0 < β < β 4 min or β > β 4 max , then f 4 ( β ) > 0 ; i.e., Π b B B > Π b B T . If β 4 min < β < β 4 max , then Π b B B < Π b B T . Otherwise, when α > α 0 , Π b B B > Π b B T . This result is consistent with Figure 5a.
(2)
Π p B B Π p B T = 472217067 β 2 s 2 + 839097798 β α s 100580645 α 2 13223282688 β . Assuming that f 5 ( β ) = 472217067 β 2 s 2 + 839097798 β α s 100580645 α 2 , then f 5 ( β ) is a continuous and concave function. Solving f 5 ( β ) = 0 , two distinct zeros are obtained:
β 5 min = 11 1816229 + 88 540905749 α 22486527 s , β 5 max = 11 1816229 + 88 540905749 α 22486527 s , in which β 5 min < 0 and β 5 max > 0 . Hence, under the given condition ( 0 , 107 α 61 s ) , the function has exactly one zero, β 5 max .
Thus, if 0 < β < β 5 max , then f 5 ( β ) < 0 ; i.e., Π p B B < Π p B T . Otherwise, if β > β 5 max , then Π p B B > Π p B T . This result is consistent with Figure 5b. □
Proof of Corollary 7.
(1)
Π b P P Π b P T = 4564 β 2 s 2 + 7626 β α s 985 α 2 71552 β . Assuming that f 6 ( β ) = 4564 β 2 s 2 + 7626 β α s 985 α 2 , then f 6 ( β ) is a continuous and concave function. Solving f 6 ( β ) = 0 , two distinct zeros are obtained: β 6 min = 3813 + 19034509 α 4564 s , β 6 max = 3813 + 19034509 α 4564 s , in which β 6 min < 0 and β 6 max > 0 . Hence, under the given condition ( 0 , 107 α 61 s ) , the function has exactly one zero, β 6 max .
Thus, if 0 < β < β 6 max , then f 6 ( β ) < 0 ; i.e., Π b P P < Π b P T . Otherwise, if β > β 6 max , then Π b P P > Π b P T . This result is consistent with Figure 6a.
(2)
Π p P P Π p P T = 5861855 α 2 4475374 β α s + 2 β 79995136 F + 5926299 s 2 β 159990272 β , assuming that f 7 ( β ) = 5861855 α 2 4475374 β α s + 2 β 79995136 F + 5926299 s 2 β , then f 7 ( β ) is a continuous and concave function. Then, f 7 ( β ) = 42 [ 314840064 F + s 20723879 a + 95301423 s β ] , and there exists an extremum point β ¯ = 11 28621824 F 1883989 α s 95301423 s 2 such that the original function is monotonically decreasing when 0 < β < β ¯ (i.e., f 7 ( β ) < 0 ) and monotonically increasing when β > β ¯ . As a result, the original function has two zeros when f 7 ( β ¯ ) < 0 . Solving this yields the range of α as 0 < α < α 1 , where α 1 = ( 256 69478210849290 + 572847872 ) F 206319641 s . Hence, two distinct zeros are obtained:
β 7 min = 79995136 F + 2237687 α s 559 20478754816 F 2 + 1145695744 F α s 206319641 α 2 s 2 11852598 s 2 ,
β 7 max = 79995136 F + 2237687 α s + 559 20478754816 F 2 + 1145695744 F α s 206319641 α 2 s 2 11852598 s 2 , in which β 7 min > 0 and β 7 max > 0 .
When α < α 1 , the zeros exist. Therefore, if 0 < β < β 7 min or β > β 7 max , then Π p P P > Π p P T . If β 7 min < β < β 7 max , then Π p P T > Π p P P . When α > α 1 , Π p P P > Π p P T . This result is consistent with Figure 6b. □
Proof of Corollary 8.
(1)
Compare the whole supply chain profits under brander-led models (Models BB and BT):
As Π B B = Π b B B + Π p B B = F + 41527 α 2 + 11998 β α s + 26215 β 2 s 2 100352 β and
Π B T = Π b B T + Π p B T + Π t B T = 27738403 α 2 + 3120414 β α s + 10924587 β 2 s 2 67465728 β F 67465728 β , Π B B Π B T = 1313105283 β 2 s 2 + 969363318 β α s + 35244275 α 2 13223282688 β > 0 . As a result, Π B B > Π B T holds universally. This result is consistent with Figure 7a.
(2)
Compare the whole supply chain profits under platform-led models (Models PP and PT):
As Π P P = Π b P P + Π p P P = 36703 α 2 86528 F β + 11182 β α s + 22799 β 2 s 2 86528 β and Π P T = Π b P T + Π p P T + Π t P T = 761260 α 2 1893376 β F + 72668 β α s + 330867 β 2 s 2 1893376 β , Π P P Π P T = 7074754 α 2 + 29070144 β α s + 28394179 β 2 s 2 319980544 β > 0 . As a result, Π P P > Π P T holds universally. This result is consistent with Figure 7b. □
Proof of Corollary 9.
(1)
It is shown in Corollaries 3, 4, and 8 that if 0 < β < β 1 min , then Π b B B > Π b P P . Furthermore, if β > β 1 min , then Π b B B < Π b P P , in which β 1 min = 11648 F + 3017 α s 8 182 11648 F 2 + 6034 F α s + 121 α 2 s 2 2863 s 2 > 0 . Moreover, Π b B T > Π b P T holds universally.
When 0 < α < α 0 :
If 0 < β < β 4 min or β > β 4 max , then f 4 ( β ) > 0 ; i.e., Π b B B > Π b B T . If β 4 min < β < β 4 max , then Π b B B < Π b B T . If α > α 0 , then Π b B B > Π b B T , in which β 4 min = 11 14784 F + 35 α s + 2 7 7805952 F 2 36960 a F s 38015 α 2 s 2 37611 s 2 and β 4 max = 11 14784 F 35 α s + 2 7 7805952 F 2 36960 a F s 38015 α 2 s 2 37611 s 2 .
As b P P = 329 α 2 + 418 β α s + 153 β 2 s 2 1664 β and b B T = 27657 s 2 β 2 2838 β α s + 84337 α 2 278784 β , Π b P P Π b B T = 26307 β 2 s 2 + 947298 β α s 379819 α 2 3624192 β . Assuming that f 8 ( β ) = 26307 β 2 s 2 + 947298 β α s 379819 α 2 , then f 8 ( β ) is a continuous and convex function. Solving f 8 ( β ) = 0 , two distinct zeros are obtained:
β 8 min = 11 14353 12 1366898 α 8769 s , β 8 max = 11 14353 + 12 1366898 α 8769 s , in which 0 < β 8 min < 107 α 61 s and β 8 max > 107 α 61 s . Hence, under the given condition β ( 0 , 107 α 61 s ) , the function has exactly one zero, β 8 min . Therefore, if 0 < β < β 8 min , then f 8 ( β ) < 0 ; i.e., Π b P P < Π b B T . If β > β 8 min , then Π b P P > Π b B T .
When α > α 2 β 8 min β 1 min = 4 25535328 F + 7 15198677 13497 1366898 α s + 17538 182 11648 F 2 + 6034 F α s + 121 α 2 s 2 25105647 s 2 < 0 .
Hence, β 1 min > β 8 min , where α 2 = 18480 F + 1584 118405 F 38015 s .
β 4 max exists when 0 < α < α 0 , and α 2 < α 0 . Moreover, when α 2 < α < α 0 , β 1 min β 4 max = 2 1964256 F 8183903 α s + 4499 7 7805952 F 2 36960 a F s 38015 α 2 s 2 + 21492 182 11648 F 2 + 6034 F α s + 121 α 2 s 2 15382899 s 2 > 0
Therefore, Π b P P > Π b B B and Π b P P > Π b B T exist when β > β 1 min . As Π b B T > Π b P T holds universally, when α > α 2 and β > β 1 min , Π b P P > max { Π b B B , Π b B T , Π b P T } .
When 0 < α < α 2 :
β 8 min β 1 min = 4 25535328 F + 7 15198677 13497 1366898 α s + 17538 182 11648 F 2 + 6034 F α s + 121 α 2 s 2 25105647 s 2 > 0 Π b P P > Π b B B and Π b P P > Π b B T exist when β > β 8 min . As Π b B T > Π b P T holds universally, when 0 < α < α 2 and β > β 8 min , Π b P P > max { Π b B B , Π b B T , Π b P T } .
In summary, when β > max { β 8 min , β 1 min } , Π b P P > max { Π b B B , Π b B T , Π b P T } . This result is consistent with Figure 8a.
(2)
It is shown in Corollaries 3, 4, and 8 that if 0 < β < β 2 min , then Π p B B < Π p P P . Otherwise, if β > β 2 min , then Π p B B > Π p P P , in which β 2 min = 1211392 F + 302053 α s 416 8479744 F 2 + 4228742 F α s 6413 α 2 s 2 303219 s 2 . Secondly, Π p B T < Π p P T holds universally. Thirdly, if 0 < β < β 5 max , then Π p B B < Π p B T . Otherwise, if β > β 5 max , then Π p B B > Π p B T , in which β 5 max = 11 1816229 + 88 540905749 α 22486527 s .
As Π p B B = 4655 s 2 β 2 + 12782 β α s + 10111 α 2 100352 β and Π p B B = 4655 s 2 β 2 + 12782 β α s + 10111 α 2 100352 β , Π p B B Π p P T = 9476061 β 2 s 2 + 41075566 β α s 16525961 α 2 185550848 β . Assuming that f 9 ( β ) = 9476061 β 2 s 2 + 41075566 β α s 16525961 α 2 , then f 9 ( β ) is a continuous and convex function. Solving f 9 ( β ) = 0 , two distinct zeros are obtained:
β 9 min = 2933969 86 731778667 α 1353723 s , β 9 max = 2933969 + 86 731778667 α 1353723 s , in which 0 < β 9 min < 107 α 61 s and β 9 max > 107 α 61 s . Hence, under the given condition β ( 0 , 107 α 61 s ) , the function has exactly one zero, β 9 min . Therefore, if 0 < β < β 9 min , then f 9 ( β ) < 0 ; i.e., Π p B B < Π p P T .
As β 9 min β 2 min = 2933969 86 731778667 α 1353723 s 1211392 F + 302053 α s 416 8479744 F 2 + 4228742 F α s 6413 α 2 s 2 303219 s 2 , if α > α 3 , then β 9 min < β 2 min , in which α 3 = 605696 114935233778433965 + 937486 3941133241299582026227 F 9148789661599884969531 s . Therefore, Π p B B > Π p P P and Π p B B > Π p P T exist when β > β 2 min . Given that Π p B T < Π p P T holds universally, there exists Π p B B > max { Π p B T , Π p P P , Π p P T } .
Otherwise, if 0 < α < α 3 , then β 9 min β 2 min > 0 ; i.e., β 9 min > β 2 min . Hence, Π p B B > Π p P P and Π p B B > Π p P T exist when β > β 9 min . Given that Π p B T < Π p P T holds universally, there exists Π p B B > max { Π p B T , Π p P P , Π p P T } .
In summary, if β > max { β 9 min , β 2 min } , then Π p B B > max { Π p B T , Π p P P , Π p P T } . This result is consistent with Figure 8b. □
Proof of Corollary 10.
  • Model BB*: As the brander independently recycles and processes the salvage in this model, there is no transfer pricing of the salvage. Thus, the profits stay the same as in the original Model BB.
  • Model BT*: The cost paid by the brander to acquire the salvage replaces the commission in the original model, while the other components of profit remain unchanged. Therefore, the game player’s maximum profit is as follows:
    Π b B T * = ( w 1 B T * c 1 B T * ) q 1 + δ [ ( w 1 B T * c 1 B T * ) ( q 2 + q t ) + ( s λ B T * ) q t ]
    Π t B T * = δ [ ( λ B T * r B T * ) q t F ]
    Π p B T * = ( p 1 B T * w 1 B T * ) q 1 + δ [ ( p 2 B T * w 2 B T * ) ( q 2 + q t ) ]
  • Model PP*: Based on the original assumptions, the brander pays the platform a transfer price λ P P * for the salvage and receives the processed salvage in return. Therefore, the game player’s maximum profit is as follows:
    Π b P P * = ( w 1 P P * c 1 P P * ) q 1 + δ [ ( w 1 P P * c 1 P P * ) ( q 2 + q t ) + ( s λ P P * ) q t ]
    Π p P P * = ( p 1 P P * w 1 P P * ) q 1 + δ [ ( p 2 P P * w 2 P P * ) ( q 2 + q t ) + ( λ P P * r p p * ) q t F ]
  • Model PT*: Similarly to Model PP*, the game player’s maximum profit is as follows:
    Π b P T * = ( w 1 P T * c 1 P T * ) q 1 + δ [ ( w 1 P T * c 1 P T * ) ( q 2 + q t ) + ( s λ P T * ) q t ]
    Π p P T = ( p 1 P T * w 1 P T * ) q 1 + δ [ ( p 2 P T * w 2 P T * ) ( q 2 + q t ) + ( λ P T * k p * ) q t F ]
    Π p B T * = δ [ ( k P * r P T * ) q t F ]
The method for the computation of the optimal equilibrium is consistent with that in Proposition 1, and the results are presented in the table below.
Table A2. Equilibrium prices and profits under different extended recycling models.
Table A2. Equilibrium prices and profits under different extended recycling models.
EquilibriumModel BB*Model BT*Model PP*Model PT*
w 1 161 s 896 + 487 α β 24409 α 4227 β s 46464 β 17 s 224 + 13 α 28 β 11 s 320 + 29 α 60 β
p 1 s 8 + 41 α 56 β 13 s 242 + 49 α 66 β 3 s 28 + 5 α 7 β 1 60 3 s + 44 α β
w 2 α 2 β α 2 β α 2 β α 2 β
p 2 9 s 64 + 295 α 448 β 255 s 3872 + 2563 α 3872 β 3 α 4 β 3 α 4 β
r 7 s 16 + 41 α 112 β 203 s 968 + 103 α 264 β 1 112 19 s + 60 α β 13 s 160 + 31 α 60 β
k --- 1 80 17 s + 44 α β
λ - 255 s 484 + 49 α 132 β 5 4 α + 5 β s 56 β 19 s 40 + 11 α 30 β
Π b 385 β 2 s 2 + 561 α 2 β ( 1792 F + 14 α s ) 1792 β 84337 α 2 2838 β α s + 27657 β 2 s 2 278784 β 48 β α s + 65 β 2 s 2 + 120 α 2 448 β 96 β α s + 496 α 2 + 129 β 2 s 2 1920 β
Π p 10111 α 2 + 12782 β α s + 4655 β 2 s 2 100352 β Ω 4 67465728 β 1744 α 2 + 720 α s β + 919 β 2 s 2 12544 F 12544 β 2976 β α s + 15136 α 2 + 3879 β 2 s 2 115200 β
Π t - F + 25 11 α 153 β s 2 11244288 β - F + 16 α + 63 β s 2 230400 β
Ω 4 = 7310699 α 2 + 4312110 β α s + 720243 β 2 s 2 .
To ensure the existence of the model’s optimal solution, the selling price must be greater than the wholesale price in each period; i.e., p i j > w i j ( i = 1 , 2 ; j = B B * , B T * , P P * , P T * ). Accordingly, the following prerequisite condition is established: α > 61 β s 107 or β < 107 α 61 s (as for the original models).
(1)
Comparing the brander’s profit in different models
Π b B B * Π b P P * = β ( 1792 F 78 α s ) + 81 α 2 + 621 β 2 s 2 1792 β > 0 ,
Π b B T * Π b P P * = 67639 α 2 89562 β α s + 450603 β 2 s 2 1951488 β > 0 ,
Π b B B * Π b P T * = 1471 α 2 1554 β α s + 3969 β 2 s 2 26880 β F 26880 β > 0 ,
Π b B T * Π b P T * = 61589 α 2 83886 β α s + 44631 β 2 s 2 1393920 β > 0 .
As a result, the brander’s profit can meet the highest level only in brander-led models (i.e., Models BB* and BT*). As Π b B B * = Π b B B and Π b B T * = Π b B T the remaining calculation follows the same procedure as in Corollary 11. This result is consistent with Figure 9a.
(2)
Comparing the platform’s profit in different models
Π p B B * Π p P T * = 287091 β 2 s 2 + 2292654 β α s 691681 α 2 22579200 β , assuming that f 12 ( β ) = 287091 β 2 s 2 + 2292654 α s β 691681 α 2 , then f 12 ( β ) is a continuous concave function. Solving f 12 ( β ) = 0 , two distinct zeros are obtained:
β 12 min = 54587 α 10 34300249 α 13671 s , β 12 max = 54587 α + 10 34300249 α 13671 s .
As β 12 min < 0 , under the given condition β ( 0 , 107 α 61 s ) , the function has exactly one zero, β 12 max . Thus, if β > β 12 max , then Π p B B * > Π p P T * .
Π p B B * Π p P T * = 22391 β 2 s 2 + β ( 100352 F + 7022 α s ) 3841 α 2 100352 β , assuming that f 13 ( β ) = 22391 β 2 s 2 + β ( 100352 F + 7022 α s ) 3841 α 2 , then f 13 ( β ) is a continuous and concave function. Solving f 13 ( β ) = 0 , two distinct zeros are obtained:
β 13 min = 50176 F + 3511 α s + 2 629407744 F 2 + 88083968 F α s + 24582738 α 2 s 2 22391 s 2 ,
β 13 max = 50176 F 3511 α s + 2 629407744 F 2 + 88083968 F α s + 24582738 α 2 s 2 22391 s 2 , in which β 13 min < 0 and β 13 max > 0 . Hence, under the given condition β ( 0 , 107 α 61 s ) , the function has exactly one zero, β 13 max . Therefore, if β > β 13 max , then Π p B B * > Π p P P * .
As β 13 max β 12 max = 685956096 F + 2 587129318 111955 34300249 α s + 27342 629407744 F 2 + 88083968 F α s + 24582738 α 2 s 2 306107361 s 2 < 0 holds universally, if β > β 12 max , then Π p B B * > Π p P T * and Π p B B * > Π p P P * .
Combined with Corollary 8, we can state that if β > β 5 max , then Π p B B * > Π p B T * , in which β 5 max = 11 1816229 + 88 540905749 α 22486527 s . As β 5 max β 12 max = 2 7574172350 1784645 34300249 + 105028 540905749 α 4879576359 s < 0 holds universally, if β > β 12 max , then Π p B B * > Π p P T * , Π p B B * > Π p P P * , and Π p B B * > Π p B T * exist at the same time; i.e., Π p B B * > max { Π p B T * , Π p P P * , Π p P T * } . This result is consistent with Figure 9b. □
Proof of Corollary 11.
(1)
From the Table A2, we have that Π b B B * = Π b B B and Π b B T * = Π b B T .
(2)
From the Table A1, we have that Π b B T * = Π b B T and Π p B T * = Π p B T .
(3)
Π b P P * Π b P P = 2605 β 2 s 2 2510 β α s + 817 α 2 11648 β .
Assuming that f 10 ( β ) = 2605 β 2 s 2 2510 β α s + 817 α 2 , then f 10 ( β ) is a continuous and convex function. Solving f 10 ( β ) = 0 two distinct zeros are obtained:
β 10 min = 1255 α 3703310 α 2605 s < 0 , β 10 max = 1255 α + 3703310 α 2605 s > 0
Hence, under the given condition β ( 0 , 107 α 61 s ) , the function has exactly one zero, β 10 max . Therefore, if 0 < β < β 10 max , then Π b P P * Π b P P > 0 ; i.e., Π b P P * > Π b P P . Otherwise, if β > β 10 max , then Π b P P * < Π b P P .
When α > 61 β s 107 , π p P P Π p P P = 370683 α 2 760506 β α s + 1476653 β 2 s 2 4239872 β < 0 holds universally; i.e., Π p P P * < Π p P P . This result is consistent with Figure 10.
(4)
π b P T Π b P T = 1611 β 2 s 2 3906 β α s + 1934 α 2 41280 β . Assuming that f 11 ( β ) = 1611 β 2 s 2 3906 β α s + 1934 α 2 , then f 11 ( β ) is a continuous and concave function. Solving f 11 ( β ) = 0 , two distinct zeros are obtained: β 11 min = 651 19 215 α 537 s , β 11 max = 651 + 19 215 α 537 s . When 0 < s < s 1 there is only one zero, β 11 min , in which s 1 = 1 537 651 + 19 215 . Meanwhile, if 0 < β < β 11 min , then f 11 ( β ) > 0 ; i.e., Π b P T * > Π b P T .
When α > 61 β s 107 , π p P T Π p P T = 6223018 α 2 12762462 β α s + 6793227 β 2 s 2 106502400 β < 0 holds universally; i.e., π p P P < Π p P P . This result is consistent with Figure 11. □

References

  1. Dou, G.; Ma, L.; Wei, K.; Zhang, Q. (Eds.) Does Implementing Trade-In and Green Technology Together Benefit the Environment? In Operations Management for Environmental Sustainability: Operational Measures, Regulations and Carbon Constrained Decisions; Springer International Publishing: Cham, Switzerland, 2023; pp. 85–121. ISBN 978-3-031-37600-9. [Google Scholar]
  2. Bai, J.; Hu, S.; Gui, L.; So, K.C.; Ma, Z. Optimal Subsidy Schemes and Budget Allocations for Government-subsidized Trade-in Programs. Prod. Oper. Manag. 2021, 30, 2689–2706. [Google Scholar] [CrossRef]
  3. Zheng, B.; Cai, Y.; Li, S. Trade-in and Resale in a Platform Supply Chain: Manufacturer’s Choice of Selling Strategies. Transp. Res. Part E Logist. Transp. Rev. 2025, 193, 103836. [Google Scholar] [CrossRef]
  4. China News Service. Last Year, the Trade-in of Old Products for New Ones Boosted the Consumption of Home Appliances by 270 Billion Yuan. Available online: https://www.china-news-online.com/lang/English/4235729.html (accessed on 8 May 2025).
  5. Wang, R.; Du, Y.; Wei, Y. Does the Bike-Sharing Platform Share Its Own Private Information or Not in a Reverse Supply Chain? Sustainability 2024, 16, 9194. [Google Scholar] [CrossRef]
  6. Ma, D.-Q.; Wang, X.-Q.; Wang, X.; Hu, J.-S. Achieving Triple Benefits in a Platform-Based Closed-Loop Supply Chain: The Optimal Combination between Recycling Channel and Blockchain. Sustainability 2023, 15, 3921. [Google Scholar] [CrossRef]
  7. Mazzucchelli, A.; Chierici, R.; Del Giudice, M.; Bua, I. Do Circular Economy Practices Affect Corporate Performance? Evidence from Italian Large-Sized Manufacturing Firms. Corp. Soc. Responsib. Environ. Manag. 2022, 29, 2016–2029. [Google Scholar] [CrossRef]
  8. Yang, V. JD.com Partners with Leading Home Brands to Launch Green Trade-in Alliance. Available online: https://jdcorporateblog.com/jd-com-partners-with-leading-home-brands-to-launch-green-trade-in-alliance/ (accessed on 12 February 2025).
  9. Hu, S.; Ma, Z.-J.; Sheu, J.-B. Optimal Prices and Trade-in Rebates for Successive-Generation Products with Strategic Consumers and Limited Trade-in Duration. Transp. Res. Part E Logist. Transp. Rev. 2019, 124, 92–107. [Google Scholar] [CrossRef]
  10. Yi, Y.; Chen, J.; Feng, J. Optimizing Prices in Trade-in Strategies for Vehicle Manufacturers. Comput. Ind. Eng. 2021, 157, 107338. [Google Scholar] [CrossRef]
  11. Xiao, Y. Choosing the Right Exchange-Old-for-New Programs for Durable Goods with a Rollover. Eur. J. Oper. Res. 2017, 259, 512–526. [Google Scholar] [CrossRef]
  12. Zhao, W.; Xia, Y.; Zhang, L.; Fan, X. Trade-in Program Options in a Retailer-led Supply Chain. In International Transactions in Operational Research; Wiley Online Library: Montreal, QC, Canada, 2024; p. itor.13574. [Google Scholar] [CrossRef]
  13. Wang, W.; Feng, L.; Chen, X.; Yang, L.; Choi, T.-M. Impacts of Selling Models: Who Should Offer Trade-in Programs in e-Commerce Supply Chains? Transp. Res. Part E Logist. Transp. Rev. 2024, 186, 103524. [Google Scholar] [CrossRef]
  14. Yu, Z.; Wang, Y.; Shen, L.; Ji, X. Decision Making by Trade-In Programs in the E-Commerce Supply Chain, Considering Cash Rebate Strategies and Platform Services. Mathematics 2024, 12, 3792. [Google Scholar] [CrossRef]
  15. Bonina, C.; Koskinen, K.; Eaton, B.; Gawer, A. Digital Platforms for Development: Foundations and Research Agenda. Inf. Syst. J. 2021, 31, 869–902. [Google Scholar] [CrossRef]
  16. Zhang, Z.; Xu, H.; Chen, K.; Zhao, Y.; Liu, Z. Channel Mode Selection for an E-Platform Supply Chain in the Presence of a Secondary Marketplace. Eur. J. Oper. Res. 2023, 305, 1215–1235. [Google Scholar] [CrossRef]
  17. Guo, L.; Zhang, Q.; Wu, J.; Santibanez Gonzalez, E.D.R. An Evolutionary Game Model of Manufacturers and Consumers’ Behavior Strategies for Green Technology and Government Subsidy in Supply Chain Platform. Comput. Ind. Eng. 2024, 189, 109918. [Google Scholar] [CrossRef]
  18. Pan, W.; Lin, M. A Two-Stage Closed-Loop Supply Chain Pricing Decision: Cross-Channel Recycling and Channel Preference. Axioms 2021, 10, 120. [Google Scholar] [CrossRef]
  19. Li, D.; Cruz, J.M.; Ke, K. Closed-Loop Supply Chain Network Model with Recycling, Re-Manufacturing, Refurbishing, and Corporate Social Responsibility. Ann. Oper. Res. 2025, 345, 247–275. [Google Scholar] [CrossRef]
  20. Yang, Q.; Sun, L. Eco-Design in Manufacturing or Remanufacturing? The Sustainable Options in a Closed-Loop Supply Chain with Outsourcing. Sustainability 2024, 16, 1633. [Google Scholar] [CrossRef]
  21. Li, Y.; Wang, K.; Xu, F.; Fan, C. Management of Trade-in Modes by Recycling Platforms Based on Consumer Heterogeneity. Transp. Res. Part E Logist. Transp. Rev. 2022, 162, 102721. [Google Scholar] [CrossRef]
  22. Cao, K.; Gong, Y.; Xu, Y.; Wang, J. Optimal Trade-in Delegation Strategy Considering Third-Party Recycler Intrusion and Used Products with Different Durability. Oper. Res. 2025, 25, 1. [Google Scholar] [CrossRef]
  23. Zheng, B.; Chu, J.; Jin, L. Recycling Channel Selection and Coordination in Dual Sales Channel Closed-Loop Supply Chains. Appl. Math. Modell. 2021, 95, 484–502. [Google Scholar] [CrossRef]
  24. Li, H.; Yi, Y. Optimal Trade-in Rebate Payment Strategies for B2C Platforms under Different Recycling Models. J. Ind. Manag. Optim. 2024, 20, 1395–1434. [Google Scholar] [CrossRef]
  25. Ferrer, G.; Swaminathan, J.M. Managing New and Remanufactured Products. Manag. Sci. 2006, 52, 15–26. [Google Scholar] [CrossRef]
  26. Yin, R.; Tang, C.S. Optimal Temporal Customer Purchasing Decisions under Trade-In Programs with up-Front Fees. Decis. Sci. 2014, 45, 373–400. [Google Scholar] [CrossRef]
  27. Quan, Y.; Hong, J.; Song, J.; Leng, M. Game-Theoretic Analysis of Trade-in Services in Closed-Loop Supply Chains. Transp. Res. Part E Logist. Transp. Rev. 2021, 152, 102428. [Google Scholar] [CrossRef]
  28. Hu, S.; Tang, Y. Impact of Product Sharing and Heterogeneous Consumers on Manufacturers Offering Trade-in Programs. Omega 2024, 122, 102949. [Google Scholar] [CrossRef]
  29. Genc, T.S.; De Giovanni, P. Optimal Return and Rebate Mechanism in a Closed-Loop Supply Chain Game. Eur. J. Oper. Res. 2018, 269, 661–681. [Google Scholar] [CrossRef]
  30. Raz, G.; Souza, G.C. Recycling as a Strategic Supply Source. Prod. Oper. Manag. 2018, 27, 902–916. [Google Scholar] [CrossRef]
  31. Yenipazarli, A. Managing New and Remanufactured Products to Mitigate Environmental Damage under Emissions Regulation. Eur. J. Oper. Res. 2016, 249, 117–130. [Google Scholar] [CrossRef]
  32. Apple. Trade In. Available online: https://www.apple.com/shop/trade-in (accessed on 8 May 2025).
  33. Vogue. Gucci and the RealReal Announce a Game-Changing Partnership. Available online: https://www.vogue.com/article/gucci-the-realreal-partnership-secondhand-consignment (accessed on 8 May 2025).
Figure 1. Decision-making sequences in four models.
Figure 1. Decision-making sequences in four models.
Sustainability 17 06158 g001
Figure 2. Profit comparison under Models BB and PP (s = 1; F = 1). (a) Brand profit comparison under Models BB and PP. (b) Platform profit comparison under Models BB and PP.
Figure 2. Profit comparison under Models BB and PP (s = 1; F = 1). (a) Brand profit comparison under Models BB and PP. (b) Platform profit comparison under Models BB and PP.
Sustainability 17 06158 g002
Figure 3. Profit comparison under Models BT and PT (s = 1; F = 1). (a) Brand profit comparison under Models BT and PT. (b) Platform profit comparison under Models BT and PT.
Figure 3. Profit comparison under Models BT and PT (s = 1; F = 1). (a) Brand profit comparison under Models BT and PT. (b) Platform profit comparison under Models BT and PT.
Sustainability 17 06158 g003
Figure 4. Comparison of supply chain profit (s = 1; F = 1). (a) Profit comparison under Models BB and PP. (b) Profit comparison under Models BT and PT.
Figure 4. Comparison of supply chain profit (s = 1; F = 1). (a) Profit comparison under Models BB and PP. (b) Profit comparison under Models BT and PT.
Sustainability 17 06158 g004
Figure 5. Profit comparison under Models BB and BT (s = 1; F = 1). (a) Brander profit comparison under Models PP and PT. (b) Platform profit comparison under Models PP and PT.
Figure 5. Profit comparison under Models BB and BT (s = 1; F = 1). (a) Brander profit comparison under Models PP and PT. (b) Platform profit comparison under Models PP and PT.
Sustainability 17 06158 g005
Figure 6. Profit comparison under Models PP and PT (s = 1; F = 1). (a) Brander profit comparison under Models PP and PT. (b) Platform profit comparison under Models PP and PT.
Figure 6. Profit comparison under Models PP and PT (s = 1; F = 1). (a) Brander profit comparison under Models PP and PT. (b) Platform profit comparison under Models PP and PT.
Sustainability 17 06158 g006
Figure 7. Comparison of supply chain profit (s = 1; F = 1). (a) Profit comparison under Models BB and BT. (b) Profit comparison under Models PP and PT.
Figure 7. Comparison of supply chain profit (s = 1; F = 1). (a) Profit comparison under Models BB and BT. (b) Profit comparison under Models PP and PT.
Sustainability 17 06158 g007
Figure 8. Profit comparison under all four models (s = 1; F = 1). (a) Brander profit comparison under all four models. (b) Platform profit comparison under all four models.
Figure 8. Profit comparison under all four models (s = 1; F = 1). (a) Brander profit comparison under all four models. (b) Platform profit comparison under all four models.
Sustainability 17 06158 g008
Figure 9. Profit comparison under four extended models (s = 1; F = 1). (a) Brander profit comparison under all four models. (b) Platform profit comparison under all four models.
Figure 9. Profit comparison under four extended models (s = 1; F = 1). (a) Brander profit comparison under all four models. (b) Platform profit comparison under all four models.
Sustainability 17 06158 g009
Figure 10. A profit comparison between the original Model PP and the extended Model PP* (s = 1; F = 1). (a) The brander profit comparison. (b) The platform profit comparison.
Figure 10. A profit comparison between the original Model PP and the extended Model PP* (s = 1; F = 1). (a) The brander profit comparison. (b) The platform profit comparison.
Sustainability 17 06158 g010
Figure 11. The profit comparison between the original Model PT and the extended Model PT* (s = 1; F = 1). (a) The brander profit comparison. (b) The platform profit comparison.
Figure 11. The profit comparison between the original Model PT and the extended Model PT* (s = 1; F = 1). (a) The brander profit comparison. (b) The platform profit comparison.
Sustainability 17 06158 g011
Figure 12. A profit comparison between the original models and the extended models with revenue sharing ( α = 10 ; s = 1; F = 1; β = 0.5 ).
Figure 12. A profit comparison between the original models and the extended models with revenue sharing ( α = 10 ; s = 1; F = 1; β = 0.5 ).
Sustainability 17 06158 g012
Table 1. A comparison between this study and the related literature.
Table 1. A comparison between this study and the related literature.
Research
Paper
Brander-LedPlatform-Led2-Period
Self-RecyclingOutsourcingSelf-RecyclingOutsourcing
Yi et al. [10]
Bai et al. [2]
Xiao [11]
Zhao et al. [12]
Wang et al. [13]
Yu et al. [14]
Zheng et al. [3]
Cao et al. [22]
Zheng et al. [23]
Li &Yi [24]
Yin & Tang [26]
Quan et al. [27]
Hu & Tang [28]
This paper
Table 2. Summary of notation.
Table 2. Summary of notation.
SymbolDefinitionSymbolDefinition
i Period i , i = 1 , 2 c i The cost for a brander to produce new products in period i
j Model j , ( j = B B , B T , P P , P T , j * is the extension model) q i New customer demand in period i
w i j In period i , brander’s wholesale price in Model j q t The demand from existing customers for trade-in programs
p i j In period i , the platform’s sales price under Model j α Market potential
r j Rebate price offered in trade-in service under Model j β Customer’s price sensitivity
s The salvage value of old products b j Brander’s profit under Model j
F Fixed costs for recycling old products p j Platform’s profit under Model j
k B The unit commission paid by the brander to the third-party recycler t j The third-party recycler’s profit under Model j
k P The unit commission paid by the platform to the third-party recycler λ j * The transfer price of salvage in the extended model
Table 3. Equilibrium prices and profits under different recycling models.
Table 3. Equilibrium prices and profits under different recycling models.
EquilibriumModel BBModel BTModel PPModel PT
Wholesale in period 1: w 1 487 α 896 β 161 s 896 24409 α 4227 β s 46464 β 271 α + 71 β s 832 β 1074 α + 197 β s 2752 β
Sales price in period 1: p 1 41 α 56 β s 8 49 α 66 β 13 s 242 37 α 52 β 7 s 52 63 α 86 β 11 s 172
Wholesale in period 2: w 2 α 2 β α 2 β 67 α + 59 β s 208 β 61 s 344 + 130 α 344 β
Sales price in period 2: p 2 9 s 64 + 295 α 448 β 255 s 3872 + 2563 α 3872 β 275 α + 59 β s 416 β 474 α + 61 β s 688 β
Rebate price: r 7 s 16 + 41 α 112 β 203 s 968 + 103 α 264 β 37 α + 45 β s 104 β 542 α + 339 β s 1376 β
Unit commission: k - 49 α 132 β 255 s 484 - 3 42 α 61 β s 344 β
Brander’s profit: Π b 385 β 2 s 2 + 561 α 2 β ( 1792 F + 14 α s ) 1792 β 84337 α 2 2838 β α s + 27657 β 2 s 2 278784 β 329 α 2 + 418 β α s + 153 β 2 s 2 1664 β 1164 α 2 + 796 β α s + 155 β 2 s 2 5504 β
Platform’s profit: Π p 10111 α 2 + 12782 β α s + 4655 β 2 s 2 100352 β Ω 1 67465728 β Ω 2 86528 β Ω 3 946688 β
The 3P’s profit: Π t - 25 11 α 153 β s 2 11244288 β F 11244288 β - 38 α 305 β s 2 1893376 β F 1893376 β
Ω 1 , Ω 2 , and Ω 3 are defined in Appendix A.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zhu, L.; Si, Y.; Han, Z. Self-Recycling or Outsourcing? Research on the Trade-In Strategy of a Platform Supply Chain. Sustainability 2025, 17, 6158. https://doi.org/10.3390/su17136158

AMA Style

Zhu L, Si Y, Han Z. Self-Recycling or Outsourcing? Research on the Trade-In Strategy of a Platform Supply Chain. Sustainability. 2025; 17(13):6158. https://doi.org/10.3390/su17136158

Chicago/Turabian Style

Zhu, Lingrui, Yinyuan Si, and Zhihua Han. 2025. "Self-Recycling or Outsourcing? Research on the Trade-In Strategy of a Platform Supply Chain" Sustainability 17, no. 13: 6158. https://doi.org/10.3390/su17136158

APA Style

Zhu, L., Si, Y., & Han, Z. (2025). Self-Recycling or Outsourcing? Research on the Trade-In Strategy of a Platform Supply Chain. Sustainability, 17(13), 6158. https://doi.org/10.3390/su17136158

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop