Skip to Content
SustainabilitySustainability
  • Article
  • Open Access

1 December 2025

47 Pages

The Impact Mechanism of Different Government Subsidy Methods on Emission Reduction Decisions in the Petroleum Supply Chain

,
and
1
School of Economics and Management, Southwest Petroleum University, Chengdu 610500, China
2
The Institute for Artificial Intelligence and Sustainable Decision-Making, Southwest Petroleum University, Chengdu 610500, China
3
School of Economics and Management, Southwest Jiaotong University, Chengdu 610032, China
*
Author to whom correspondence should be addressed.

Abstract

Global climate change presents severe challenges to both ecosystems and human societies, necessitating profound structural adjustments in high-emission industries, such as petroleum. This paper develops a two-tier supply chain model comprising a single raw material supplier (leader) and a single refining manufacturer (follower) to examine the effects of R&D cost subsidies and per-unit emission reduction subsidies on supply chain equilibrium decisions and social welfare under three cooperation modes: non-cooperative, cost-sharing, and full cooperation. The results indicate that both types of subsidies effectively elevate wholesale prices, retail prices, emission reduction levels, and supply chain profits. However, per-unit emission reduction subsidies prove more effective in stimulating short-term emission reductions by manufacturers, whereas R&D cost subsidies are more conducive to fostering long-term technological innovation and enhancing emission reduction benefits across the supply chain. Furthermore, social welfare is significantly higher under the R&D cost subsidy strategy than under the per-unit emission reduction subsidy. The study also demonstrates that the full cooperation mode achieves optimal emission reduction performance in the supply chain, and the R&D cost subsidy strategy strengthens overall supply chain profitability and market competitiveness. When designing subsidy policies, governments should account for the roles, technological capabilities, and cost structures of supply chain members, prioritize full cooperation modes, and mitigate risks associated with cost-sharing models under high subsidy intensities to maximize policy effectiveness.

1. Introduction

Global climate change has evolved into an undeniable global issue, with the frequent occurrence of extreme weather events, rising sea levels, and deteriorating ecosystems profoundly impacting the Earth’s ecological systems and human society’s developmental trajectory [1]. This change not only threatens human quality of life and living environments but also imposes immense pressure on the sustainable development of the global economy [2]. Against this backdrop, the international community has widely recognized that addressing climate change and reducing greenhouse gas emissions—particularly carbon dioxide emissions—is a shared responsibility of all humanity. The signing of international agreements, such as the Paris Agreement, marks a crucial step in global efforts to combat climate change [3]. These agreements set specific emission reduction targets for various countries and accelerate the global process of achieving carbon neutrality. However, for high-carbon-emitting industries, such as petroleum enterprises, this transformation task is particularly arduous [4]. Petroleum enterprises have long relied on fossil fuels, placing them under enormous pressure to reduce emissions. On one hand, the urgent global demand for carbon emission reduction requires petroleum enterprises to adopt effective measures to reduce their carbon footprint from the source—such as upgrading production processes, improving energy efficiency, and developing low-carbon technologies. However, these measures demand substantial capital investment and technical support, which constitute a heavy burden for petroleum enterprises. On the other hand, the rapid development and popularization of new energy technologies (e.g., solar energy, wind energy, and the widespread application of electric vehicles) are gradually undermining the market competitiveness of petroleum products, reducing the demand for such products, and forcing petroleum enterprises to confront the severe reality of eroding market share [5,6]. To sustain competitiveness and long-term development, oil companies must actively invest in the new energy sector, which further intensifies the challenges of transformation. In this context, they not only need to respond to drastic external changes but also implement profound internal strategic adjustments, including production optimization, product quality improvement, and enhanced technological innovation. Nevertheless, it is difficult for individual companies to accomplish this task alone, making adjustments to supply chain cooperation mechanisms essential [7,8].
In recent years, participants in petroleum supply chains have increasingly emphasized sustainability, seeking to lower their carbon footprint through technological innovation, resource sharing, and process optimization [9]. Emerging trends include establishing green supply chain management systems to promote environmental information exchange and technical collaboration between upstream and downstream firms, jointly investing in low-carbon technologies and clean energy solutions, implementing carbon trading mechanisms to incentivize emission reduction and optimize resource allocation, and strengthening cooperation with international organizations and multinational enterprises to address global climate challenges [10]. Governments play a critical role in this process by formulating and implementing subsidy policies—such as direct subsidies, tax incentives, R&D support, and carbon trading subsidies—to encourage oil companies to invest in clean energy, energy-saving technologies, and emission reduction initiatives, thereby enhancing the sustainability of their supply chains [11,12]. However, designing and implementing effective subsidy policies is highly complex, as different approaches vary in terms of incentive effects, operational complexity, and sustainability, creating diverse impacts on the emission reduction decisions of supply chain actors. For example, cost subsidies may stimulate R&D enthusiasm but also risk fostering overreliance on government support and weakening market mechanisms [13,14]. Although per-unit production emission reduction subsidies can precisely incentivize abatement, determining appropriate subsidy standards to prevent market distortions remains a major challenge [15]. Therefore, in the context of carbon reduction, optimizing both internal cooperation models within petroleum supply chains and government subsidy mechanisms constitutes a critical force for advancing the low-carbon transformation of the energy industry [16,17]. Studying these dynamics and their effects is thus of great importance for designing sound energy policies and promoting sustainable economic development.
The objective of this study is to investigate how different subsidy mechanisms influence carbon reduction decisions within the petroleum supply chain. Specifically, it addresses the following research question: how do government subsidies—such as R&D cost subsidies versus per-unit emission reduction subsidies—affect the decision-making of supply chain participants, and what role do different cooperation modes (non-cooperative, cost-sharing, and full cooperation) play in shaping these decisions? To address this question, the study employs game theory modeling, a methodological approach widely used to analyze strategic interactions among multiple stakeholders, including governments, manufacturers, and consumers. Game theory is particularly suitable for this research, as it enables the simulation of stakeholder interactions under various subsidy schemes, leading to equilibrium outcomes that can inform policy design [18]. This method has been extensively applied across multiple sectors to examine both competitive and cooperative decision-making—for instance, in modeling price competition in electricity markets and developing cooperative emission reduction strategies in international trade [19,20]. In the context of the petroleum supply chain, game theory facilitates a detailed analysis of the interactions among suppliers, manufacturers, and governments in response to subsidies and carbon reduction initiatives. By applying this framework, the study models how stakeholders make strategic decisions under two specific subsidy mechanisms—R&D cost subsidies and per-unit emission reduction subsidies—and compares their effectiveness in promoting emission reductions and influencing economic outcomes. Additionally, the study examines how different cooperation modes (non-cooperative, cost-sharing, and full cooperation) affect overall emission reduction performance and economic returns. Although existing literature has explored the role of government subsidies in advancing environmental sustainability, few studies have systematically compared the effects of these subsidy mechanisms within the petroleum supply chain. Moreover, while some research has investigated cooperation among supply chain members under government incentives, there remains a gap in the systematic evaluation of how different cooperation modes influence subsidy effectiveness and emission reduction outcomes. This study aims to fill these gaps by developing a two-level petroleum supply chain model, comparing the impacts of both subsidy strategies, and offering theoretical insights to assist oil companies in making informed emission reduction investment decisions, while supporting the design and implementation of effective government subsidy policies.
The main structure of this paper is as follows: Section 1 of the paper introduces the research background and describes the purpose of the work and its significance. Section 2 summarizes the relevant literature and presents the contributions of scholars in this field and the research innovation of this paper. Section 3 provides the problem description and assumptions’ conditions. Subsequently, Section 4 introduces the decision-making model among supply chain members under government R&D cost subsidies. In Section 5, the paper introduces the decision-making model among supply chain members under government unit product emission reduction subsidy. In Section 6, the paper compares the impact of different government subsidies on the supply chain’s environmental performance and social welfare. Finally, the results are illustrated with a numerical example.

2. Literature Review

As a key means to achieve the “dual carbon” goals (carbon peaking and carbon neutrality) and sustainable development, the development of green supply chains is inseparable from the guidance and support of government policies [21,22]. Government subsidies, as a common incentive measure, influence the green behavior decisions of supply chain entities through direct or indirect financial support [23,24,25], thereby promoting emission reduction for enterprises or supply chains [26,27]. Our study aims to relate two streams of literature: one is the research on supply chain decision-making under government subsidy policies, and the other is the impact of supply chain contract design on decision-making.

2.1. Supply Chain Decision-Making Under Government Subsidy Policies

How government subsidies affect the formulation of optimal supply chain decisions is an important issue that has attracted widespread attention from scholars. Existing studies generally conduct research mainly from two perspectives: subsidy recipients [28,29,30] and subsidy types [31,32]. Regarding subsidy targets or objects, existing research mainly covers three types of entities, including manufacturers, consumers, and mixed subsidies. The first scenario centers on subsidizing manufacturers, which is currently the most concentrated area of research. For instance, Meng et al. [33] used Stackelberg game theory to explore a dual-channel green supply chain model that considers consumer green preferences and channel preferences. Research has found that government subsidies to manufacturers lower product prices, promote the sale of green products, benefit manufacturers, and affect retailers’ profits based on subsidy levels. Yang et al. [34] developed a game model involving a government and two competing enterprises, examining the role of government subsidies in improving green product technology. They found that production subsidies for manufacturers effectively motivate technological upgrades, expanding market share for green products and resolving the prisoner’s dilemma, promoting mutual profit improvement. Barman et al. [35] proposed a three-layer green supply chain model with a dual-channel structure, analyzing optimal pricing strategies under both subsidized and non-subsidized conditions. Using the Stackelberg game method, they discovered that government subsidies to manufacturers reduce green product costs, benefiting both manufacturers and suppliers, and enhancing the competitiveness of green products. Li et al. [36] analyzed recycling and emission reduction models under the dual constraints of carbon taxes and government subsidies, finding that manufacturer subsidies can effectively promote enterprises to invest in green technologies, but their effectiveness is constrained by carbon tax rates. In the context of consumer subsidies, Bian et al. [37] compared the effects of subsidizing consumers versus manufacturers on supply chain emission reduction levels, concluding that consumer subsidies are more conducive to driving emission reduction by directly stimulating market demand for green products. Wang et al. [38] primarily investigated the impact of government subsidies on consumer demand for green products within a green supply chain, with a specific focus on how demand forecast information sharing between the retailer and the manufacturer influences the supply chain decisions. They analyzed the dynamics of a decentralized supply chain where the retailer has private demand forecast information, and the potential benefits and challenges of information sharing in a context where the government subsidizes green products to stimulate consumption. Lai et al. [39] believed that consumer subsidies may improve environmental and economic performance in the supply chain compared to producer exclusive subsidies but may not always maximize social welfare when the unit costs associated with green innovation for supply chain members are low. Another scenario extends the research perspective to include multiple stakeholders simultaneously and compares the differences among different subsidy recipients. In the comparison of different subsidy recipients, most scholars have found that the government tends to subsidize manufacturers more in the supply chain field; for example, Meng et al. [40] identified three subsidy scenarios and found that governments prefer subsidizing core manufacturers at the supply chain level. Subsidizing manufacturers enhances green innovation, benefiting the environment, economy, and social welfare. As subsidies and the proportion of green consumers rise, the government favors manufacturer subsidies for greater environmental and economic benefits. Xiao et al. [41] also found that subsidizing manufacturers outperforms retailer subsidies in promoting green quality and profit, especially with direct sales, while retailer subsidies may harm supply chain profits. Some scholars also believe that subsidies for manufacturers and retailers have different effects. For instance, Wen and Liu [42] constructed a differential game model of the secondary pharmaceutical supply chain consisting of pharmaceutical enterprises and medical institutions in the context of centralized drug procurement policy. They compared four subsidy strategies and found that the government’s subsidies to pharmaceutical companies have a significant impact on their low-carbon investments and low-carbon marketing of medical institutions, while subsidies to medical institutions increase the demand and profits for low-carbon drugs. He et al. [43] studied the interactive effects of subsidies and corporate social responsibility on omnichannel supply chain (OSC) emissions. Research has found that different subsidies have different impacts on supply chain profits. Li [44] explored a manufacturer’s strategies for acquiring information and subsidizing retailers in a supply chain with competing retailers having private demand information. They found that subsidizing retailers is always beneficial for the manufacturer, while direct information acquisition is profitable only when the cost is low. Xia et al. [45] built a game theoretical model that includes an ordinary supplier, a low-carbon supplier, and a retailer. They found that the government’s subsidy policy depends on the abatement level of the product. Wang et al. [46] considered four potential subsidy strategies for the government’s supply chain: supplier subsidies, manufacturer subsidies, consumer subsidies, and low-carbon infrastructure subsidies, and analyzed the applicability of each subsidy.
In terms of government subsidy types, existing literature mainly analyzes cost subsidies, R&D subsidies, innovation subsidies, etc., and compares the above issues. Li et al. [47] considered the impact of three subsidy strategies on social welfare, namely, no green subsidies, green product subsidies, and green innovation subsidies. They indicated that green product subsidies are not always the best choice for the government, but in terms of maximizing subsidy efficiency, green innovation subsidies are always superior to green product subsidies. Chen et al. [48] analyzed government subsidies (per-unit production subsidy and innovation effort subsidy) in a research joint venture (RJV) between a manufacturer and retailer. They found that the government should not use both subsidies simultaneously, that it is indifferent to RJV formation when effort subsidies are present, and that manufacturers may be worse off initiating the RJV under certain conditions. Shi and Min [49] demonstrated that a product unit-based subsidy is more effective than a lump sum remanufacturing system implementation subsidy if our goal is to encourage a longer operational period of the remanufacturing system. Li et al. [50] explored the impact of two types of government subsidies—based on fixed investment costs and emission reductions—on green decision-making in the supply chain under the cap-and-trade mechanism. The study found that while government subsidies can promote green marketing cooperation, they are not sufficient to ensure investment in green technologies and emission reduction. Zheng et al. [51] analyzed subsidy modes for new energy vehicles, comparing product, innovation, and product–innovation subsidies. They found that both the government and firms prefer product–innovation subsidies, with trade-offs in subsidy payments and pricing strategies. Investment strategies correlate with consumer recognition and R&D capabilities, with innovation subsidies becoming more viable as these factors grow. Yi et al. [52] studied the impact of bank financing without government subsidies, bank financing with interest subsidies, and bank financing with direct subsidies on the decision-making of supply chain members in the agricultural supply chain.

2.2. The Impact of Supply Chain Contract Design on Decision-Making

Supply chain members typically employ cost-sharing contracts in supply chain management to address the double marginalization problem caused by information asymmetry among decentralized members [53,54]. In their study, Ma et al. [24] evaluated the impact of government subsidy policies on the investment decisions and cooperation of members in manufacturing supply chains. They proposed a cost-sharing contract and found that under such an arrangement, carbon emission reduction and supply chain performance consistently improved. Herbon and David [55] analyzed an integrated inventory supply chain, exploring the optimal production lot, production rate, and shipment cost-sharing ratio between manufacturers and retailers. It demonstrates that a specific shipment cost-sharing ratio can achieve total costs similar to those in an integrated model, providing insights and optimal solutions through mathematical analysis. Zhou et al. [56] studied a three-tier supply chain consisting of suppliers, manufacturers, and retailers, and proposed two types of unilateral cost-sharing contracts. The results show that through reasonable cost-sharing cooperation strategies, the supply chain can achieve Pareto improvement. Xu et al. [57] explored contract design between manufacturers and dealerships for low-carbon products, proposing one-way and two-way cost-sharing contracts to promote low-carbon efforts. They analyzed how various factors affect their cooperation and offered insights for designing long-term supply chain contracts. Furthermore, some scholars have found that cooperative contracts among supply chain members yield better outcomes in terms of both profitability and environmental performance compared to cost-sharing contracts. For example, Cheng and Zhang [58] showed that compared to decentralized decision-making, centralized decision-making has improved the overall emission reduction level of the supply chain. Sinayi and Rasti-Barzoki [59] presented a two-tier supply chain model with government intervention, focusing on economic, social, and environmental sustainability. It found that manufacturer–retailer cooperation leads to greener products, higher profits, and increased consumer surplus. Government policies, including taxes and subsidies, significantly impact profits and sustainability. Wang et al. [60] studied the emission reduction problem of the construction supply chain under government subsidies, explored differential game models of centralized decision-making, decentralized decision-making, and cost-sharing contracts, and compared the optimal feedback equilibrium strategies of both parties under three decision-making scenarios. The study found that the total profit and total emission reduction level of the supply chain under centralized decision-making were influenced by decentralized decision-making and cost-sharing contracts.

2.3. Research Gaps

Although the academic community has carried out a number of studies on government subsidy and contract design, there is still room for further expansion, as follows: (1) Government subsidies play a crucial role in emission reduction incentives within the petroleum industry. However, previous studies have primarily focused on general supply chains or agricultural supply chains, with most lacking analysis specific to the petroleum supply chain. This research expands the scope of government subsidies to include the petroleum supply chain and further examines the effects of subsidy standards. (2) Most prior studies extensively discussed government subsidies and supply chain contract design separately, and there is limited exploration of how different subsidy types interact with specific contract models in promoting long-term green innovation and sustainability within supply chains. This study bridges the research gap by integrating the impact of government subsidies with supply chain contract design, investigating how R&D cost subsidies and per-unit emission reduction subsidies affect supply chain equilibrium decisions and social welfare under non-cooperative, cost-sharing, and cooperative models. These characteristics reflect the main variations between the present study and previous research.

3. Problem Description and Assumptions

3.1. Problem Description

The petroleum supply chain exhibits a significant power asymmetry: upstream suppliers dominate the supply chain through resource monopolies and bargaining advantages, while small and medium-sized refining manufacturers remain constrained by costs and strategically passive [6]. Against this backdrop, this paper constructs a two-echelon supply chain game-theoretic model comprising a dominant raw material supplier (Stackelberg leader) and a follower refining manufacturer (Stackelberg follower). As the leader, the supplier controls decision-making by setting wholesale prices and emission reduction requirements, while the manufacturer, as the follower, responds by determining retail prices and adjusting its own emission reduction investments. The government, as a key driver of green transformation, provides two types of subsidies to incentivize manufacturers’ emission reduction behaviors: R&D-based subsidies and per-unit emission reduction subsidies. Within this policy framework, supply chain members need to adjust their cooperation strategies according to the subsidy type, including non-cooperative strategies, cost-sharing strategies, and cooperative strategies. Non-cooperative mode represents the scenario where the supply chain members make decisions independently without any formal agreement or cost-sharing arrangements. Full cooperation mode illustrates the case where both parties work together, sharing both the benefits and costs, aiming to maximize the overall system efficiency. Cost-sharing mode is an intermediate scenario, where the partners share specific costs (such as emission reduction or R&D expenses), balancing the need for cooperation with individual cost control. Although suppliers hold the dominant position in the game, manufacturers can still achieve strategic initiative by sharing part of the suppliers’ emission reduction costs. This cost-sharing mechanism is well-justified: firstly, it incentivizes suppliers to enhance emission reduction efforts, improving product green competitiveness and stimulating market demand; secondly, it enables manufacturers to obtain wholesale price concessions, thereby reducing procurement costs; finally, from a long-term perspective, it helps shape a green brand image, securing sustainable competitive advantages in the low-carbon market [10].
Therefore, this study develops decision-making models for the petroleum supply chain under three scenarios: non-cooperation, cost-sharing, and full cooperation. Using Stackelberg game methodology, it conducts a comparative analysis of the differential impacts of two subsidy mechanisms (R&D cost subsidies and per-unit emission reduction subsidies) on the emission reduction behaviors, profit distribution, and overall welfare of both suppliers and manufacturers.

3.2. Basic Assumptions

Due to the complexity and uncertainty of many factors in the real world, it is simplified here, and a series of reasonable assumptions are made for the corresponding model construction and decision-making problems:
Assumption 1. 
In a secondary green supply chain consisting of a single raw material supplier and a single refining manufacturer, the supplier is the leader and the manufacturer is the follower, and both parties make decisions with the goal of optimizing their own profits.
Assumption 2. 
According to the widely adopted demand function in existing literature [61,62], we assume that the market demand is linear to the emission reduction effect of the final product, which is given by:
d = a − b p + α x m + β x s .
In the model, a denotes the potential aggregate demand. The variables x m and x S stand for the emission reduction levels undertaken by the manufacturer and the supplier, respectively. Their effectiveness in translating these reductions into increased market demand is measured by coefficients α and β , under the given condition that α > β > 0 [63]. This critical assumption captures the greater consumer appeal and market premium often associated with greener upstream production inputs compared to internal manufacturing efficiencies.
Assumption 3. 
In the petroleum product supply chain, assume that the wholesale price is set at ω based on the green emission reduction production cost, and the refining manufacturer sets the retail price at p based on market and consumer capacity, and then the unit production cost of the raw material supplier in actual production is c .
Assumption 4. 
Following the literature [64], we assume that the production cost per-unit is identical for suppliers and manufacturers when producing emission-reducing products. However, undertaking emission reduction requires additional R&D investment. Consistent with prior studies [65,66], we model the R&D investment cost as a quadratic function of the emission reduction level to reflect increasing marginal costs. Specifically, the supplier’s emission reduction investment cost is   E s = 1 2 μ s x s 2 , and the manufacturer’s is  E m = 1 2 μ m x m 2 , where  μ s  and  μ m  are the marginal cost coefficients for their respective emission reduction investments. These cost functions satisfy the standard economic properties of increasing marginal costs [67]:
E s ′ ( x s ) > 0 , E s ″ ( x s ) > 0 , E m ′ ( x m ) > 0 , E m ″ ( x m ) > 0 .
This convexity assumption ensures that achieving higher emission reduction levels becomes progressively more expensive.
Assumption 5. 
In order to encourage suppliers to invest in carbon reduction, manufacturers share the cost of carbon reduction of suppliers according to the proportion k ( 0 < k < 1 ) . In this case, the supplier only needs to bear the proportional ( 1 − k ) carbon reduction cost [68].
Assumption 6. 
To incentivize emission reduction, the government provides financial subsidies to the manufacturer.We consider two distinct subsidy strategies. (1) R&D cost-sharing subsidy: The government subsidizes a fraction ε ( 0 < ε < 1 )  of the manufacturer’s R&D investment cost. This direct subsidy is aimed at enhancing the manufacturer’s willingness to invest in green technologies [69]. The resulting government expenditure is  1 2 ε μ m x m 2 . (2) Per-unit emission reduction subsidy: The government provides a subsidy of  σ  per-unit of emission reduction achieved by the manufacturer [48]. The total subsidy expenditure is thus  σ x m , where  σ ( 0 < σ < 1 )  is the unit subsidy coefficient. The theoretical subsidy mechanisms discussed in this paper find their parallels in real-world policies. For instance, R&D cost subsidies are reflected in the United States’ tax credit policies for green technology innovation and China’s direct funding for key emission-reduction R&D projects under its Five-Year Plans [70]. These policies aim to lower the entry barrier for innovation, aligning with our model’s assumption. Similarly, the logic of per-unit emission reduction subsidies can be observed in Australia’s petroleum industry. In Australia refiners receive rebates reducing their per ton CO2 cost, showing a real-world linkage between unit emissions and financial incentives [71].

4. Establishment and Solution of Supply Chain Model Under R&D Cost Subsidy

4.1. Non-Cooperative Decisions by Supply Chain Members Under R&D Cost Subsidies

According to the analysis and assumptions in Section 3, when the supplier and the manufacturer are not cooperating, the supplier’s profit π s 21 and the manufacturer’s profit π m 21 are expressed as, respectively:
π s 21 = ( ω − c ) ( a − b p + α x m + β x s ) − 1 2 μ s x s 2 ,
π m 21 = ( p − ω ) ( a − b p + α x m + β x s ) − 1 2 ( 1 − ε ) μ m x m 2 .
First, the manufacturer’s reaction function for the sales price is derived. Subsequently, based on this function, the supplier determines its emission reduction level and optimal wholesale price to maximize its own profit. Substituting these equilibrium values back into the manufacturer’s reaction function yields the optimal sales price. Based on the solution method of Ma et al. [24], Lemma 1 can be derived as follows:
Lemma 1. 
For a supply chain under non-cooperative decision-making with R&D cost subsidies, the optimal wholesale price, sales price, supplier emission reduction, manufacturer emission reduction, supplier profit, and manufacturer profit are given by:
ω 21 * = ( ε − 1 ) μ m [ 2 ( a + b c ) μ s − β 2 c ] + ( a + b c ) α 2 μ s ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s ,
p 21 * = ( ε − 1 ) μ m [ ( 3 a + b c ) μ s − β 2 c ] + ( a + b c ) α 2 μ s ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) − α 2 μ s ,
x s 21 * = ( ε − 1 ) μ m β ( a − b c ) ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s ,
x m 21 * = − μ s α ( a − b c ) ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s ,
π s 21 * = ( ε − 1 ) μ m μ s ( a − b c ) 2 [ ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) − α 2 μ s ] 2 ,
π m 21 * = [ 2 b ( ε − 1 ) μ m + α 2 ] ( ε − 1 ) ( a − b c ) 2 μ m μ s 2 2 [ ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s ] 2 .
Among them, the superscript 21 represents the non-cooperative scenario between the supplier and the manufacturer under government R&D cost subsidies.
The proof of Lemma 1 is shown in Appendix A.
Based on the above optimal results, we can derive Corollary 1:
Corollary 1. 
Under the non-cooperative mode with government R&D cost subsidies, (1) an increase in the cost coefficient of emission reduction diminishes the emission reduction efforts of both the supplier and the manufacturer, thereby lowering the overall supply chain profit. (2) Conversely, a higher-influence coefficient of emission reduction on demand enhances their emission reduction efforts and increases the total supply chain profit. (3) As the government subsidy coefficient rises, the emission reduction levels and profits of both the supplier and the manufacturer increase correspondingly.
The simple derivation of Corollary 1 is as follows: Only when ω 21 ∗ , p 21 ∗ , x s 21 ∗ , x m 21 ∗ , π s 21 ∗ and π m 21 ∗ are greater than 0 will the manufacturer and supplier engage in production. Therefore, the results follow directly from the first-order conditions of the profit maximization problem. By finding the first-order derivative in Equations (4)–(9), it can be seen that the first-order derivatives are all less than 0:
∂ x s 21 * ∂ μ s < 0 , ∂ x m 21 * ∂ μ s < 0 , ∂ x s 21 * ∂ μ m < 0 , ∂ x m 21 * ∂ μ m < 0 ∂ π s 21 * ∂ μ s < 0 , ∂ π m 21 * ∂ μ s < 0 , ∂ π s 21 * ∂ μ m < 0 , ∂ π m 21 * ∂ μ m < 0 .
Taking the first-order derivatives of α and β , we can see that their first-order derivatives are all greater than 0:
∂ x s 21 * ∂ α > 0 , ∂ x m 21 * ∂ α > 0 , ∂ x s 21 * ∂ β > 0 , ∂ x m 21 * ∂ β > 0 ∂ π s 21 * ∂ α > 0 , ∂ π m 21 * ∂ α > 0 , ∂ π s 21 * ∂ β > 0 , ∂ π m 21 * ∂ β > 0 .
By taking the first derivative of the optimal emission reduction results and profits of suppliers and manufacturers with respect to ε , we can conclude that:
∂ x s 21 ∗ ∂ ε > 0 , ∂ x m 21 ∗ ∂ ε > 0 , ∂ π s 21 ∗ ∂ ε > 0 , ∂ π m 21 ∗ ∂ ε > 0 .
Therefore, Corollary 1 is established.
The results demonstrate that (1) emission reduction performance in the petroleum supply chain demonstrates an inverse relationship with the marginal cost coefficient of abatement investments. As marginal costs rise, refiners encounter amplified challenges in deploying emission reduction R&D, necessitating substantially higher capital commitments. Concurrently, raw material suppliers reduce their abatement efforts when facing elevated marginal costs, thereby diminishing the overall effectiveness of emission reduction initiatives across the supply chain. Consequently, the increasing marginal costs systematically suppress emission reduction investments throughout the petroleum supply chain. Furthermore, rising marginal cost coefficients directly inflate operational expenses for both manufacturers and suppliers, ultimately compressing profit margins. (2) The growing coefficient of emission reduction’s impact on market demand reflects intensified consumer focus on low-carbon products. This expanding green market segment drives petroleum product manufacturers and suppliers to correspondingly increase resource allocation toward clean technology and low-carbon product development [56]. The enhanced market demand subsequently generates improved profitability for both suppliers and manufacturers, creating a self-reinforcing cycle of environmental and economic gains. (3) As government subsidy coefficients for manufacturers’ emission reduction R&D increase, a positive correlation emerges with multiple performance indicators: manufacturers’ emission reduction levels, supplier profits, and manufacturer profits all show synchronous improvement. Government subsidies effectively lower the R&D cost burden for manufacturers, stimulating increased abatement investments and enhanced profitability. Simultaneously, supplier emission reduction levels rise correspondingly with increasing subsidies. Some scholars attribute this spillover effect to the “free-rider” phenomenon [57], where suppliers benefit from the market externalities generated by manufacturers’ improved product sustainability without bearing the full cost of the corresponding R&D investment.

4.2. Supply Chain Members Cooperate in Decision-Making Under R&D Cost Subsidies

Under the cooperative scenario, supply chain members make joint decisions with the objective of maximizing total profit. The specific decision sequence is as follows: First, the central decision-maker of the supply chain simultaneously determines all variables, including the emission reduction levels of both the supplier and the manufacturer, as well as the market price of the product. The optimal decisions are derived through a centralized optimization model, with the profit function expressed as follows:
π t o t a l 22 = ( p − c ) ( a − b p + α x m + β x s ) − 1 2 ( 1 − ε ) μ m x m 2 − 1 2 μ s x s 2 .
Based on the above analysis, Lemma 2 can be expressed as follows:
Lemma 2. 
For a supply chain under cooperative decision-making with R&D cost subsidies, the optimal sales price, supplier emission reduction, manufacturer emission reduction, supplier profit, and supply chain profit are:
p 22 * = [ ( a + b c ) μ s − c β 2 ] ( ε − 1 ) μ m + α 2 c μ s ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + α 2 c μ s ,
x s 22 * = ( ε − 1 ) μ m β ( a − b c ) ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + α 2 c μ s ,
x m 22 * = ( ε − 1 ) μ s α ( a − b c ) ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + α 2 c μ s ,
π t 22 * = ( ε − 1 ) μ m μ s ( a − b c ) ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 2 c μ s .
Among them, the superscript 22 denotes the cooperative scenario between the supplier and the manufacturer under government R&D cost subsidies.
The proof of Lemma 2 is shown in Appendix A.
We can get Corollary 2 from the above results.
Corollary 2. 
Under the fully cooperative model with government R&D cost subsidies for the manufacturer, (1) an increase in the marginal cost coefficient of emission reduction lowers the emission reduction levels of both the supplier and the manufacturer, thereby decreasing the total supply chain profit. (2) Conversely, a higher-impact coefficient of emission reduction on demand enhances their emission reduction efforts and strengthens the overall profitability of the supply chain. (3) As the government subsidy coefficient increases, the emission reduction levels of both members and the total supply chain profit see corresponding improvements.
The results follow directly from the first-order conditions of the profit maximization problem. By finding the first derivative of μ s , μ m in Equations (11)–(14), it can be seen that the first derivatives are less than 0:
∂ x s 22 * ∂ μ s < 0 , ∂ x m 22 * ∂ μ s < 0 , ∂ x s 22 * ∂ μ m < 0 , ∂ x m 22 * ∂ μ m < 0 , ∂ π t 22 * ∂ μ s < 0 , ∂ π t 22 * ∂ μ m < 0 .
For the sum of α and β, we can find that the first-order derivatives are all greater than 0:
∂ x s 22 * ∂ α > 0 , ∂ x m 22 * ∂ α > 0 , ∂ x s 22 * ∂ β > 0 , ∂ x m 22 * ∂ β > 0 , ∂ π s 22 * ∂ α > 0 , ∂ π m 22 * ∂ β > 0 .
By taking the first derivative of the optimal emission reduction results and profits of supply chain with respect to ε , we can conclude that:
∂ x s 22 ∗ ∂ ε > 0 , ∂ x m 22 ∗ ∂ ε > 0 , ∂ π t 22 ∗ ∂ ε > 0 .
Therefore, Corollary 2 is established.
The results show that (1) in the scenario where supply chain members cooperate in decision-making under R&D cost subsidies, the level of emission reduction is inversely related to the marginal cost coefficients of both suppliers and manufacturers. An increase in the supplier’s marginal cost raises the unit cost of emission reduction, leading to a decrease in emission reduction investment. Similarly, an increase in the manufacturer’s marginal cost raises R&D expenditure, compresses profits, and reduces the overall emission reduction capacity. However, the cooperative model cannot fully offset these impacts. Ultimately, a simultaneous increase in the marginal costs of both the supplier and manufacturer results in higher overall emission reduction costs and lower total supply chain profits. (2) Similar to the non-cooperative model, as the coefficient representing the impact of emission reduction on demand increases, indicating greater market attention to low-carbon products, the demand for low-carbon products rises. This motivates supply chain members to increase their investment in emission reductions, thereby increasing the overall supply chain profit. (3) Under the cooperative model, both the manufacturer’s and supplier’s emission reduction R&D levels increase with the subsidy amount. The subsidy reduces the manufacturer’s emission reduction costs, increases the marginal return on R&D, and encourages the manufacturer to expand investment in emission reduction. Similarly, the supplier also increases their emission reduction investment. At the same time, the government subsidy reduces the manufacturer’s emission reduction R&D costs, thus improving the overall supply chain profit.

4.3. Cost-Sharing Decisions for Supply Chain Members Under R&D Cost Subsidies

Under the cost-sharing scenario, the supplier invests in emission reduction technology, while the manufacturer bears a portion of the emission reduction technology investment cost for the supplier. Assuming the manufacturer’s cost-sharing ratio is k , and the supplier bears the remaining 1 − k portion of the emission reduction technology investment, with 0 < k < 1 , the profit functions of the supplier and the manufacturer can be expressed as follows:
π s 23 = ( ω − c ) ( a − b p + α x m + β x s ) − ( 1 − k ) 2 μ s x s 2 ,
π m 23 = ( p − ω ) ( a − b p + α x m + β x s ) − 1 2 ( 1 − ε ) μ m x m 2 − 1 2 k μ s x s 2 .
The specific sequence of decisions is as follows: First, the manufacturer determines the cost-sharing ratio for emission reduction investment, its own emission reduction level, and the market price. Subsequently, the supplier decides on the emission reduction level and the wholesale price. Both parties aim to maximize their profits, and the optimal decisions and profits are derived using backward induction. Therefore, Lemma 3 can be obtained as follows:
Lemma 3. 
Under the cost-sharing situation with R&D cost subsidies, the optimal cost-sharing proportion, wholesale price, sales price, supplier emission reduction, manufacturer emission reduction, supplier profit, and manufacturer profit are shown below:
k 23 * = β 2 μ m ( ε − 1 ) + μ s α 2 8 b μ m ( ε − 1 ) μ s + 3 μ s α 2 ,
ω 23 * = 2 [ 8 b 2 c μ s + ( 8 a c μ s − 5 c β 2 ) b − a β 2 ] ( ε − 1 ) 2 μ m 2 + 2 α 2 ( 5 a c μ s − 2 c β 2 + 3 a μ s ) ( ε − 1 ) μ m + 2 c α 4 μ s 4 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 4 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 4 μ s ,
p 23 * = [ 8 b 2 c μ s + ( 8 a c μ s − 5 c β 2 ) b − a β 2 ] ( ε − 1 ) 2 μ m 2 + 2 α 2 ( 5 a c μ s − 2 c β 2 + 3 a μ s ) ( ε − 1 ) μ m + 2 c α 4 μ s 4 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 4 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 4 μ s ,
x s 23 * = [ 8 b ( ε − 1 ) μ m + 3 α 2 ] β ( ε − 1 ) μ m ( a − b c ) 4 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 4 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 4 μ s ,
x m 23 ∗ = − α [ ( 8 b μ s − β 2 ) ( ε − 1 ) μ m + 2 μ s α 2 ] ( a − b c ) 4 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 4 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 4 μ s ,
π s 23 * = 4 ( 8 b μ s − 3 β 2 ) ( ε − 1 ) 2 μ m 2 ( a − b c ) 2 [ b ( ε − 1 ) μ m + 2 α 2 ] [ ( 8 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 2 μ s ] [ 2 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 2 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + α 4 μ s ] 2 ,
π m 23 * = [ ( 8 b μ s − β 2 ) ( ε − 1 ) μ m + 4 μ s α 2 ] ( a − b c ) 2 ( ε − 1 ) μ m 16 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 16 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 8 α 4 μ s .
Among them, the superscript 23 indicates the cost-sharing scenario between the supplier and the manufacturer under government R&D cost subsidies.
The proof of Lemma 3 is shown in Appendix A.
We can get Corollary 3 from the above results.
Corollary 3. 
Under the cost-sharing model with government R&D cost subsidies, (1) an increase in the emission reduction cost coefficient reduces both the emission reduction levels and the individual profits of the supplier and the manufacturer. (2) Conversely, a higher-impact coefficient of emission reduction on demand strengthens their emission reduction efforts and raises their respective profits in the supply chain. (3) An increase in the subsidy coefficient improves both the emission reduction investments and profits of the two members, while correspondingly lowering the cost-sharing coefficient.
From Appendix A, and by finding the first derivative of μ s , μ m in Equations (18)–(23), it can be seen that the first derivative is less than 0:
∂ x s 23 * ∂ μ s < 0 , ∂ x m 23 * ∂ μ s < 0 , ∂ x s 23 * ∂ μ m < 0 , ∂ x m 23 * ∂ μ m < 0 ∂ π s 23 * ∂ μ s < 0 , ∂ π m 23 * ∂ μ s < 0 , ∂ π s 23 * ∂ μ m < 0 , ∂ π m 23 * ∂ μ m < 0 .
For the sum of α and β, we can find that the first-order derivatives are all greater than 0:
∂ x s 23 * ∂ α > 0 , ∂ x m 23 * ∂ α > 0 , ∂ x s 23 * ∂ β > 0 , ∂ x m 23 * ∂ β > 0 ∂ π s 23 * ∂ α > 0 , ∂ π m 23 * ∂ α > 0 , ∂ π s 23 * ∂ β > 0 , ∂ π m 23 * ∂ β > 0 .
Calculating the first derivative of the subsidy coefficient, we can obtain:
∂ x m 23 ∗ ∂ ε > 0 , ∂ x s 23 ∗ ∂ ε > 0 , ∂ π s 23 ∗ ∂ ε > 0 , ∂ π m 23 ∗ ∂ ε > 0 , ∂ k 23 ∗ ∂ ε < 0 .
Therefore, Corollary 3 is established.
The results indicate that (1) consistent with the non-cooperative and cooperative models, under the manufacturer cost-sharing model, both the emission reduction R&D levels and profits of supply chain members show a negative correlation with marginal costs. This indicates that contractual relationships between supply chain members do not mitigate the negative effects caused by increasing marginal costs. (2) Similarly, aligning with the patterns observed in non-cooperative and cooperative models, under the cost-sharing model, the emission reduction R&D levels and profits of supply chain members demonstrate a positive correlation with the emission reduction demand coefficient. (3) While both emission reduction R&D levels and profits of supply chain members increase with higher government subsidy coefficients, we observe that the manufacturer’s cost-sharing coefficient (k) for the supplier decreases as subsidy levels rise. This occurs because government subsidies partially substitute for the manufacturer’s sharing responsibility. Under rational decision-making, the sharing coefficient k naturally declines when the subsidy coefficient reaches a certain threshold [25].

5. Establishment and Solution of Supply Chain Model Under Per-Unit Product Emission Reduction Subsidy

5.1. Non-Cooperative Decision-Making of Supply Chain Members Under Unit Product Emission Reduction Subsidies

The scenario with government per-unit emission reduction subsidies follows the same framework as presented in Section 4. When the supplier and the manufacturer are not cooperating, the supplier’s profit π s 31 and the manufacturer’s profit π m 31 are expressed as, respectively:
π s 31 = ( ω − c ) ( a − b p + α x m + β x s ) − 1 2 μ s x s 2 ,
π m 31 = ( p − ω + σ x m ) ( a − b p + α x m + β x s ) − 1 2 μ m x m 2 .
The solution program for this scenario follows a method similar to that in Section 4. Specifically, the manufacturer’s reaction function for the sales price is first derived. Subsequently, based on this reaction function, the supplier determines its optimal emission reduction level and wholesale price to maximize its profit. Finally, substituting these equilibrium values into the manufacturer’s reaction function yields the optimal sales price. Following the solution method of Ma et al. [24], Lemma 4 is derived as follows:
Lemma 4. 
Under the context of non-cooperative decision-making with per-unit emission reduction subsidies, the optimal wholesale price, sales price, supplier emission reduction, manufacturer emission reduction, supplier profit, and manufacturer profit are:
ω 31 * = ( a + b c ) ( b 2 σ 2 + 2 b α σ + α 2 − 2 b μ m ) μ s + b c μ m β 2 2 b ( b 2 σ 2 + 2 b α σ + α 2 − 2 b μ m ) μ s + b μ m β 2 ,
p 31 * = ( 2 a σ 2 + a c σ − c μ m ) b 2 + ( 3 α σ μ s + α 2 c − 3 a μ m ) b + a α 2 μ s + b c μ m β 2 2 b ( b 2 σ 2 μ s + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ) + b μ m β 2 ,
x s 31 * = − β μ m ( a − b c ) 2 2 ( b 2 σ 2 μ s + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ) + μ m β 2 ,
x m 31 * = − ( α + σ ) μ m ( a − b c ) 2 ( b 2 σ 2 μ s + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ) + μ m β 2 ,
π s 31 * = − μ s μ m ( a − b c ) 2 2 ( 2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 ) ,
π m 31 * = − ( b 2 σ 2 + 2 b α σ + α 2 − 2 b μ m ) μ s 2 μ m ( a − b c ) 2 2 ( 2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 ) 2 .
Among them, the superscript 31 represents the non-cooperative scenario between the supplier and the manufacturer under government per-unit emission reduction subsidies.
The proof of Lemma 4 is shown in Appendix A.
By finding the first derivative of μ s and μ m in Equations (26)–(31), it can be seen that the first derivatives are less than 0:
∂ x s 31 * ∂ μ s < 0 , ∂ x m 31 * ∂ μ s < 0 , ∂ x s 31 * ∂ μ m < 0 , ∂ x m 31 * ∂ μ m < 0 ∂ π s 31 * ∂ μ s < 0 , ∂ π m 31 * ∂ μ s < 0 , ∂ π s 31 * ∂ μ m < 0 , ∂ π m 31 * ∂ μ m < 0 .
For the sum of α and β, we can find that the first-order derivatives are all greater than 0:
∂ x s 31 * ∂ α > 0 , ∂ x m 31 * ∂ α > 0 , ∂ x s 31 * ∂ β > 0 , ∂ x m 31 * ∂ β > 0 ∂ π s 31 * ∂ α > 0 , ∂ π m 31 * ∂ α > 0 , ∂ π s 31 * ∂ β > 0 , ∂ π m 31 * ∂ β > 0 .
The first-order derivatives of the supplier’s and the manufacturer’s emission reduction investments and profits with respect to the government’s per-unit emission subsidy coefficient σ are given as follows:
∂ x s 31 * ∂ σ > 0 , ∂ x m 31 * ∂ σ > 0 , ∂ π s 31 * ∂ σ > 0 , ∂ π m 31 * ∂ σ > 0 .
Therefore, Corollary 4 is established.
Corollary 4. 
Under a government subsidy program for the manufacturer’s per-unit emission reduction with non-cooperative supply chain members, (1) an increase in the impact coefficient of emission reduction on demand enhances the emission reduction efforts of both the supplier and manufacturer, thereby increasing their respective profits. (2) Conversely, a higher emission reduction cost coefficient reduces their emission reduction efforts and corresponding profits. (3) As the government subsidy coefficient increases, both members’ emission reduction levels and respective profits see a corresponding rise.
The conclusions drawn from Corollary 4 are consistent with the analytical framework established in Corollary 1 (Section 4.1), reinforcing the same fundamental relationships.

5.2. Cooperative Decision-Making Among Supply Chain Members Under Per-Unit Product Emission Reduction Subsidies

Under the scenario of government per-unit product emission reduction subsidies with cooperative supply chain members, the supply chain makes joint decisions with the objective of maximizing total profit. The supply chain profit function is expressed as follows:
π t o t a l 32 = ( p − c + σ x m ) ( a − b p + α x m + β x s ) − 1 2 μ m x m 2 − 1 2 μ s x s 2 .
Similar to the solution approach in Section 4.2, we can obtain Lemma 5:
Lemma 5. 
Under the context of cooperative decision-making with per-unit product emission reduction subsidies, the optimal sales price, supplier emission reduction, manufacturer emission reduction, supplier profit, and supply chain profit are:
p 32 * = [ ( a σ 2 + α c σ − c μ m ) b + a α σ + α 2 c − a μ m ] μ s + c β 2 μ m μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s - 2 b μ s μ m + β 2 μ m ,
x s 32 * = − β ( a − b c ) μ m μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m + β 2 μ m ,
x m 32 * = − ( α + σ ) ( a − b c ) μ s μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m + β 2 μ m ,
π t 32 * = − ( a − b c ) 2 μ s μ m 2 [ μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m + β 2 μ m ] .
Among them, the superscript 32 denotes the cooperative scenario between the supplier and the manufacturer under government per-unit emission reduction subsidies.
The proof of Lemma 5 is shown in Appendix A.
By finding the first derivative of μ m , μ s in Equations (33)–(36), it can be seen that the first derivatives are less than 0:
∂ x s 32 * ∂ μ s < 0 , ∂ x m 32 * ∂ μ s < 0 , ∂ x s 32 * ∂ μ m < 0 , ∂ x m 32 * ∂ μ m < 0 , ∂ π s 32 * ∂ μ s < 0 , ∂ π m 32 * ∂ μ s < 0 , ∂ π s 32 * ∂ μ m < 0 , ∂ π m 32 * ∂ μ m < 0 .
For the sum of α and β, we can find that the first-order derivatives are all greater than 0:
∂ x s 32 * ∂ α > 0 , ∂ x m 32 * ∂ α > 0 , ∂ x s 32 * ∂ β > 0 , ∂ x m 32 * ∂ β > 0 ∂ π s 32 * ∂ α > 0 , ∂ π m 32 * ∂ α > 0 , ∂ π s 32 * ∂ β > 0 , ∂ π m 32 * ∂ β > 0 .
Taking the first-order derivatives of the supply chain members’ emission reduction investments and the supply chain profit with respect to the subsidy coefficient σ yields:
∂ x s 32 * ∂ σ > 0 , ∂ x m 32 * ∂ σ > 0 , ∂ π t 32 * ∂ σ > 0 .
Therefore, Corollary 5 is established.
Corollary 5. 
Under a government per-unit product emission reduction subsidy and with supply chain members in a cooperative mode, (1) an increase in the impact coefficient of emission reduction on demand enhances the emission reduction efforts of both the supplier and the manufacturer, thereby increasing the overall supply chain profit. (2) Conversely, a rise in the emission reduction cost coefficient dampens these efforts and diminishes the supply chain profit. (3) Furthermore, an increase in the government subsidy coefficient correspondingly raises the emission reduction levels of both members and the total supply chain profit.
These findings align with the conclusions established in Corollary 2. Therefore, a detailed analysis is omitted here to avoid redundancy.

5.3. Cost-Sharing Decisions for Supply Chain Members Under Unit Product Emission Reduction Subsidies

Under the cost-sharing scenario with government per-unit product emission reduction subsidies, the supplier implements emission reduction measures while the manufacturer bears a portion of the associated emission reduction costs. We can obtain the profit functions of suppliers and manufacturers separately, as follows.
π s 33 = ( ω − c ) ( a − b p + α x m + β x s ) − ( 1 − k ) 2 μ s x s 2 ,
π m 33 = ( p − ω + σ x m ) ( a − b p + α x m + β x s ) − 1 2 μ m x m 2 − 1 2 k μ s x s 2 .
The decision-making sequence follows the same structure as in Section 4.3: First, the manufacturer determines the cost-sharing ratio for emission reduction investment, its own emission reduction level, and the market price. Subsequently, the supplier decides on its emission reduction level and the wholesale price. Using backward induction, Lemma 6 can be derived as follows:
Lemma 6. 
Under the cost-sharing situation with per-unit emission reduction subsidies, the optimal cost-sharing proportion, wholesale price, sales price, supplier emission reduction, manufacturer emission reduction, supplier profit, and manufacturer profit are:
k 33 * = − β 2 μ m 4 μ s ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) ,
ω 33 * = β 2 μ m ( 5 b c + a ) + 4 ( a + b c ) ( b 2 σ 2 μ s + 2 b α σ μ s − 2 b μ m μ s + α 2 μ s ) 2 b ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ,
x s 33 * = − 2 ( a − b c ) β μ m 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ,
p 33 * = 4 ( 2 a σ 2 + c α σ - c μ m μ s ) μ s σ 2 b 4 + [ ( 12 c α 2 σ 2 + ( 28 α σ 3 - 16 c μ m σ ) α - 28 a μ m σ 2 + 8 c μ m 2 ) μ s + 4 c β 2 μ m σ 2 ] b 3 + [ ( 12 c α 3 σ + ( 36 a σ 2 − 12 c μ m ) α 2 − 48 a α μ m σ + 24 a μ m 2 ) μ s + ( 2 a σ 2 + 9 c μ m ) β 2 μ m ] b 2 + [ ( 20 c α 3 σ + 4 c α 4 - 20 a α 2 μ m ) μ s + β 2 ( 3 a α σ − 5 c α 2 − a μ m ) μ m ] b + a α 2 ( 4 σ 2 μ s + β 2 μ m ) 8 b ( 4 b 2 σ 2 μ s + 8 b α σ μ s - 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ - 2 b μ m + α 2 ) ,
x m 33 * = − ( a − b c ) ( b σ + α ) [ ( 4 b 2 σ 2 + 8 b α σ − 8 b μ m + 4 α 2 ) μ s + β 2 μ m ] 2 [ ( 4 b 2 σ 2 + 8 b α σ − 8 b μ m + 4 α 2 ) μ s + 3 β 2 μ m ] ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) ,
π s 33 * = − ( a − b c ) 2 μ m ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) 8 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) ,
π m 33 * = − ( a − b c ) 2 μ m ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) 4 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) .
Among them, the superscript 33 represents the cost-sharing scenario between the supplier and the manufacturer under government per-unit emission reduction subsidies.
The proof of Lemma 6 is shown in Appendix A.
By finding the first derivative of μ s and μ m in the above results, it can be seen that the first derivative is less than 0:
∂ x s 33 * ∂ μ s < 0 , ∂ x m 33 * ∂ μ s < 0 , ∂ x s 33 * ∂ μ m < 0 , ∂ x m 33 * ∂ μ m < 0 ∂ π s 33 * ∂ μ s < 0 , ∂ π m 33 * ∂ μ s < 0 , ∂ π s 33 * ∂ μ m < 0 , ∂ π m 33 * ∂ μ m < 0 .
For the sum of α and β, we can find that the first-order derivatives are all greater than 0:
∂ x s 33 * ∂ α > 0 , ∂ x m 33 * ∂ α > 0 , ∂ x s 33 * ∂ β > 0 , ∂ x m 33 * ∂ β > 0 ∂ π s 33 * ∂ α > 0 , ∂ π m 33 * ∂ α > 0 , ∂ π s 33 * ∂ β > 0 , ∂ π m 33 * ∂ β > 0 .
By differentiating the supply chain members’ emission reduction investments and profit with respect to the subsidy coefficient σ , we derive the following first-order conditions:
∂ x s 33 * ∂ σ > 0 , ∂ x m 33 * ∂ σ > 0 , ∂ π s 33 * ∂ σ > 0 , ∂ π m 33 * ∂ σ > 0 , ∂ k 33 * ∂ σ < 0 .
Therefore, Corollary 6 is established.
Corollary 6. 
Under a government per-unit product emission reduction subsidy with a cost-sharing model among supply chain members, (1) an increase in the demand coefficient for emission reduction enhances both members’ reduction efforts and their respective profits. (2) Conversely, a higher marginal cost coefficient for emission reduction diminishes their reduction levels and individual profits. (3) An increase in the government subsidy coefficient raises both members’ reduction levels and respective profits, while correspondingly lowering the cost-sharing coefficient.
The findings reveal a consistent influence mechanism of subsidy policies on supply chain decisions across different subsidy instruments. First, marginal costs show a persistent negative correlation with both emission reduction investments and profitability. Neither R&D cost subsidies nor per-unit product emission reduction subsidies can fully offset the constraining effect of rising marginal costs on supply chain performance. Second, the emission reduction demand coefficient maintains a stable positive effect: under both subsidy schemes, increased demand sensitivity to emission reduction enhances green premium effects, thereby boosting emission reduction efforts and profitability across the supply chain. Third, a clear substitution relationship exists between subsidy intensity and cost-sharing ratios—increased public subsidies lead to corresponding reductions in manufacturers’ cost-sharing commitments, demonstrating how government subsidies systematically reshape responsibility allocation in supply chain collaboration.

6. Comparison of Emission Reduction Decisions and Effects

This section presents a systematic comparison of optimal outcomes under varying government subsidy mechanisms. The analysis proceeds in two key dimensions: First, we examine how different cooperative modes within the supply chain influence optimal decision-making under each specific subsidy scheme. Second, we look at the impact of different government subsidies on the social welfare of the supply chain.

6.1. Comparison of Government Subsidy Allocation Strategy Under Different Supply Chain Member Reduction Decisions

6.1.1. Comparison of Decisions Made by Supply Chain Members Under Different Cooperation Modes with R&D Cost Subsidies

Considering the government subsidy based on development costs, the profitability and unit reduction quantity comparison of suppliers and manufacturers under different cooperative models, for ease of comparison, the development cost coefficients are set as μ s = μ m = μ , and the demand sensitivity coefficient as α = β [25,51]. Under the framework of R&D cost subsidies, we first analyze the variations in carbon emission reductions across different cooperation modes. The comparative results are formally summarized in Proposition 1:
Proposition 1. 
When the government provides subsidies based on R&D costs, the comparison of suppliers’ carbon emission reductions under different cooperation modes of supply chain members is as follows: x S 22 * > x S 21 * > x S 23 * .
The proof is presented in Appendix B.
As revealed in Proposition 1, under the government’s R&D cost subsidy policy directed at the manufacturer, the supplier’s carbon emission reduction achieves its highest level in the full cooperation scenario (denoted as mode 22). The non-cooperative scenario (mode 21) yields a moderate outcome, while the cost-sharing scenario (mode 23) results in the least effective emission reduction performance. This hierarchy underscores the critical role of cooperation intensity in amplifying the effectiveness of government subsidies on supply chain emission reduction outcomes.
Under the same R&D cost subsidy mechanism, we further examine the manufacturer’s emission reduction performance across different cooperation modes. The comparative outcomes are formally established in Proposition 2:
Proposition 2. 
When the government provides subsidies based on R&D costs, the comparison of manufacturers’ carbon emission reductions under different cooperation modes of supply chain members is as follows: x m 22 * > x m 21 * > x m 23 * .
The proof is presented in Appendix B.
Proposition 2 reveals that under government R&D subsidies for manufacturers, the refinery manufacturer’s emission reduction level reaches its optimum in the supplier–manufacturer cooperation scenario, followed by the non-cooperative scenario, with the cost-sharing scenario yielding the least favorable outcome. This pattern aligns with the supplier’s emission reduction behavior observed in Proposition 1, demonstrating that full cooperation consistently achieves the highest environmental performance for both supply chain members. The consistency further confirms that cooperative intensity serves as a key determinant in enhancing the effectiveness of emission reduction efforts under government subsidy policies.
Building upon the comparative analysis of emission reduction performance under different cooperation modes, we now extend our investigation to evaluate the economic implications. The following proposition systematically examines the profitability of supply chain members across various cooperation frameworks:
Proposition 3. 
When the government provides subsidies based on R&D costs, the profit order of the supply chain members under different cooperation modes is:  π t 22 * > π t 23 * > π t 21 * .
The proof is presented in Appendix B.
This hierarchical structure demonstrates that while full cooperation generates the highest economic returns, the cost-sharing mechanism still yields better financial outcomes than the non-cooperative scenario, though it remains suboptimal compared to complete collaboration.
Based on Propositions 1 through 3, government-provided R&D cost subsidies serve as an effective policy instrument to incentivize supply chain cooperation, simultaneously enhancing emission reduction performance and economic returns. In the full cooperation mode, suppliers and manufacturers achieve optimal outcomes in both emission reductions and total profits through integrated technology sharing and resource coordination. The non-cooperative mode ranks second in performance, while the cost-sharing model yields the least favorable results in both environmental and economic dimensions. The superior performance of the full cooperation mode can be attributed to its ability to foster deeper collaboration, optimize resource allocation, and improve operational efficiency—particularly when facing high carbon abatement costs. In contrast, although the cost-sharing mechanism partially distributes emission reduction expenses, it fails to establish comprehensive collaboration frameworks and effective incentive structures, ultimately limiting its effectiveness in driving meaningful environmental or financial improvements.
These findings suggest that policymakers should prioritize support for cooperative models within supply chains, as they facilitate closer integration among members and enable more efficient emission reduction achievements while maximizing economic benefits. Such an approach not only aligns environmental objectives with economic incentives but also creates sustainable conditions for long-term green transformation in industrial sectors.

6.1.2. Comparison of Decision-Making Among Supply Chain Members Under Different Cooperation Modes for Per-Unit Production Emission Reduction Subsidies

Building upon the analysis of R&D cost subsidies, we now turn to examining the effects of an alternative subsidy mechanism—government support based on per-unit production emission reductions. This form of incentive ties financial support directly to emission reduction outcomes. Under this subsidy framework, we analyze how different cooperation modes influence the optimal decisions of suppliers and manufacturers regarding profits and emission reductions. The following propositions formalize these relationships:
Proposition 4. 
When the government provides subsidies based on per-unit product emission reductions, the comparison of suppliers’ carbon emission reductions under different cooperation modes of supply chain members is as follows: x s 32 * > x s 33 * > x s 31 * .
The proof is presented in Appendix B.
Proposition 4 demonstrates that under the government’s per-unit emission reduction subsidy policy targeting the refinery manufacturer, the supplier, as the supply chain leader, exhibits distinct emission reduction behaviors across different decision-making structures. The highest emission reduction is achieved under the centralized decision-making model (full cooperation), attributable to the optimal resource allocation driven by the system-wide objective. The cost-sharing model yields intermediate results, reflecting the partial incentive effect of risk-sharing on emission reduction investments. In contrast, the decentralized model (non-cooperation) leads to the poorest performance due to insufficient motivation for emission reduction under individual profit-maximization objectives.
Proposition 5. 
When the government provides subsidies based on per-unit product emission reductions, the comparison of manufacturers’ carbon emission reductions under different cooperation modes of supply chain members is as follows: x m 33 * > x m 32 * > x m 31 * .
The proof is presented in Appendix B.
Proposition 5 reveals that under the government’s per-unit emission reduction subsidy policy for refinery manufacturers, the manufacturer achieves its optimal emission reduction performance in the centralized decision-making model (full cooperation), followed by the cost-sharing model, with the non-cooperative scenario yielding the least favorable results. This pattern can be attributed to the efficient resource allocation and coordinated environmental objectives under the fully collaborative framework, which enables maximal utilization of subsidy incentives. The cost-sharing model provides moderate improvements through risk distribution, while the absence of coordination mechanisms in the non-cooperative scenario leads to suboptimal environmental outcomes despite the presence of government subsidies.
Proposition 6. 
When the government provides subsidies based on per-unit product emission reductions, the profit order of the supply chain members under different cooperation modes is:  π t 32 * > π t 33 ∗ > π t 31 * .
The proof is presented in Appendix B.
Proposition 6 demonstrates that under the government’s per-unit emission reduction subsidy policy, different decision-making models exert significant impacts on supply chain profitability. The study reveals that the centralized decision-making model achieves the highest profit, followed by the cost-sharing model, while the decentralized model yields the lowest financial returns.

6.2. Comparison of Supply Chain Members’ Emission Reduction Decisions Under Different Government Subsidy Methods

6.2.1. Comparative Analysis of Emission Reduction and Profitability Under Different Subsidy Methods in Non-Cooperative Supply Chains

We now examine how different government subsidy mechanisms influence supply chain decisions when members choose non-cooperative, cooperative, or cost-sharing strategies. This comparative analysis holds particular value for policymakers seeking to optimize subsidy design in real-world contexts. The following propositions systematically compare the effects of two primary subsidy approaches—R&D cost subsidies and per-unit product emission reduction subsidies—on environmental and economic outcomes:
Proposition 7. 
Under the non-cooperative mode, the impacts of different government subsidy schemes on supply chain emission reduction effects and profits are compared as follows: x m 31 * > x m 21 * , x s 31 * < x s 21 * , π m 31 * < π m 21 * , π s 31 * < π s 21 * .
Proposition 7 reveals significant differences in emission reduction outcomes and profitability under different subsidy types within the non-cooperative supply chain. In terms of emission reduction, the per-unit emission reduction subsidy leads to higher manufacturer emission reductions compared to the R&D subsidy ( x m 31 * > x m 21 * ), yet results in lower supplier emission reductions ( x s 31 * < x s 21 * ). This indicates that while the per-unit emission reduction subsidy is more effective in stimulating manufacturer emission reductions, it generates weaker incentives for suppliers. Regarding profitability, the per-unit emission reduction subsidy yields lower profits for both manufacturers and suppliers compared to the R&D subsidy ( π m 31 * < π m 21 * and π s 31 * < π s 21 * ). This suggests that although the per-unit emission reduction subsidy demonstrates advantages in promoting emission reductions, it provides weaker profit incentives for supply chain members. In summary, while the per-unit emission reduction subsidy can enhance emission reduction performance—particularly for manufacturers—it offers limited incentives for suppliers and results in lower overall profitability relative to the R&D subsidy in non-cooperative settings.

6.2.2. Comparative Analysis of Emission Reduction and Profitability Under Different Subsidy Methods in Fully Cooperative Supply Chains

We now explore the influence of two government subsidy approaches on supply chain decisions under full cooperation. The key findings are as follows:
Proposition 8. 
Under the full cooperation mode of supply chain members, the comparison of emission reduction effects and profits under different government subsidy methods satisfies the following relationships: x m 32 * > x m 22 * , x s 32 * < x s 22 * , π t 32 * < π t 22 * .
According to Proposition 8, in the full cooperation mode, the per-unit emission reduction subsidy demonstrates better incentives for manufacturer emission reduction compared to the R&D subsidy, while the supplier’s emission reduction level under the per-unit subsidy is lower than under the R&D cost subsidy. This indicates that although the per-unit subsidy more effectively promotes manufacturer emission reduction, it provides relatively weaker incentives for suppliers. In contrast, the R&D cost subsidy shows more significant effects on supplier emission reduction. Regarding profitability, the total supply chain profit under the per-unit emission reduction subsidy is lower than under the R&D cost subsidy model. This finding suggests that while the per-unit subsidy has certain advantages in promoting emission reduction, it is less effective than the R&D cost subsidy in enhancing overall supply chain profitability. Therefore, policymakers need to balance emission reduction effectiveness and economic benefits when selecting the most appropriate subsidy approach based on policy objectives.

6.2.3. Comparative Analysis of Emission Reduction and Profitability Under Different Subsidy Methods in Cost-Sharing Supply Chains

Finally, we examine the impact of different government subsidy mechanisms under the cost-sharing model. This study helps to understand how government subsidies coordinate the environmental and economic goals of supply chain members within a collaborative framework of risk sharing. Therefore, Proposition 9 can be obtained:
Proposition 9. 
Under the cost-sharing mode, the impacts of different government subsidy approaches on supply chain emission reduction and profits are as follows: x m 33 * > x m 23 * , x s 33 * < x s 23 * , π m 33 * < π m 23 * , π s 33 * < π s 23 * .
The findings reveal that in the cost-sharing context, per-unit product emission reduction subsidies prove more effective in promoting manufacturer emission reductions, while showing weaker incentive effects for suppliers. In contrast, R&D cost subsidies demonstrate stronger effectiveness in driving supplier emission reductions. Regarding profitability, both manufacturer profits ( π m 33 * ) and supplier profits ( π s 33 * ) under per-unit emission reduction subsidies remain lower than those achieved under R&D cost subsidies. These results suggest that although per-unit product emission reduction subsidies show advantages in promoting emission reductions, they appear less effective in enhancing overall supply chain profitability within cost-sharing frameworks. This imbalance in profit distribution between manufacturers and suppliers may potentially affect the stability of supply chain cooperation, suggesting policymakers should consider both environmental effectiveness and economic equity when designing subsidy schemes.

6.3. Social Welfare Analysis

6.3.1. Social Welfare Model Construction

The optimal social welfare under government R&D cost subsidies and per-unit product emission reduction subsidies are [37]:
S W ( ε ) = π + ( a − b p + α x m + β x s ) 2 2 b − 1 2 ε μ m x m 2 ,
S W ( σ ) = π + ( a − b p + α x m + β x s ) 2 2 b − σ x m ( a − b p + α x m + β x s ) .
According to the optimal decision variables mentioned above, social welfare can be found as:
S W ( ε ) 21 = μ s μ m ( a − b c ) 2 [ ( ε − 1 ) 2 ( 7 b μ s − β 2 ) μ m + α 2 μ s ( ε − 3 ) ] 2 [ ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 2 μ s ] 2 ,
,
S W ( ε ) 22 = μ s μ m ( a − b c ) 2 [ ( ε − 1 ) 2 ( 3 b μ s − β 2 ) μ m − α 2 μ s ( ε − 3 ) ] 2 [ ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + α 2 μ s ] 2
S W ( ε ) 23 = μ m ( a − b c ) 2 [ ( ε − 1 ) 4 b ( 448 b 2 μ s 2 − 176 b μ s β 2 + β 4 ) μ m 3 + ( ε − 1 ) 2 [ 32 μ s 2 ( 5 ε − 9 ) b 2 − 4 μ s ( 15 ε − 23 ) β 2 b − β 4 ( ε + 1 ) ] α 2 μ m − ( ε − 1 ) μ s [ 4 μ s ( ε + 15 ) b + ( 5 ε − 13 ) β 2 ] α 4 μ m − 4 α 6 μ s w ( ε + 1 ) } 2 [ ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + α 2 μ s ] 2 , S W ( σ ) 31 = μ s μ m ( a − b c ) 2 [ 5 b 2 σ 2 + ( 8 α σ − 7 μ m ) b + 3 α 2 ) μ s − ( 4 b σ 2 + 4 α σ − μ m ) β 2 ] − 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b μ s + α 2 ) + μ m β 2 ] 2 ,
S W ( σ ) 32 = μ s μ m ( a − b c ) 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) ( 2 b 2 σ 2 + 3 ( α σ − μ m ) b + α 2 ) μ s 2 + μ m ( 5 b 2 σ 2 + ( 8 α σ − 7 μ m ) b + 3 α 2 ) μ s β 2 + β 4 μ m 2 ] − 2 [ ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 ] 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 ] , S W ( σ ) 33 = μ m ( a − b c ) 2 [ 80 μ s 2 σ 6 b 6 + 16 μ s 2 ( 28 α σ − 27 μ m ) σ 4 b 5 + [ 16 ( − 100 α μ m σ 3 + 65 α 2 σ 4 + 48 σ 2 μ m 2 ) μ s 2 + 56 β 2 σ 4 μ s μ m ] b 4 + [ 16 ( 80 α 3 σ 3 − 138 α 2 μ m σ 2 + 88 α σ μ m 2 − 28 μ m 3 ) μ s + 8 μ m β 2 σ 2 ( 52 α σ − 25 μ m ) μ s b 3 + [ 16 ( − 84 μ m σ α 3 + 55 α 4 σ 2 + 40 α 2 μ m 2 ) μ s 2 ) + 16 μ m β 2 ( 18 σ 2 α 2 − 23 α σ μ m + 11 μ m 2 ) μ s + 5 β 4 μ m 2 σ 2 ] b 2 + [ 16 ( 20 α 5 σ − 19 α 4 μ m ) μ s 2 + 8 α 2 β 2 μ m ( 22 α σ − 21 μ m ) μ s + ( 8 α σ − 7 μ m ) μ m 2 β 4 ] b + 48 α 6 μ s 2 + 3 α 2 β 4 μ m 2 + 40 α 4 β 2 μ m μ s − 2 [ ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 ] 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 ] .
In order to further solve the optimal subsidy coefficient of the government, the second-order partial derivation about sum can be obtained:
∂ 2 ( S W 21 ) ∂ ε 2 < 0 , ∂ 2 ( S W 22 ) ∂ ε 2 < 0 , ∂ 2 ( S W 23 ) ∂ ε 2 < 0 , ∂ 2 ( S W 31 ) ∂ σ 2 < 0 , ∂ 2 ( S W 32 ) ∂ σ 2 < 0 , ∂ 2 ( S W 33 ) ∂ σ 2 < 0 .
That is, there is only one optimal subsidy coefficient to maximize social welfare. Because the above-mentioned quadratic partial derivative formula is too complex, it is not expressed in this paper, but it does not affect the analysis of the results.

6.3.2. Comparative Analysis of Social Welfare

This section compares the differences in optimal social welfare between the government’s R&D cost subsidy and the per-unit emission reduction subsidy, examining the magnitude of optimal decisions under each subsidy scheme.
Proposition 10. 
The comparison of social welfare under the two subsidy methods in the non-cooperative mode between suppliers and manufacturers satisfies the following relationship: S W 21 * > S W 31 * .
The proof is presented in Appendix B.
Proposition 10 indicates that under the non-cooperative supply chain mode, social welfare under R&D cost subsidies exceeds that achieved under per-unit emission reduction subsidies. R&D cost subsidies (ε) enhance social welfare through two complementary channels: first, by promoting technological upgrades at the supplier level that reduce emission abatement costs across the industry; second, by creating learning effects for manufacturers that accelerate technology adoption and transformation cycles.
We further examine how different subsidy mechanisms affect social welfare in fully cooperative scenarios, as shown in Proposition 11:
Proposition 11. 
The comparison of social welfare under the two subsidy methods in the fully cooperative mode between suppliers and manufacturers satisfies the following relationship: S W 22 * > S W 32 * .
The proof is presented in Appendix B.
Proposition 11 demonstrates that even under full cooperation, R&D cost subsidies continue to generate higher social welfare compared to per-unit production emission reduction subsidies. This persistent advantage stems from the fundamental characteristics of R&D cost subsidies: they not only stimulate immediate emission reductions but also foster long-term technological innovation and knowledge accumulation within the supply chain. The cooperative environment further amplifies these benefits by enabling better coordination in research activities and more efficient dissemination of technological advancements between suppliers and manufacturers.
Proposition 12 demonstrates the impact of different government subsidies on supplier members under the cost-sharing model, as shown below:
Proposition 12. 
Under the cost-sharing model between suppliers and manufacturers within the supply chain, the comparison of social welfare under the two subsidy methods is as follows: S W 23 * > S W 33 * .
The proof is presented in Appendix B.
Proposition 12 demonstrates that the welfare superiority of R&D cost subsidies persists even in the cost-sharing context. This finding is particularly significant, as it reveals that while cost-sharing facilitates risk distribution between supply chain members, it cannot overcome the fundamental limitations of output-based subsidy designs.
The consistent results from Propositions 10–12 demonstrate that R&D cost subsidies consistently achieve superior Pareto improvements compared to per-unit production emission reduction subsidies across all three modes. This advantage originates from fundamentally different incentive mechanisms: R&D cost subsidies directly reduce the marginal cost of innovation, thereby stimulating long-term technological progress and generating positive externalities, including knowledge spillovers and sustained efficiency gains [70]. In contrast, unit-based subsidies, while raising the marginal return on emission reduction, often lead to superficial compliance, where firms prioritize achieving short-term quantitative targets over investing in substantive innovation. Consequently, although both subsidy types contribute to emission reduction, R&D subsidies more effectively enhance overall social welfare by promoting genuine technological advancement and establishing a foundation for sustainable transition.

7. Numerical Analysis

This section aims to provide a numerical illustration of the theoretical models developed in the previous sections. The simulations are designed to visually compare the performance of the three cooperation modes under different government subsidies and to derive managerial insights. It is crucial to note that the analysis employs a theoretical numerical example rather than empirical data. The parameter values (e.g., a = 100 , b = 2 , c = 10 , α = 0.8 , β = 0.6 , μ m = 1.2 , μ s = 1.0 , ε ∈ [ 0 , 0.8 ] , σ ∈ [ 0 , 0.6 ] ) are set primarily for illustrative purposes, with their magnitudes and relationships calibrated to be theoretically consistent (i.e., satisfying all model assumptions for interior solutions) and practically plausible, drawing from established practices in relevant literature [37,47,72,73]. The simulation was executed according to the following methodological procedure: First, the analytical equilibrium solutions derived in the preceding sections were computationally implemented using MATLAB R2023a. The core algorithm involved systematically varying the key subsidy parameters—the R&D cost subsidy coefficient ( ε ) and the per-unit emission reduction subsidy ( σ )—across their defined ranges ( ε ∈ [ 0 , 0.8 ] , σ ∈ [ 0 , 0.6 ] ). For each discrete value within these ranges, the corresponding equilibrium outcomes for profits and emission reductions were calculated by substituting the full set of parameter values into the analytical model. This process generated the dataset underpinning the comparative figures (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8). To ensure the robustness of the observed trends, a sensitivity analysis was conducted on critical parameters, confirming that the ordinal rankings and central insights remain stable under reasonable parameter variations.
Figure 1. The impact of the government R&D cost subsidy coefficient on supply chain profits.
Figure 2. The impact of the government R&D cost subsidy coefficient on supplier emission reduction.
Figure 3. The impact of the government R&D cost subsidy coefficient on manufacturers’ emission reductions.
Figure 4. The impact of the government per-unit production emission reduction subsidy coefficient on supply chain profits.
Figure 5. The impact of the subsidy coefficient of government per-unit production emission reduction on supplier emission reduction.
Figure 6. The impact of the subsidy coefficient of government per-unit production emission reduction on the emission reduction of manufacturers.
Figure 7. The impact of the R&D cost subsidy coefficient on social welfare.
Figure 8. The impact of the per-unit production emission reduction subsidy coefficient on social welfare.

7.1. The Impact of Subsidy Coefficients on the Profits and Emission Reductions of Supply Chain Members

This section investigates the impact of the R&D cost subsidy coefficient and the per-unit emission reduction subsidy coefficient on the profits and emission reductions of supply chain members. The main findings are presented in Figure 1, Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6.
As shown in Figure 1, as the government’s R&D cost subsidy coefficient increases, the profits of supply chain members also rise accordingly. Under the government’s R&D cost subsidy, the full cooperation mode yields the highest profits for supply chain members, followed by the non-cooperative mode, while the cost-sharing mode results in the lowest profits. When the government R&D cost subsidy coefficient is small, the profit growth for supply chain members is relatively slow. However, as the subsidy coefficient continues to increase, the profits of supply chain members grow at a much faster pace. This indicates that higher levels of R&D subsidies provide a strong incentive for supply chain members to enhance their collaboration and improve their profitability. These findings highlight the important role that government subsidies play in driving both profitability and collaboration within the supply chain. Full cooperation consistently results in the highest profits, while the non-cooperative model provides lower profits. The significant increase in profits as the subsidy coefficient rises further underscores the importance of government support in stimulating economic performance and fostering a more cooperative and innovative supply chain environment.
As shown in Figure 2, suppliers’ emission reductions increase with the government R&D cost subsidy coefficient. Under this subsidy, the full cooperation mode achieves the highest emission reductions, followed by the non-cooperative mode, while the cost-sharing mode yields the lowest profits. When the subsidy coefficient is relatively small, the growth rate of emission reductions is slower across all three modes. Once the coefficient exceeds a threshold ( ε > 0.5 ), however, the non-cooperative mode exhibits a faster growth rate in emission reductions than full cooperation. This indicates that higher subsidies may accelerate emission reductions in non-cooperative settings, although full cooperation still offers the greatest overall environmental and financial benefits. The findings emphasize that while full cooperation provides superior long-term profits and emission reductions, non-cooperative models can achieve faster emission reductions under certain conditions, especially with substantial R&D cost subsidies. This underscores the need to optimize both subsidy mechanisms and cooperation strategies to enhance environmental sustainability and economic efficiency in supply chains.
Figure 3 shows that as the government’s R&D cost subsidy coefficient increases, manufacturers’ emission reductions also increase. Under the government’s subsidy based on research and development costs, the profit is highest in the fully cooperative mode of supply chain members, followed by the non-cooperative mode, and the cost sharing is minimized. These findings highlight that government R&D cost subsidies not only influence profits but also play a critical role in determining the effectiveness of emission reductions within the supply chain. While full cooperation continues to provide the highest emission reductions, non-cooperative modes become more effective at reducing emissions as the subsidy coefficient increases, suggesting that larger subsidies can drive faster emission reductions, even in less collaborative environments.
Figure 4, Figure 5 and Figure 6 depict the impact of the government’s per-unit emission reduction subsidy on supply chain profits and member emission levels. As shown in Figure 4, supply chain profits increase with a higher subsidy coefficient. The fully cooperative mode yields the highest profits under this policy, while the non-cooperative and cost-sharing modes produce comparable, lower returns. Profit growth is modest at lower subsidy levels across all modes but accelerates markedly as the subsidy intensifies, particularly under full cooperation. This underscores the critical role of government subsidies in enhancing supply chain profitability, with higher subsidies acting as a powerful incentive for profit growth, especially within cooperative frameworks.
As shown in Figure 5, suppliers’ emission reductions increase with a higher government per-unit emission reduction subsidy coefficient. Among the three cooperation modes, the fully cooperative model achieves the highest level of supplier emission reduction, followed by the cost-sharing model, while the non-cooperative model yields the lowest. This demonstrates that the subsidy’s effectiveness in incentivizing upstream suppliers is significantly enhanced by closer supply chain collaboration.
A comparison of the three cooperation modes in Figure 6 reveals that the fully cooperative model leads to the highest manufacturer emission reductions, followed by the cost-sharing and non-cooperative models. In all cases, emission levels exhibit a positive correlation with the government’s per-unit emission reduction subsidy coefficient, underscoring the dual importance of effective collaboration and adequate subsidy intensity.

7.2. The Impact of Subsidy Coefficients on Social Welfare Under Different Subsidy Methods

This section compares the impact of different government subsidy methods on social welfare.
As shown in Figure 7, social welfare initially increases steadily or slightly declines as the government R&D cost subsidy coefficient rises. Beyond a certain threshold, however, social welfare under the full cooperation mode begins to decrease rapidly. A similar pattern is observed in the non-cooperative and cost-sharing modes, where social welfare grows slowly at first but eventually drops sharply. These results indicate that although government R&D cost subsidies can enhance social welfare up to a point, excessive or indiscriminate subsidies may eventually become counterproductive. Therefore, it is essential for the government to carefully design and balance such subsidy policies to avoid unintended negative welfare effects.
In Figure 8, social welfare consistently improves across all three cooperation models—non-cooperative, full cooperation, and cost-sharing—as the per-unit emission reduction subsidy coefficient increases. Specifically, under both the full cooperation and non-cooperative modes, social welfare rises steadily with higher subsidy levels. In contrast, under the cost-sharing mode, social welfare increases initially but declines sharply after reaching a peak. This suggests that when implementing per-unit product emission reduction subsidies, supply chain participants should avoid the cost-sharing mode to prevent a significant welfare loss. These findings reinforce the conclusion that the most advantageous strategies, in terms of both profitability and social welfare, are achieved under the full cooperation and non-cooperative models.

8. Conclusions

This paper constructed a two-level supply chain model consisting of a single raw material supplier (leader) and a single refining manufacturer (follower) to systematically study the impact of government subsidies targeting manufacturers—R&D cost subsidies (RS) and per-unit emission reduction subsidies (PS)—on supply chain equilibrium decisions (including wholesale prices, retail prices, and emission reduction efforts) and social welfare under three modes: non-cooperation, cost-sharing, and full cooperation. The core findings are as follows:
(1)
Positive incentive effects of subsidy coefficients. Both R&D cost subsidies and per-unit emission reduction subsidies positively influence wholesale prices, retail prices, emission reductions, and supply chain profits. R&D cost subsidies incentivize manufacturers to increase emission reduction efforts by lowering their marginal emission reduction costs, whereas per-unit emission reduction subsidies stimulate greater investment in emission reduction by directly enhancing manufacturers’ returns on such efforts.
(2)
Comparative analysis of equilibrium decisions under two subsidy strategies. In terms of emission reduction outcomes, manufacturers’ equilibrium emission reductions under the per-unit subsidy strategy are significantly higher than under the R&D cost subsidy strategy, whereas suppliers’ equilibrium emission reductions demonstrate a greater advantage under the R&D cost subsidy strategy. Regarding profit distribution, manufacturer profits, supplier profits, and overall supply chain profits under the R&D cost subsidy strategy are significantly superior to those under the per-unit subsidy strategy. In terms of pricing mechanisms, the equilibrium wholesale price under the R&D cost subsidy strategy is generally higher, reflecting the potential premium associated with technological R&D, while the equilibrium retail price is comparatively lower, indicating that R&D subsidies enhance market competitiveness by reducing long-term emission reduction costs.
(3)
Differential impact mechanisms and applicability of subsidy strategies. (1) Emission reduction incentive dimension: R&D cost subsidies focus on stimulating long-term emission reduction potential and innovation momentum by reducing the fixed costs of technological R&D, though their direct stimulation of short-term emission reduction behavior is limited. Per-unit emission reduction subsidies, on the other hand, provide stronger immediate incentives for manufacturers’ short-term emission reduction actions but are less effective in promoting continuous technological innovation and long-term emission reduction efficiency. Regardless of the subsidy strategy, the full cooperation mode consistently achieves the optimal emission reduction outcome, highlighting the importance of in-depth collaboration. (2) Supply chain profit dimension: R&D subsidies are more conducive to improving the overall profit level of the supply chain, providing stronger economic sustainability for the green transition.
(4)
Social welfare comparison and subsidy intensity threshold effects. Under the same cooperation mode, the social welfare generated by R&D cost subsidies is significantly higher than that of per-unit emission reduction subsidies. As the intensity of R&D cost subsidies increases, social welfare shows an initial increase followed by stabilization. However, under the cost-sharing mode, when the intensity of per-unit emission reduction subsidies exceeds a specific threshold, social welfare experiences a precipitous decline.
In summary, per-unit emission reduction subsidies help manufacturers achieve short-term emission reduction targets but have limited effects on improving supply chain profits and supplier emission reductions. R&D cost subsidies, in contrast, better enhance long-term supply chain profits and emission reduction investments, promoting deeper collaboration within the supply chain. Governments should design differentiated subsidy policies based on the roles, emission reduction cost structures, and technological capabilities of supply chain members. Particularly when implementing high-intensity per-unit emission reduction subsidies, it is essential to guard against the opportunistic behaviors and coordination costs that may arise in cost-sharing models, prioritize the guidance of full cooperation modes, and encourage collaborative innovation among supply chain members to maximize policy effectiveness.
This study advances the theoretical understanding of subsidy mechanisms in supply chain dynamics by systematically comparing R&D cost subsidies and per-unit emission reduction subsidies across different cooperation modes. It elucidates how these subsidies differentially influence manufacturer and supplier behaviors, offering novel insights into the interplay between economic incentives and environmental objectives in asymmetric power structures. From a managerial perspective, the findings provide actionable guidance for enterprises to align pricing strategies and cooperation models with government incentives, while offering policymakers a framework to design differentiated subsidy mechanisms that balance short-term emission reduction targets with long-term green transformation. Future research should extend this work by examining multi-tier supply chains, incorporating external factors like market volatility and regulatory changes, and validating theoretical findings through empirical data and dynamic modeling. In conclusion, this study demonstrates that while per-unit subsidies effectively drive immediate emission reductions, R&D subsidies are superior for fostering long-term innovation and supply chain profitability. Governments must tailor policies to specific supply chain characteristics, and subsequent research should continue refining these models to address the evolving landscape of environmental governance and supply chain complexity.

Author Contributions

Conceptualization, H.X. and D.T.; methodology, H.X.; software, H.X.; validation, H.X. and X.C.; formal analysis, X.C.; resources, X.C.; writing—original draft preparation, X.C.; writing—review and editing, H.X.; visualization, H.X.; supervision, D.T.; project administration, H.X.; funding acquisition, H.X. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Humanities and Social Sciences Foundation for Youth Scholars of the Ministry of Education of China (Grant No. 22YJC630171), the Natural Science Foundation of Sichuan Province (Grant No. 2023NSFSC1055), the Cultivation Project of Humanities and Social Sciences General Fund of Southwest Petroleum University (Grant No. 2022-2023RW020), and the Students Innovation Training Program Project of Sichuan (Grant No. 202410615064).

Institutional Review Board Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Acknowledgments

The authors express their gratitude to the anonymous reviewers and editors for their helpful comments about this paper.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Appendix A.1

Proof of Lemma 1. 
For Equation (3), the second-order partial derivatives and mixed partial derivatives of the retail price p and the emission reduction effort x m are calculated, and the Hessian matrix is given as:
H m 21 = ∂ 2 π m ∂ p 2 ∂ 2 π m ∂ p ∂ x m ∂ 2 π m ∂ x m ∂ p ∂ 2 π m ∂ x m 2 = − 2 b α 0 − ( 1 − ε ) μ m .
When the first principal minor of the determinant satisfies − 2 b < 0 and the second principal minor satisfies 2 b ( 1 − ε ) μ m − α 2 > 0 , the matrix is negative definite. Therefore, the profit function π m 21 is strictly concave with respect to p and xm. □
For Equation (2), taking the first-order derivatives of the retail price p and the emission reduction effort xm, setting them equal to 0, and solving the system of equations yields:
p 21 * = a + b ω + α x m + β x s 2 b ;   x m = − α ( p − ω ) ( 1 − ε ) μ m .
Substituting Equation (A1) into (1), we obtain:
π s 21 = ( ω − c ) b ( 1 − ε ) μ m A + β ( α x s − b ω ) B − 1 2 μ s x s 2 ,
where:
A = 2 b ( 1 − ε ) μ m − α 2 ;   B = 2 μ s A − ( 1 − ε ) μ m β 2 .
For Equation (A2), the second-order partial derivatives and mixed partial derivatives of the wholesale price ω and the emission reduction effort x s are calculated, and the Hessian matrix is given as:
H s 21 = ∂ 2 π s ∂ ω 2 ∂ 2 π s ∂ ω ∂ x s ∂ 2 π s ∂ x s ∂ ω ∂ 2 π s ∂ x s 2 = − 2 b b ( 1 − ε ) μ m A β 1 − ( 1 − ε ) μ m β 2 B β 1 − ( 1 − ε ) μ m β 2 B − μ s .
When the first principal minor satisfies − 2 b < 0 and the second principal minor satisfies 2 b ( 1 − ε ) μ m − α 2 > 0 , the matrix is negative definite. Therefore, the profit function π s 21 is strictly concave with respect to ω and x s .
For Equation (A2), we first take the first-order derivatives of the wholesale price ω and the emission reduction effort x s , set them equal to zero, and solve the system of equations. This yields the following optimal wholesale price and emission reduction effort:
ω 21 * = ( ε − 1 ) μ m [ 2 ( a + b c ) μ s − β 2 c ] + ( a + b c ) α 2 μ s ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s ,   x s 21 * = ( ε − 1 ) μ m β ( a − b c ) ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s .
Equation (A4) provides the optimal emission reduction level of the supplier x s 21 * and the corresponding optimal wholesale price ω 21 * . Substituting Equation (A4) into Equation (A1), we obtain the optimal emission reduction of the manufacturer and the optimal retail price p 21 * :
p 21 * = ( ε − 1 ) μ m [ ( 3 a + b c ) μ s − β 2 c ] + ( a + b c ) α 2 μ s ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) − α 2 μ s ,   x m 21 * = − μ s α ( a − b c ) ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s .
By substituting Equations (A4) and (A5) into Equations (2) and (3), we derive the supplier’s expected optimal profit π s 21 * and the manufacturer’s expected optimal profit π m 21 * :
x m 21 * = − μ s α ( a − b c ) ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) + 2 α 2 μ s ,
π s 21 * = ( ε − 1 ) μ m μ s ( a − b c ) 2 [ ( ε − 1 ) ( 4 b μ m μ s − β 2 μ m ) − α 2 μ s ] 2 .
Proof of Lemma 2. 
For Equation (10), the second-order partial derivatives and mixed partial derivatives of retail price p , emission reduction effort x m , and emission reduction effort x s are obtained, yielding the following Hessian matrix:
H t 22 = ∂ 2 π t ∂ p 2 ∂ 2 π t ∂ p ∂ x m ∂ 2 π t ∂ p ∂ x s ∂ 2 π t ∂ x m ∂ p ∂ 2 π t ∂ x m 2 0 ∂ 2 π t ∂ p ∂ x s 0 ∂ 2 π t ∂ x s 2 = − 2 b α β α − ( 1 − ε ) μ m 0 β 0 − μ s .
Since the first principal minor satisfies − 2 b < 0 , the second principal minor satisfies 2 b ( 1 − ε ) μ m − α 2 > 0 , and the third principal minor satisfies − 2 b [ ( 1 − ε ) μ m μ s ] + α 2 μ s + β 2 ( 1 − ε ) μ m < 0 , the matrix is negative definite. Therefore, the overall supply chain profit function π t 22 is strictly concave with respect to p , x m , and x s . □
For Equation (10), taking the first-order derivatives of the retail price and the emission reduction efforts and setting them equal to zero, we obtain:
p = a + b ω + α x m + β x s 2 b ,   x m = α ( p − ω ) ( 1 − ε ) μ m ,   x s = β ( p − c ) μ s .
By combining these expressions, we can obtain p 22 * , x s 22 * , x m 22 * , π t 22 * .
Proof of Lemma 3. 
For Equation (16), the second-order partial derivatives and mixed partial derivatives of retail price p , emission reduction effort x m , and cost-sharing coefficient k are calculated. The corresponding Hessian matrix is:
H m 23 ∂ 2 π m ∂ p 2 ∂ 2 π m ∂ p ∂ x m ∂ 2 π m ∂ p ∂ k ∂ 2 π m ∂ x m ∂ p ∂ 2 π m ∂ x m 2 ∂ 2 π m ∂ x m ∂ k ∂ 2 π m ∂ k ∂ p ∂ 2 π m ∂ k ∂ x m ∂ 2 π m ∂ k 2 = − 2 b α β α − ( 1 − ε ) μ m 0 β 0 − μ s x s 2 .
Since the first principal minor satisfies − 2 b < 0 , the second principal minor satisfies 2 b ( 1 − ε ) μ m m − α 2 > 0 , and the third principal minor satisfies − 2 b [ ( 1 − ε ) μ m μ s x s 2 ] + α 2 μ s + β 2 ( 1 − ε ) μ m < 0 , the matrix is negative definite. Therefore, the manufacturer’s profit function π m 23 * is strictly concave in p , x m , and k . □
For Equation (15), taking the first-order conditions with respect to retail price p and emission reduction effort x m , setting them equal to zero, and solving the system yields:
p = ω + a − b ω + β x s 2 b , x m = α ( a − b ω + β x s ) 2 b ( 1 − ε ) μ m .
Substituting Equation (A10) into Equation (15), we obtain:
π m 23 = ( ω − c ) b ( 1 − ε ) μ m β ( a + β x s − b ω ) B − 1 2 μ s x s 2 .
For Equation (15), taking the first-order derivative with respect to the cost-sharing coefficient k and setting it equal to zero yields:
k = 1 − β ( a − b ω + β x s ) 2 b μ s x s − α 2 ( 1 − ε ) μ m .
For Equation (A12), the second-order partial derivatives and mixed partial derivatives of wholesale price ω and emission reduction effort x s are calculated, and the Hessian matrix is:
H s 23 ∂ 2 π s ∂ ω 2 ∂ 2 π s ∂ ω ∂ x s ∂ 2 π s ∂ x s ∂ ω ∂ 2 π s ∂ x s 2 = − 2 b β ( ω − c ) β ( ω − c ) − μ s ( 1 − k 23 * ) .
Since the first principal minor satisfies − 2 b < 0 , and the second principal minor satisfies 2 b μ s ( 1 − k 23 * ) − β 2 ( 1 − k 23 * ) 2 > 0 , the matrix is negative definite. Therefore, the supplier’s profit function π s 23 is strictly concave in ω and x s .

Appendix A.2

Proof of Lemma 4. 
For Equation (25), the second-order partial derivatives and mixed partial derivatives of the retail price p and the emission reduction effort x m are obtained. The Hessian matrix is:
H m 31 ∂ 2 π m ∂ p 2 ∂ 2 π m ∂ p ∂ x m ∂ 2 π m ∂ x m ∂ p ∂ 2 π m ∂ x m 2 = − 2 b α − b ω α − b ω 2 α ω − μ m .
The profit function π m 31 satisfies the first principal minor − 2 b < 0 and the second principal minor 2 b μ m − ( α + σ ) 2 − 4 b α ω > 0 . Under these conditions, the Hessian matrix is negative definite, and hence the manufacturer’s profit function π m 31 is strictly concave with respect to p and x m . □
For Equation (25), taking the first-order derivatives of retail price p and emission reduction effort x m , setting them equal to zero, and solving the system yields:
p = a − b ω + ( α − b ω ) x m + β x s 2 b , x m = ( α − b ω ) ( a − b ω + β x s ) 2 b μ m − ( α + σ ) 2 .
Substituting Equation (A16) into Equation (23) gives:
π m 31 = ( ω − c ) a − b ω + ( α − b ω ) x m + β x s 2 b + α ( α − b ω ) ( a − b ω + β x s ) 2 b μ m − ( α + σ ) 2 − 1 2 μ s x s 2
For Equation (A17), taking the second-order partial derivatives and mixed partial derivatives of the wholesale price ω and emission reduction effort x s , the Hessian matrix is:
H s 31 = ∂ 2 π s ∂ x s 2 ∂ 2 π s ∂ x s ∂ ω ∂ 2 π s ∂ ω ∂ x s ∂ 2 π s ∂ ω 2 = − μ s β β − b .
The profit function π s 31 satisfies the first principal minor − μ s < 0 and the second principal minor b μ s − β 2 > 0 . Therefore, μ s must be sufficiently large, i.e., μ s > β 2 / b . Under these conditions, the matrix is negative definite, and thus the supplier’s profit function π s 31 is strictly concave with respect to p and x s .
For Equation (A17), taking the first-order derivatives of the wholesale price ω and the emission reduction effort x s , setting them equal to zero, and solving the system yields:
ω 31 * = ( a + b c ) ( b 2 σ 2 + 2 b α σ + α 2 − 2 b μ m ) μ s + b c μ m β 2 2 b ( b 2 σ 2 + 2 b α σ + α 2 − 2 b μ m ) μ s + b μ m β 2 ,
x s 31 * = − β μ m ( a − b c ) 2 2 ( b 2 σ 2 μ s + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ) + μ m β 2 .
Equations (A19) and (A20) represent the optimal supplier emission reduction level x s 31 * and the optimal wholesale price ω 31 * . Substituting Equations (A18) and (A19) into Equation (A16), the optimal manufacturer emission reduction level π m 31 * and optimal retail price p 31 * are obtained:
p 31 * = ( 2 a σ 2 + a c σ − c μ m ) b 2 + ( 3 α σ μ s + α 2 c − 3 a μ m ) b + a α 2 μ s + b c μ m β 2 2 b ( b 2 σ 2 μ s + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ) + b μ m β 2 ,
π m 31 * = − ( b 2 σ 2 + 2 b α σ + α 2 − 2 b μ m ) μ s 2 μ m ( a − b c ) 2 2 ( 2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 ) 2 .
By substituting Equations (A19)–(A22) into Equations (23) and (24), the supplier’s expected optimal profit π s 31 * and the manufacturer’s expected optimal profit π m 31 * are obtained.
Proof of Lemma 5. 
For Equation (32), the second-order partial derivatives and mixed partial derivatives of the retail price p , the emission reduction effort x m , and x s are obtained. The corresponding Hessian matrix is:
H t 32 = ∂ 2 π t ∂ p 2 ∂ 2 π t ∂ p ∂ x m ∂ 2 π t ∂ p ∂ x s ∂ 2 π t ∂ x m ∂ p ∂ 2 π t ∂ x m 2 0 ∂ 2 π t ∂ p ∂ x s 0 ∂ 2 π t ∂ x s 2 = − 2 b α − b σ β α − b σ 2 α σ + σ 2 − μ m β σ β β σ − μ s .
The profit function π t 32 satisfies the first principal minor − 2 b < 0 , the second principal minor 2 b ( 2 α ω + σ 2 − μ m ) − ( α − b ω ) 2 > 0 , and the third principal minor condition det ( H t 32 ) < 0 . Hence, the matrix is negative definite, and the overall supply chain profit function π t 32 is strictly concave with respect to p , x m , and x s . □
For Equation (32), taking the first-order conditions of retail price and emission reduction efforts, setting them equal to zero, and solving the system yields:
p = a + b ω + ( α − b ω ) x m + β x s 2 b , x m = ( a − b c ) ( α + σ ) + β ( α + σ ) x s 2 b μ m − ( α + σ ) 2 , x s = β ( p − α x m ) μ s .
Combining these expressions, the equilibrium results can be derived as:
p 32 * = [ ( a σ 2 + α c σ − c μ m ) b + a α σ + α 2 c − a μ m ] μ s + c β 2 μ m μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s - 2 b μ s μ m + β 2 μ m ,
x s 32 ∗ = − μ s β ( a − b c ) 2 2 [ μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ] + μ m β 2 ,
x m 32 ∗ = μ s β ( a − b c ) ( α + σ ) 2 [ μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m ] + μ m β 2 .
The corresponding profit function of the supply chain is:
π t 32 * = − ( a − b c ) 2 μ s μ m 2 [ μ s b 2 σ 2 + 2 b α σ μ s + α 2 μ s − 2 b μ s μ m + β 2 μ m ] .
Proof of Lemma 6. 
For Equation (37), the second-order partial derivatives and mixed partial derivatives of the retail price, emission reduction effort, and cost-sharing coefficient are obtained. The corresponding Hessian matrix is:
H s 33 = ∂ 2 π m ∂ p 2 ∂ 2 π m ∂ p ∂ x m ∂ 2 π m ∂ p ∂ k ∂ 2 π m ∂ x m ∂ p ∂ 2 π m ∂ x m 2 ∂ 2 π m ∂ x m ∂ k ∂ 2 π m ∂ k ∂ p ∂ 2 π m ∂ k ∂ x m ∂ 2 π m ∂ k 2 = − 2 b α + b ω 0 α + b ω − μ m − β x s 0 − β x s 0 .
The profit function satisfies the first principal minor − 2 b < 0 , the second principal minor condition 2 b μ m − ( α + b ω ) 2 > 0 , and the third principal minor condition − ( α + b ω ) β 2 x s 2 < 0 . Therefore, the Hessian matrix is negative definite, and the manufacturer’s profit function π m 33 is strictly concave in p , x m , and k . □
For Equation (38), taking the first-order derivatives of retail price p and emission reduction effort x m , setting them equal to zero, and solving the system yields:
p = a + b ω + ( α − b ω ) x m + β x s 2 b , x m = ( a − b ω + β x s ) 2 b μ m − ( α + σ ) 2 .
Substituting Equation (A29) into Equation (38) gives:
π m 33 = ( ω − c ) a − b ω + ( α − b ω ) x m + β x s 2 + ( α + σ ) ( a − b ω + β x s ) 2 b μ m − ( α + σ ) 2 − 1 2 k μ s x s 2 .
For Equation (A31), the second-order partial derivatives and mixed partial derivatives of the wholesale price ω and emission reduction effort x s are calculated. The corresponding Hessian matrix is:
H s 33 = ∂ 2 π s ∂ ω 2 ∂ 2 π s ∂ ω ∂ x s ∂ 2 π s ∂ x s ∂ ω ∂ 2 π s ∂ x s 2 = − b β β − μ s ( 1 − k 33 * ) .
The supplier’s profit function π s 33 satisfies the first principal minor − b < 0 and the second principal minor b μ s ( 1 − k 33 * ) − β 2 > 0 . Thus, the condition k 33 ∗ < ( 1 − β 2 ) / μ s must hold. Under these conditions, the Hessian matrix is negative definite, and the supplier’s profit function π s 33 is strictly concave in p and x s .
For Equation (37), taking the first-order derivatives of the wholesale price ω and the emission reduction effort x s , setting them equal to zero, and solving yields:
ω 33 * = β 2 μ m ( 5 b c + a ) + 4 ( a + b c ) ( b 2 σ 2 μ s + 2 b α σ μ s − 2 b μ m μ s + α 2 μ s ) 2 b ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ,
x s 33 ∗ = − 2 ( a − b c ) β μ m ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) .
Equation (A33) provides the optimal emission reduction level and the optimal wholesale price. Substituting Equations (A33) and (A34) into Equation (38), the optimal manufacturer emission reduction level x m 33 ∗ and the optimal retail price p 33 * are obtained. Then, substituting Equations (40) and (42) into Equations (37) and (38), the expected optimal profits of suppliers and manufacturers are obtained as:
π m 33 * = − ( a − b c ) 2 μ m ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) 4 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) , π s 33 * = − ( a − b c ) 2 μ m ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) 8 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) .

Appendix B

Proof of Proposition 1. 
Based on Lemma 1, Lemma 2, and Lemma 3, the following conclusions can be drawn:
x s 23 * − x s 21 * = ( a − b c ) β 3 ( ε − 1 ) [ 4 μ ( ε + 2 ) ( ε − 1 ) b + β 2 ( ε + 3 ) ] 2 [ 4 b ( ε − 1 ) μ − β 2 ( ε − 2 ) ] [ 16 b 2 ( ε − 1 ) 2 μ 2 − 2 b ( 3 ε − 72 ) ( ε − 1 ) β 2 μ − β 4 ( 2 ε − 3 ) ] ,
x s 22 * − x s 23 * = ( a − b c ) β ( ε − 1 ) [ 2 μ ( ε − 1 ) b + β 2 ] [ 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) ] [ 4 b ( ε − 1 ) μ − β 2 ( ε − 3 ) ] ,
x s 22 * − x s 21 * = ( a − b c ) 2 μ ( ε − 1 ) [ 10 b 2 μ 2 ( ε − 1 ) 2 − 2 ( ε − 1 ) ( 3 ε − 8 ) b μ β 2 + ( 2 ε 2 − 10 ε + 13 ) β 4 ] [ 4 b ( ε − 1 ) μ − β 2 ( ε − 3 ) ] 2 [ 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) ] .
The optimal solution for each decision variable must be greater than 0 [53]. Based on the optimal solutions of the decision variables calculated in Section 4 and Section 5, the following inequalities can be obtained: 16 b 2 ( ε − 1 ) 2 μ 2 − 2 b ( 3 ε − 72 ) ( ε − 1 ) β 2 μ − β 4 ( 2 ε − 3 ) > 0 , 4 μ ( ε + 2 ) ( ε − 1 ) b + β 2 ( ε + 3 ) < 0 , 4 b ( ε − 1 ) μ − β 2 ( ε − 3 ) < 0 , 2 μ ( ε − 1 ) b + β 2 < 0 , and 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) < 0 . □
From these inequalities, we obtain:
x s 23 * − x s 21 * < 0 , x s 22 * − x s 23 * > 0 , x s 22 * − x s 21 * > 0 .
Thus:
x s 22 * > x s 21 * > x s 23 * .
Proof of Proposition 2. 
Based on Lemma 1, Lemma 2, and Lemma 3, the following conclusions can be drawn:
x m 23 * − x m 21 * = − ( a − b c ) β 3 [ ( ε 2 − 2 ε + 3 ) β 2 + b ( ε − 1 ) μ ] 2 { b ( ε − 1 ) μ [ 16 b ( ε − 1 ) μ − 2 ( 3 ε − 7 ) β 2 ] + ( 3 − 2 ε ) β 4 } [ ( 3 − ε ) β 2 + 4 b ( ε − 1 ) μ ] ,
x m 22 * − x m 23 * = − ( a − b c ) β ( ε − 1 ) { b μ [ 16 μ ( ε − 1 ) b − 2 β 2 ( ε − 3 ) ] − ε β 4 } 2 { 2 b ( ε − 1 ) μ [ b ( ε − 1 ) μ − ( 3 ε − 7 ) β 2 ] + β 4 ( 3 − 2 ε ) } [ 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) ] ,
x m 22 * − x m 21 * = − ( a − b c ) β [ 2 μ ( ε − 1 ) b + β 2 ] [ 4 b ( ε − 1 ) μ − ( ε − 3 ) β 2 ] [ 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) ] .
Each decision variable’s optimal solution must satisfy: ( ε 2 − 2 ε + 3 ) β 2 + b ( ε − 1 ) μ < 0 , 2 b ( ε − 1 ) μ [ b ( ε − 1 ) μ − ( 3 ε − 7 ) β 2 ] + β 4 ( 3 − 2 ε ) > 0 , 2 μ ( ε − 1 ) b + β 2 < 0 , ( 3 − ε ) β 2 + 4 b ( ε − 1 ) μ < 0 , b ( ε − 1 ) μ [ 16 b ( ε − 1 ) μ − 2 ( 3 ε − 7 ) β 2 ] < 0 , 16 μ ( ε − 1 ) b − 2 β 2 ( ε − 3 ) < 0 , 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) < 0 , 2 b ( ε − 1 ) μ − β 2 ( ε − 2 ) < 0 , and 4 b ( ε − 1 ) μ − ( ε − 3 ) β 2 < 0 . □
Thus, the product ranking under different cooperation models for the supply chain members is:
x m 23 * − x m 21 * < 0 ,   x m 22 * − x m 23 * > 0 ,   x m 22 * − x m 21 * > 0 .
Thus, x m 22 * > x m 21 * > x m 23 * .
Proof of Proposition 3. 
Based on Lemma 1, Lemma 2, and Lemma 3, we have:
π t 21 * = π s 21 * + π m 21 * = − μ ( ε − 1 ) ( a − b c ) 2 [ 2 b μ ( ε − 1 ) − β 2 ( ε − 2 ) ] 2 [ 4 b ( ε − 1 ) μ − β 2 ( ε − 3 ) ] 2 .
For the fully cooperative model, the equation becomes:
π t 22 * = μ 2 ( ε − 1 ) ( a − b c ) 2 ( 2 b μ − β 2 ) ( ε − 1 ) μ + 2 β 2 μ .
The optimal result of cost sharing can be obtained as follows:
π t 23 * = π s 23 * + π m 23 * ,
π t 23 * = [ 384 b 3 ( ε − 1 ) 3 μ 3 − 32 b 2 ( ε − 1 ) 2 ( 5 ε − 13 ) β 2 μ 2 + 2 ( ε − 15 ) b ( ε − 1 ) ( 3 ε − 5 ) β 4 μ + β 6 ( ε 2 − 15 ε + 18 ) ] μ ( ε − 1 ) ( a − b c ) 2 16 [ 16 b 2 ( ε − 1 ) 2 μ 2 − 2 b ( ε − 1 ) ( 3 ε − 7 ) β 2 μ − ( 2 ε − 3 ) β 4 ] 2 ,
π t 23 * − π t 21 * = μ ( ε − 1 ) ( a − b c ) 2 [ 2 b μ ( ε − 1 ) − β 2 ( ε − 2 ) ] 2 [ 4 b ( ε − 1 ) μ − β 2 ( ε − 3 ) ] 2 + [ 384 b 3 ( ε − 1 ) 3 μ 3 − 32 b 2 ( ε − 1 ) 2 ( 5 ε − 13 ) β 2 μ 2 + 2 ( ε − 15 ) b ( ε − 1 ) ( 3 ε − 5 ) β 4 μ + β 6 ( ε 2 − 15 ε + 18 ) ] μ ( ε − 1 ) ( a − b c ) 2 16 [ 16 b 2 ( ε − 1 ) 2 μ 2 − 2 b ( ε − 1 ) ( 3 ε − 7 ) β 2 μ − ( 2 ε − 3 ) β 4 ] 2 ,
π t 22 * − π t 23 * = μ ( ε − 1 ) 2 ( a - b c ) 2 { [ 64 b 3 ( ε − 1 ) 2 μ 3 ( 2 b ( ε − 1 ) μ − 1 ) ] + 2 b 2 μ 2 ( ε − 1 ) [ 64 b 2 ( ε − 1 ) 2 μ 2 − 21 ( ε + 1 ) ( 7 ε − 9 ) ] + β 6 ε [ 6 b ( ε − 1 ) ( ε − 2 ) + 1 ] } 16 [ 2 b ( ε − 1 ) μ − ( ε − 2 ) β 2 ] [ 16 b 2 ( ε − 1 ) 2 μ 2 − 2 b ( ε − 1 ) ( 3 ε − 7 ) β 2 μ − ( 2 ε − 3 ) β 4 ] 2 ,
π t 22 * − π t 21 * = ( ε − 1 ) ( a − b c ) 2 { [ 16 b 2 ( ε − 1 ) 2 μ 3 + 4 b μ 2 ( ε − 1 ) [ ( − 2 β 2 + b ) ε + 6 β 2 − b ] − [ ( 4 b − β 2 ) ε 2 + ( 6 β 2 − 12 b ) ε − 9 β 2 + 8 b ] β 2 μ + β 4 ( ε − 2 ) 2 } 2 [ 4 b ( ε − 1 ) μ − ( ε − 3 ) β 2 ] 2 [ 2 b ( ε − 1 ) μ − ( ε − 2 ) β 2 ] .
Since the optimal solutions of the decision variables must be greater than 0, there exists:
( 2 a σ 2 + a c σ − c μ m ) b 2 + ( 3 α σ μ s + α 2 c − 3 a μ m ) b + a α 2 μ s + b c μ m β 2 < 0 ,
2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 < 0 ,
2 a σ 2 + a c σ − c μ m < 0 ,
4 ( a + b c ) ( b 2 σ 2 μ s + 2 b α σ μ s − 2 b μ m μ s + α 2 μ s ) + β 2 μ m ( 5 b c + a ) < 0 ,
3 α σ μ s + α 2 c − 3 a μ m < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m < 0 ,
b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m < 0 .
Since μ s = μ m = μ , a = β , let 2 b ( ε − 1 ) μ m + α 2 = 2 b ( ε − 1 ) μ + β 2 = X < 0 , then:
( a + b c ) [ 2 b ( ε − 1 ) μ + β 2 ] − β 2 b c ( ε − 1 ) < 0 = ( a + b c ) X − β 2 b c ( ε − 1 ) < 0 ,
( a + b c ) X < β 2 b c ( ε − 1 ) < 0 ,
64 b 3 ( ε − 1 ) 2 μ 3 ( 2 b ( ε − 1 ) μ − 1 ) < 0 ,
2 b 2 μ 2 ( ε − 1 ) [ 64 b 2 ( ε − 1 ) 2 μ 2 − 21 ( ε + 1 ) ( 7 ε − 9 ) ] < 0 .
Therefore, we have π t 23 * − π t 21 * < 0 , π t 22 * − π t 23 * > 0 , π t 22 * − π t 21 * > 0 . □
That is, π t 22 * > π t 21 * > π t 23 * .
Proof of Proposition 4. 
From Lemma 4, Lemma 5, and Lemma 6, we can get following equations:
x s 33 ∗ − x s 31 ∗ = ( a − b c ) { − 4 b 3 c μ σ 2 + 4 [ − 2 σ ( c μ + σ ) β + a μ σ 2 + 2 c μ 2 ] b 2 + [ − ( 7 c μ + 16 σ ) β 2 + 4 ( 2 a σ + 4 ) μ β − 8 a μ 2 ] b + 7 a β 2 μ − 12 β 3 } 2 [ 2 b 2 σ 2 + 4 b ( β σ − μ ) + 3 β 2 ] ( 4 b 2 σ 2 + 8 b ( β σ − μ ) + 7 β 2 μ ) ,
x s 33 ∗ − x s 32 ∗ = β ( a − b c ) [ 2 b 2 σ 2 + 4 ( β σ − μ ) b + 3 β 2 ] [ 4 b 2 σ 2 + 8 ( β σ − μ ) b + 7 β 2 ] [ b 2 σ 2 + 2 ( β σ − μ ) b + β 2 ] .
Since the optimal solution of each decision variable must be greater than 0, there exists:
( 2 a σ 2 + a c σ − c μ m ) b 2 + ( 3 α σ μ s + α 2 c − 3 a μ m ) b + a α 2 μ s + b c μ m β 2 < 0 ,
4 ( a + b c ) ( b 2 σ 2 μ s + 2 b α σ μ s - 2 b μ m μ s + α 2 μ s ) + β 2 μ m ( 5 b c + a ) < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m < 0 ,
2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 < 0 ,
3 α σ μ s + α 2 c − 3 a μ m < 0 ,
b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m < 0 ,
2 a σ 2 + a c σ − c μ m < 0 ,   4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m < 0 .
Because μ s = μ m = μ , α = β , let b 2 σ 2 + 2 b α σ − 2 b μ + β 2 = A , z = a − b c , and we can get:
x s 33 ∗ − x s 31 ∗ = z [ 4 A ( μ z − 2 β ) + 3 z β 2 μ − 8 β 3 ] ( 4 A + 2 β 2 ) ( 4 A + 3 β 2 ) < 0 ,
x s 33 ∗ − x s 32 ∗ = 2 ( a − b c ) ( 2 A + β 2 ) β ( 4 A + 3 β 2 ) ( A + β 2 ) < 0 .
Then x s 33 ∗ − x s 31 ∗ > 0 , x s 33 ∗ − x s 32 ∗ < 0 , which means x s 32 ∗ > x s 33 ∗ > s 31 ∗ . □
Proof of Proposition 5. 
Based on Lemma 4, Lemma 5, and Lemma 6, the following conclusions can be drawn:
x m 33 * − x m 32 * = ( β + b σ ) ( a − b c ) [ 4 b 4 σ 4 + 16 ( β σ 3 − μ σ 2 ) b 3 + ( 25 β 2 σ 2 − 32 β μ σ + 16 μ 2 ) b 2 − 18 β 2 ( μ − β σ ) b + β 4 ] 2 ( b 2 σ 2 + 2 b ( β σ − μ ) + β 2 ) [ 4 b 2 σ 2 + 8 b ( β σ − μ ) + 7 β 2 ] ( b 2 σ 2 + 2 b ( β σ − μ ) + 2 β 2 ) ,
x m 33 * − x m 31 * = − ( β + b σ ) ( a − b c ) β 4 2 ( b 2 σ 2 + 2 b ( β σ − μ ) + β 2 ) [ 4 b 2 σ 2 + 8 b ( β σ − μ ) + 7 β 2 ] ( 2 b 2 σ 2 + 4 b ( β σ − μ ) + 3 β 2 ) ,
x m 32 − x m 31 = ( a − b c ) ( β + b σ ) ( b 2 σ 2 + 2 b ( β σ − μ ) + β 2 ) ( 2 b 2 σ 2 + 4 b ( β σ − μ ) + 3 β 2 ) ( b 2 σ 2 + 2 b ( β σ − μ ) + β 2 ) .
Since the optimal solution of each decision variable must be greater than 0, there exists:
2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 < 0 ,
2 a σ 2 + a c σ − c μ m < 0 ,
( 2 a σ 2 + a c σ − c μ m ) b 2 + ( 3 α σ μ s + α 2 c − 3 a μ m ) b + a α 2 μ s + b c μ m β 2 < 0 ,
b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m < 0 ,
4 ( a + b c ) ( b 2 σ 2 μ s + 2 b α σ μ s − 2 b μ m μ s + α 2 μ s ) + β 2 μ m ( 5 b c + a ) < 0 ,
3 α σ μ s + α 2 c − 3 a μ m < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m < 0 .
Because μ s = μ m = μ , α = β , let b 2 σ 2 + 2 b α σ − 2 b μ + β 2 = A . Then:
x m 33 ∗ − x m 32 ∗ = ( β + b σ ) ( a − b c ) { 9 A β 2 + σ 2 b 2 [ 4 A + 12 β 2 + 8 ( β σ − μ ) b ] − 16 ( 2 β σ − μ ) b 2 μ − 5 β 4 } 2 A ( 4 A + 3 β 2 ) ( A + β 2 ) < 0 ,
x m 33 * − x m 31 * = − ( β + b σ ) ( a - b c ) β 4 2 A ( 4 A + 3 β 2 ) ( 2 A + β 2 ) > 0 ,
x m 32 − x m 31 = ( a − b c ) ( β + b σ ) − ( 2 A + β 2 ) > 0 .
Which means x m 32 * > x m 33 * > x m 31 * . □
Proof of Proposition 6. 
Based on Lemma 4, Lemma 5, and Lemma 6, the following results can be drawn:
π t 33 * − π t 32 * = ( a − b c ) 2 μ 3 ( A − β 2 ) 2 ( 4 A + 3 β 2 ) ( A + β 2 ) < 0 ,
π t 32 * − π t 31 * = ( a − b c ) 2 μ 2 [ b 2 σ 2 + b ( 2 β σ − 2 μ ) + 2 β 2 ] [ ( 7 b 2 σ 2 + 14 b β σ − 14 b μ + 7 β 2 ) μ 2 − 8 β 2 ] − 4 ( 2 b 2 σ 2 + 4 b β σ − 4 b μ + 3 β 2 ) 2 ( b 2 σ 2 + 2 b β σ − 2 b μ + 2 β 2 ) ,
π t 33 * − π t 31 ∗ = − ( a − b c ) 2 μ [ 12 ( b 2 σ 2 + 2 b α σ − 2 b μ ) + 13 β 2 ] 8 [ 4 ( b 2 σ 2 + 2 b β σ − 2 b μ ) + 7 β 2 ] [ b 2 σ 2 + 2 b β σ − 2 b μ + β 2 ] + ( a − b c ) 2 μ 2 [ 3 ( b 2 σ 2 + 2 b α σ − 2 b μ + β 2 ) μ + β 2 μ ] 2 [ 2 ( b 2 σ 2 + 2 b β σ − 2 b μ + β 2 ) μ + β 2 μ ] 2 .
Since the optimal solution of each decision variable must be greater than 0, there exists:
( 2 a σ 2 + a c σ − c μ m ) b 2 + ( 3 α σ μ s + α 2 c − 3 a μ m ) b + a α 2 μ s + b c μ m β 2 < 0 ,
2 b 2 σ 2 μ s + 4 b α σ μ s + 2 α 2 μ s − 4 b μ s μ m + μ m β 2 < 0 ,
2 a σ 2 + a c σ − c μ m < 0 ,
4 ( a + b c ) ( b 2 σ 2 μ s + 2 b α σ μ s - 2 b μ m μ s + α 2 μ s ) + β 2 μ m ( 5 b c + a ) < 0 ,
3 α σ μ s + α 2 c − 3 a μ m < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m < 0 ,
b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 < 0 ,
4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m < 0 .
Because μ s = μ m = μ , α = β , let b 2 σ 2 + 2 b α σ − 2 b μ + β 2 = A , then:
b 2 σ 2 + 2 b α σ − 2 b μ + β 2 + β 2 2 = A + β 2 2 < 0 ,
4 b 2 σ 2 + 8 b β σ − 8 b μ + 7 β 2 = 4 A + 3 β 2 < 0 ,
4 ( a + b c ) ( b 2 σ 2 + 2 b α σ - 2 b μ + β 2 ) μ + β 2 μ ( 5 b c + a ) = 4 A ( a + b c ) + β 2 ( 5 b c + a ) < 0 ,
π t 33 * − π t 32 * = ( a − b c ) 2 μ 3 ( A − β 2 ) 2 ( 4 A + 3 β 2 ) ( A + β 2 ) < 0 ,
π t 32 * − π t 31 ∗ = − ( a − b c ) 2 μ 2 ( 7 A − β 2 ) 4 ( 2 A + β 2 ) 2 > 0 ,
π t 33 * − π t 31 ∗ = ( a − b c ) 2 μ ( − A β 4 ) β 2 8 ( 16 A 4 + 16 A 2 β 4 + 3 A β 6 ) > 0 .
Which means π t 32 * > π t 33 ∗ > π t 31 * . □
Proof of Proposition 7. 
From Lemma 1 and Lemma 4, we can obtain the following equations:
x s 31 * − x s 21 * = 2 β ( a − b c ) μ s μ m [ σ 2 ( ε − 1 ) 2 b 2 + 2 α σ ( ε − 1 ) b + α 2 ε ] [ ( 4 b ( ε − 1 ) μ m + 2 α 2 ) μ s − β 2 μ m ( ε − 1 ) ] [ 2 ( b σ + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] ,
x m 31 * − x m 21 * = μ s ( a − b c ) [ ( σ ( ε − 1 ) b + α ε ) β 2 μ m + 2 ( ( 2 ( ε − 1 ) μ m + α σ ) b σ + α 2 σ − 2 α ε μ m ) b μ s ] [ ( 4 ( ε − 1 ) μ m + 4 α 2 ) μ s − β 2 μ m ( ε − 1 ) ] [ 2 ( b σ + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] ,
π m 31 * − π m 21 * = − ( a − b c ) 2 μ s 2 μ m [ σ 2 ( ε − 1 ) 2 b 2 + 2 α σ ( ε − 1 ) b + α 2 ε ] [ 8 μ s 2 μ m σ 2 ( ε − 1 ) b 3 + 4 μ s 2 ( 4 ( ε − 1 ) μ m 2 + 4 σ α ( ε − 1 ) μ m + α 2 σ 2 ) b 2 + 8 μ s 2 ( ( ε − 2 ) μ m + σ α ) α 2 b + 4 α 4 μ s 2 + β 4 μ m 2 ( ε − 1 ) ] 2 [ ( 4 b ( ε − 1 ) μ m + 2 α 2 ) μ s − β 2 μ m ( ε − 1 ) ] 2 [ 2 ( b σ + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] 2 ,
π s 31 * − π s 21 * = − ( a − b c ) 2 μ s 2 μ m [ σ 2 ( ε − 1 ) b 2 + 2 α σ ( ε − 1 ) b + α 2 ε ] 8 [ ( 4 b ( ε − 1 ) μ m + 2 α 2 ) μ s − β 2 μ m ( ε − 1 ) ] [ 2 ( b σ + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] .
Among them, according to the previous conditions, each decision variable must be greater than 0, that is: x m 31 * > x m 21 * , x s 31 * < x s 21 * , π m 31 * < π m 21 * , π s 31 * < π s 21 * . □
If the supplier’s emission reductions are the same in this model, i.e., x s 31 * = x s 21 * , then ε = b σ ( b σ + 2 α ) b σ + 2 b α σ + α 2 , and bringing the ε into the above equation, the profit size comparison of the supply chain is π m 31 * − π m 21 * = 0 , π s 31 * − π s 21 * = 0 , x m 31 * − x m 21 * > 0 .
The government R&D cost subsidy and per-unit product emission reduction subsidy expenditure are, respectively:
π g 21 = 1 2 ε ( x m 21 * ) 2 ,
π g 31 = σ x m 31 * ( a − b p + α x m 31 * + β x s 31 * ) .
Bring the ε into the Formulas (A67) and (A68) and subtract the two formulas:
π g 31 − π g 21 = − b σ μ s μ m ( α + b σ ) ( a − b c ) 2 [ b ( b 2 σ 2 + 3 b α σ + α 2 ) μ s + 2 β 2 α 2 ] 2 [ 2 ( b 2 σ 2 + ( 2 α σ − 2 μ m ) b + α 2 ) μ s ) + β 2 μ m ] 2 < 0 .
So, π g 31 < π g 21 . It can be seen that under the same emission reduction, the R&D cost subsidy expenditure is greater than the subsidy expenditure per-unit product emission reduction, so the above proof results can be obtained to be correct.
Proof of Proposition 8. 
From Lemma 2 and Lemma 5, we can obtain the following equations:
x m 32 * − x m 22 * = μ s ( a − b c ) [ b ( ( 2 ( 1 − ε ) μ m + α σ ) b σ + α 2 σ − 2 α ε μ m ) μ s + ( σ ( ε − 1 ) b + α ε ) β 2 μ m ] [ ( 2 b ( ε − 1 ) μ m + α 2 ) μ s − β 2 μ m ( ε − 1 ) ] [ ( b 2 σ 2 + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] ,
x s 32 * − x s 22 * = 2 β ( a − b c ) μ s μ m [ σ 2 ( ε − 1 ) 2 b 2 + 2 α σ ( ε − 1 ) b + α 2 ε ] [ ( 2 b ( ε − 1 ) μ m + 2 α 2 ) μ s − β 2 μ m ( ε − 1 ) ] [ ( b 2 σ 2 + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] ,
π t 32 * − π t 22 * = − ( a − b c ) 2 μ s 2 μ m [ σ 2 ( ε − 1 ) b 2 + 2 α σ ( ε − 1 ) b + α 2 ε ] 2 [ ( 2 b ( ε − 1 ) μ m + α 2 ) μ s − β 2 μ m ( ε − 1 ) ] [ 2 ( b 2 σ 2 + ( 2 σ α − 2 μ m ) b + α 2 ) μ s + β 2 μ m ] .
Which means x m 32 * > x m 22 * , x s 32 * < x s 22 * , π t 32 * < π t 22 * . □
If in this model the supplier’s emission reduction is the same, i.e., x s 32 * = x s 22 * , so ε = b σ ( b σ + 2 α ) b σ + 2 b α σ + α 2 , then the ε is brought into the above equation, the profit size of the supply chain is compared: π t 32 * − π t 22 * = 0 , x m 32 * − x m 22 * > 0 , and the government R&D cost subsidy and per-unit production emission reduction sub-sidy expenditure are:
π g 22 = 1 2 ε ( x m 22 * ) 2 ,
π g 32 = σ x m 32 * ( a - b p + α x m 32 * + β x s 32 * ) .
Bring the ε into the Formulas (A72) and (A73) and subtract the two formulas:
π g 32 − π g 22 = − b σ μ s μ m ( α + b σ ) ( a − b c ) 2 [ 2 b 4 σ 4 μ s + 2 ( 5 α σ − 2 μ m ) b 3 σ 3 μ s + ( 2 ( 8 α 2 σ 2 − 6 α σ μ m ) μ s + β 2 σ 2 μ m ) b 2 + ( 2 ( 5 α 2 σ 2 − 2 α 2 μ m ) μ s + 3 β 2 α σ μ m ) b + 2 α 4 μ s ] α 2 [ ( b 2 σ 2 + ( 2 α σ − 2 μ m ) b + α 2 ) μ s ) + β 2 μ m ] 2 [ 2 ( b 2 σ 2 + ( 2 α σ − 2 μ m ) b + α 2 ) μ s ) + β 2 μ m + 2 α 2 μ s ] .
The calculation process is the same as above, π g 32 − π g 22 < 0 ,   π g 32 < π g 22 , and it can be seen that under the same emission reduction, the R&D subsidy expenditure is greater than the subsidy expenditure per-unit production emission reduction, so the above proof results can be obtained to be correct.
Proof of Proposition 9. 
From Lemma 3 and Lemma 6, we can obtain the following equations:
x m 33 * − x m 23 * = − α [ ( 8 b μ s − β 2 ) ( ε − 1 ) μ m + 2 μ s α 2 ] ( a − b c ) 4 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 4 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 4 μ s − ( a − b c ) ( b σ + α ) ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) 2 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) ,
x s 33 * − x s 23 * = − [ 8 b ( ε - 1 ) μ m + 3 α 2 ] β ( ε − 1 ) μ m ( a − b c ) 4 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 4 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 4 μ s − 2 ( a − b c ) β μ m 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + 3 β 2 μ m ,
π m 33 * − π m 23 * = − [ ( 8 b μ s − β 2 ) ( ε − 1 ) μ m + 4 μ s α 2 ] ( a − b c ) 2 ( ε − 1 ) μ m 16 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 16 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 8 α 4 μ s − ( a − b c ) 2 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) μ m 4 ( 4 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) ,
π s 33 * − π s 23 * = 4 ( 8 b μ s − 3 β 2 ) [ b ( ε − 1 ) μ m + 2 α 2 ] [ ( 8 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 2 μ s ] ( ε − 1 ) 2 μ m 2 ( a − b c ) 2 − [ 2 b ( ε − 1 ) 2 ( 8 b μ s − 3 β 2 ) μ m 2 + 2 α 2 ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + α 4 μ s ] 2 − ( a − b c ) 2 ( 4 b 2 σ 2 μ s + 8 b α σ μ s − 8 b μ m μ s + 4 α 2 μ s + β 2 μ m ) μ m 8 ( 4 ( b 2 σ 2 + 2 b ( α σ − μ m ) + α 2 ) μ s + 3 β 2 μ m ) ( b 2 σ 2 + 2 b ( α σ − μ m ) + α 2 ) .
Among them, according to the previous conditions, each decision variable must be greater than 0, thus satisfying: x m 33 * − x m 23 * > 0 , x s 33 * − x s 23 * < 0 , π m 33 * − π m 23 * < 0 , π s 33 * − π s 23 * < 0 , so: x m 33 * > x m 23 * , x s 33 * < x s 23 * , π m 33 * < π m 23 * , π s 33 * < π s 23 * . □
If the supplier’s emission reductions are the same in this model, i.e., x s 33 * = x s 23 * , the ε = b σ ( b σ + 2 α ) b σ + 2 b α σ + α 2 , and bringing the ε into the above equation, the profit size comparison of the supply chain is π m 33 * − π m 23 * = 0 , π s 33 * − π s 23 * = 0 , x m 33 * − x m 23 * > 0 .
The government R&D cost subsidy and per-unit production emission reduction sub-sidy expenditure are, respectively:
π g 23 = 1 2 ε ( x m 23 * ) 2 ,
π g 33 = σ x m 33 * ( a - b p + α x m 33 * + β x s 33 * ) .
Bringing the ε into the Formulas (A78) and (A79) and subtracting the two equations, we can get the fol-lowing results: π g 33 − π g 23 < 0 , so π g 33 < π g 23 . It can be seen that under the same emission reduction, the R&D subsidy expenditure is greater than the subsidy expenditure per-unit product emission reduction, so the above proof results can be obtained to be correct.
Proof of Proposition 10. 
From Section 6.3.1, we can get following equation:
Δ S W 1 = S W 21 ∗ − S W 31 ∗ = μ s μ m ( a − b c ) 2 [ ( ε − 1 ) 2 ( 7 b μ s − β 2 ) μ m + α 2 μ s ( ε − 3 ) ] 2 [ ( 4 b μ s − β 2 ) ( ε − 1 ) μ m + 2 α 2 μ s ] 2 + μ s μ m ( a − b c ) 2 [ 5 b 2 σ 2 + ( 8 α σ − 7 μ m ) b + 3 α 2 ) μ s − ( 4 b σ 2 + 4 α σ − μ m ) β 2 ] 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b μ s + α 2 ) + μ m β 2 ] 2 .
Among the above equation, according to the previous conditions, each decision vari-able must be greater than 0, thus satisfying:
( ε − 1 ) 2 ( 7 b μ s − β 2 ) μ m + α 2 μ s ( ε − 3 ) > 0 ,
5 b 2 σ 2 + ( 8 α σ − 7 μ m ) b + 3 α 2 ) μ s − ( 4 b σ 2 + 4 α σ − μ m ) β 2 > 0 .
From the above conditions, S W 21 ∗ − S W 31 ∗ > 0 can be obtained. □
Proof of Proposition 11. 
Δ S W 2 = S W 22 * − S W 32 * = μ s μ m ( a − b c ) 2 [ ( ε − 1 ) 2 ( 3 b μ s − β 2 ) μ m − α 2 μ s ( ε − 3 ) ] 2 [ ( 2 b μ s − β 2 ) ( ε − 1 ) μ m + α 2 μ s ] 2 + μ s μ m ( a − b c ) 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) ( 2 b 2 σ 2 + 3 ( α σ − μ m ) b + α 2 ) μ s 2 + μ m ( 5 b 2 σ 2 + ( 8 α σ − 7 μ m ) b + 3 α 2 ) μ s β 2 + β 4 μ m 2 ] 2 [ ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 ] 2 [ 2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 ] .
Among the above equation, according to the previous conditions, each decision vari-able must be greater than 0, thus satisfying:
( ε − 1 ) 2 ( 3 b μ s − β 2 ) μ m − α 2 μ s ( ε − 1 ) > 0 ,
2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 < 0 ,
2 ( b 2 σ 2 + 2 ( α σ − μ m ) b + α 2 ) μ s + μ m β 2 < 0 .
Therefore, S W 22 * − S W 32 * > 0 . □
Proof of Proposition 12. 
Δ S W 3 = S W 23 * − S W 33 * = − μ m ( a − b c ) 2 [ 12 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + β 2 μ m ] [ 4 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + 3 β 2 μ m ] ( b 2 σ 2 + 2 b α σ − 2 b μ m + α ) + b μ m ( a - b c ) 2 [ μ m − 2 σ ( α + b σ ) ] [ 4 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + β 2 μ m ] 2 8 [ 4 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + 3 β 2 μ m ] 2 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α ) 2 .
Among the above equation, according to the previous conditions, each decision vari-able must be greater than 0, thus satisfying:
12 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + β 2 μ m < 0 ,
b 2 σ 2 + 2 b α σ − 2 b μ m + α < 0 ,
4 ( b 2 σ 2 + 2 b α σ − 2 b μ m + α 2 ) μ s + 3 β 2 μ m < 0 ,
μ m − 2 σ ( α + b σ ) > 0 .
and S W 23 * − S W 33 * > 0 can be obtained. □

References

  1. Chen, Q.; Zhang, H.; Lau, Y.Y.; Wang, T.; Wang, W.; Zhang, G. Climate change, carbon peaks, and carbon neutralization: A bibliometric study from 2006 to 2023. Sustainability 2023, 15, 5723. [Google Scholar] [CrossRef] [Scilit]
  2. Abbasi, K.R.; Zhang, Q.; Alotaibi, B.S.; Abuhussain, M.A.; Alvarado, R. Toward sustainable development goals 7 and 13: A comprehensive policy framework to combat climate change. Environ. Impact Assess. Rev. 2024, 105, 107415. [Google Scholar] [CrossRef] [Scilit]
  3. Adun, H.; Ampah, J.D.; Bamisile, O.; Hu, Y. The synergistic role of carbon dioxide removal and emission reductions in achieving the Paris Agreement goal. Sustain. Prod. Consump. 2024, 45, 386–407. [Google Scholar] [CrossRef] [Scilit]
  4. Popescu, C.; Apostu, S.A.; Rădulescu, I.G.; Mureșan, J.D.; Brezoi, A.G. Energizing the Now: Navigating the Critical Landscape of Today’s Energy Challenges—An In-Depth Review. Energies 2024, 17, 675. [Google Scholar] [CrossRef] [Scilit]
  5. Jiang, X.; Xie, F. Decomposition analysis of carbon emission drivers and peaking pathways for key sectors under China’s dual carbon goals: A case study of Jiangxi Province, China. Sustainability 2024, 16, 5811. [Google Scholar] [CrossRef] [Scilit]
  6. Zhao, S.; Song, Q.; Liu, L.; Li, J.; Zhao, D. Uncovering the lifecycle carbon emissions and its reduction pathways: A case study of petroleum refining enterprise. Energy Convers. Manag. 2024, 301, 118048. [Google Scholar] [CrossRef] [Scilit]
  7. Hasan, J.; Thomas, A.; Tomos, O. Sustainable supply chain practices in the oil and gas industry: A case study. Sustainability. 2024, 16, 1720. [Google Scholar] [CrossRef] [Scilit]
  8. Wu, W.; Bi, S.; Zhan, Y.; Gu, X. Supply chain digitalization and energy efficiency (gas and oil): How do they contribute to achieving carbon neutrality targets? Energy Econ. 2025, 142, 108140. [Google Scholar] [CrossRef] [Scilit]
  9. Chen, J.; Wang, W.; Sun, W.; Jiao, Y.; He, Y.; Li, D.; Gong, J. Tapping into the potential CO2 emission reduction of a crude oil transportation system from carbon footprint perspective. J. Clean. Prod. 2023, 413, 137409. [Google Scholar] [CrossRef] [Scilit]
  10. Abdussalam, O.; Fello, N.; Chaabane, A. Exploring options for carbon abatement in the petroleum sector: A supply chain optimization-based approach. Int. J. Syst. Sci. Oper. Logist. 2023, 10, 2005174. [Google Scholar] [CrossRef] [Scilit]
  11. Wang, J.; Shao, X.; Liu, G.; Yuan, Z. Domestic substitution and supply chain performance: The impact of technology and price subsidies. Transport. Res. E-Log. 2025, 201, 104218. [Google Scholar] [CrossRef] [Scilit]
  12. Yan, Z.; Hu, H.; Wang, Z.; Liang, Z.; Kong, W. The effect of government subsidies on cooperative green innovation of supply chains from the perspective of cost sharing. J. Bus. Ind. Mark. 2025, 40, 415–431. [Google Scholar] [CrossRef] [Scilit]
  13. Jegers, M.; Buts, C.; Van Puyvelde, S. Welfare effects of variable and fixed cost subsidies in profit, non-profit and mixed markets. Manag. Decis. Econ. 2024, 45, 4138–4156. [Google Scholar] [CrossRef] [Scilit]
  14. Jin, W.; Yang, J.; Wang, C. Cost subsidy or environmental regulation? The effects of government interventions on environmental quality and 3BL performance. Int. J. Prod. Econ. 2024, 270, 109180. [Google Scholar] [CrossRef] [Scilit]
  15. Zhang, H.; Duan, X.; Jiang, J. Fixed rebate subsidy vs. unit price subsidy: Incentive effect on the biomass power supply chain. Energy Policy 2024, 187, 114031. [Google Scholar] [CrossRef] [Scilit]
  16. Zhang, X.; Yousaf, H.A.U. Green supply chain coordination considering government intervention, green investment, and customer green preferences in the petroleum industry. J. Clean. Prod. 2020, 246, 118984. [Google Scholar] [CrossRef] [Scilit]
  17. Van Asselt, H.; Fragkos, P.; Peterson, L.; Fragkiadakis, K. The environmental and economic effects of international cooperation on restricting fossil fuel supply. Int. Environ. Agreem. Politics Law Econ. 2024, 24, 141–166. [Google Scholar] [CrossRef] [Scilit]
  18. Xu, S.; Govindan, K.; Wang, W.; Yang, W. Supply chain management under cap-and-trade regulation: A literature review and research opportunities. Int. J. Proc. Econ. 2024, 271, 109199. [Google Scholar] [CrossRef] [Scilit]
  19. Huang, Y.; Wang, Y.; Liu, N. A two-stage energy management for heat-electricity integrated energy system considering dynamic pricing of Stackelberg game and operation strategy optimization. Energy 2022, 244, 122576. [Google Scholar] [CrossRef] [Scilit]
  20. Amiri-Pebdani, S.; Alinaghian, M.; Khosroshahi, H. Energy pricing and investigating international trade law considering renewable energy investments under the cap-and-trade system: A game theoretic approach. Appl. Energy 2025, 381, 125155. [Google Scholar] [CrossRef] [Scilit]
  21. Huang, X.; Lu, X.; Sun, Y.; Yao, J.; Zhu, W. A Comprehensive Performance Evaluation of Chinese Energy Supply Chain under “Double-Carbon” Goals Based on AHP and Three-Stage DEA. Sustainability 2022, 14, 10149. [Google Scholar] [CrossRef] [Scilit]
  22. Liu, W.; Gao, Y.; Yuan, C.; Wang, D.; Tang, O. The impact of carbon neutrality policies on the stock market from a supply chain perspective. Ind. Manag. Data. Syst. 2023, 123, 2631–2658. [Google Scholar] [CrossRef] [Scilit]
  23. Fu, C.; Yu, W.; Wang, Y.; Feng, W.; Chen, J. Low-carbon strategy analysis in a construction supply chain considering government subsidies and contract design. J. Clean. Prod. 2024, 470, 143326. [Google Scholar] [CrossRef] [Scilit]
  24. Ma, S.; He, Y.; Gu, R.; Li, S. Sustainable supply chain management considering technology investments and government intervention. Transport. Res. E-Log. 2021, 149, 102290. [Google Scholar] [CrossRef] [Scilit]
  25. Che, C.; Chen, Y.; Zhang, X.; Zhang, Z. The Impact of Different Government Subsidy Methods on Low-Carbon Emission Reduction Strategies in Dual-Channel Supply Chain. Complexity 2021, 1, 6668243. [Google Scholar] [CrossRef] [Scilit]
  26. Hussain, J.; Pan, Y.; Ali, G.; Xiaofang, Y. Pricing behavior of monopoly market with the implementation of green technology decision under emission reduction subsidy policy. Sci. Total Environ. 2020, 709, 136110. [Google Scholar] [CrossRef] [Scilit]
  27. Liu, X.; Lin, K.; Wang, L.; Ding, L. Pricing decisions for a sustainable supply chain in the presence of potential strategic customers. Sustainability 2020, 12, 1655. [Google Scholar] [CrossRef] [Scilit]
  28. Yu, J.J.; Tang, C.S.; Shen, Z.J.M. Improving consumer welfare and manufacturer profit via government subsidy programs: Subsidizing consumers or manufacturers? Manuf.-Serv. Oper. Manag. 2018, 20, 752–766. [Google Scholar] [CrossRef] [Scilit]
  29. Su, C.; Liu, X.; Du, W. Green supply chain decisions considering consumers’ low-carbon awareness under different government subsidies. Sustainability 2020, 12, 2281. [Google Scholar] [CrossRef] [Scilit]
  30. Fan, R.; Dong, L. The dynamic analysis and simulation of government subsidy strategies in low-carbon diffusion considering the behavior of heterogeneous agents. Energy Policy. 2018, 117, 252–262. [Google Scholar] [CrossRef] [Scilit]
  31. Alizamir, S.; Iravani, F.; Mamani, H. An analysis of price vs. revenue protection: Government subsidies in the agriculture industry. Manag. Sci. 2019, 65, 32–49. [Google Scholar] [CrossRef] [Scilit]
  32. Zu, Y.; Chen, L.; Fan, Y. Research on low-carbon strategies in supply chain with environmental regulations based on differential game. J. Clean. Prod. 2018, 177, 527–546. [Google Scholar] [CrossRef] [Scilit]
  33. Meng, Q.; Li, M.; Liu, W.; Li, Z.; Zhang, J. Pricing policies of dual-channel green supply chain: Considering government subsidies and consumers’ dual preferences. Sustain. Prod. Consump. 2021, 26, 1021–1030. [Google Scholar] [CrossRef] [Scilit]
  34. Yang, R.; Tang, W.; Zhang, J. Technology improvement strategy for green products under competition: The role of government subsidy. Eur. J. Oper. Res. 2021, 289, 553–568. [Google Scholar] [CrossRef] [Scilit]
  35. Barman, A.; De, P.K.; Chakraborty, A.K.; Lim, C.P.; Das, R. Optimal pricing policy in a three-layer dual-channel supply chain under government subsidy in green manufacturing. Math. Comput. Simulat. 2023, 204, 401–429. [Google Scholar] [CrossRef] [Scilit]
  36. Li, Y.; Jia, Z.; Qin, C.; Li, Z.; Ying, J. Analysis of recycling and emission reduction models under carbon taxes and government subsidies. Int. Rev. Econ. Financ. 2025, 99, 104052. [Google Scholar] [CrossRef] [Scilit]
  37. Bian, J.; Zhang, G.; Zhou, G. Manufacturer vs. Consumer Subsidy with Green Technology Investment and Environmental Concern. Eur. J. Oper. Res. 2020, 287, 832–843. [Google Scholar] [CrossRef] [Scilit]
  38. Wang, W.; Lin, W.; Cai, J.; Chen, M. Impact of demand forecast information sharing on the decision of a green supply chain with government subsidy. Ann. Oper. Res. 2023, 329, 953–978. [Google Scholar] [CrossRef] [Scilit]
  39. Lai, Z.; Lou, G.; Guo, Y.; Tu, X.; Zhao, Y. Optimal supply chain green innovation and subsidy strategy: Producer subsidy versus consumer subsidy. Ind. Manag. Data Syst. 2024, 124, 3350–3378. [Google Scholar] [CrossRef] [Scilit]
  40. Meng, Q.; Wang, Y.; Zhang, Z.; He, Y. Supply chain green innovation subsidy strategy considering consumer heterogeneity. J. Clean. Prod. 2021, 281, 125199. [Google Scholar] [CrossRef] [Scilit]
  41. Xiao, Q.; Gao, Z.; Zhang, Q.; Xia, Z. Pricing policies of dual-channel green supply chain: Considering manufacturers’ dual behavioural preferences and government subsidies. Int. J. Syst. Sci. Oper. Logist. 2024, 11, 2417347. [Google Scholar] [CrossRef] [Scilit]
  42. Wen, Y.; Liu, L. Comparative study on low-carbon strategy and government subsidy model of pharmaceutical supply chain. Sustainability 2023, 15, 8345. [Google Scholar] [CrossRef] [Scilit]
  43. He, X.; Jiang, J.; Hu, W. Cross effects of government subsidies and corporate social responsibility on carbon emissions reductions in an omnichannel supply chain system. Comput. Ind. Eng. 2023, 175, 108872. [Google Scholar] [CrossRef] [Scilit]
  44. Li, G.; Zheng, H.; Sethi, S.P.; Guan, X. Inducing downstream information sharing via manufacturer information acquisition and retailer subsidy. Decis. Sci. 2020, 51, 691–719. [Google Scholar] [CrossRef] [Scilit]
  45. Xia, Y.; Hui, Y.; Hongfu, H.; Siyuan, Z.; Qingling, Y. Low-carbon supplier selection in the presence of government subsidy. Manag. Decis. Econ. 2025, 46, 183–200. [Google Scholar] [CrossRef] [Scilit]
  46. Wang, S.; Song, H.; Liao, Y.; Diabat, A. Low-carbon innovation with government subsidy for achieving SDGs in a complementary supply chain. Ann. Oper. Res. 2025, 352, 325–349. [Google Scholar] [CrossRef] [Scilit]
  47. Li, Y.; Tong, Y.; Ye, F.; Song, J. The choice of the government green subsidy scheme: Innovation subsidy vs. product subsidy. Int. J. Prod. Res. 2020, 58, 4932–4946. [Google Scholar] [CrossRef] [Scilit]
  48. Chen, J.Y.; Dimitrov, S.; Pun, H. The impact of government subsidy on supply Chains’ sustainability innovation. Omega 2019, 86, 42–58. [Google Scholar] [CrossRef] [Scilit]
  49. Shi, W.; Min, K.J. Remanufacturing decisions and implications under material cost uncertainty. Int. J. Prod. Res. 2015, 53, 6421–6435. [Google Scholar] [CrossRef] [Scilit]
  50. Li, Z.; Pan, Y.; Yang, W.; Ma, J.; Zhou, M. Effects of government subsidies on green technology investment and green marketing coordination of supply chain under the cap-and-trade mechanism. Energy Econ. 2021, 101, 105426. [Google Scholar] [CrossRef] [Scilit]
  51. Zheng, J.; Zhao, H.; Fu, J. Diverse government subsidy modes in a supply chain considering different innovation dimensions. Soft. Comput. 2024, 28, 3973–3986. [Google Scholar] [CrossRef] [Scilit]
  52. Yi, Z.; Chen, Y.; Luo, S.; Huang, H. Agricultural supply chain finance considering interest or direct subsidy by government. Transport. Res. E-Log. 2025, 195, 103992. [Google Scholar] [CrossRef] [Scilit]
  53. Wang, Y.; Xu, X.; Zhu, Q. Carbon emission reduction decisions of supply chain members under cap-and-trade regulations: A differential game analysis. Comput. Ind. Eng. 2021, 162, 107711. [Google Scholar] [CrossRef] [Scilit]
  54. Xia, L.; Bai, Y.; Ghose, S.; Qin, J. Differential game analysis of carbon emissions reduction and promotion in a sustainable supply chain considering social preferences. Ann. Oper. Res. 2022, 310, 257–292. [Google Scholar] [CrossRef] [Scilit]
  55. Herbon, A.; David, I. Optimal manufacturer’s cost sharing ratio, shipping policy and production rate–A two-echelon supply chain. Oper. Res. Perspect. 2023, 10, 100264. [Google Scholar] [CrossRef] [Scilit]
  56. Zhou, H.; Liu, M.; Tan, Y. Long-term emission reduction strategy in a three-echelon supply chain considering government intervention and Consumers’ low-carbon preferences. Comput. Ind. Eng. 2023, 186, 109697. [Google Scholar] [CrossRef] [Scilit]
  57. Xu, C.; Jing, Y.; Shen, B.; Zhou, Y.; Zhao, Q.Q. Cost-sharing contract design between manufacturer and dealership considering the customer low-carbon preferences. Expert Syst. Appl. 2023, 213, 118877. [Google Scholar] [CrossRef] [Scilit]
  58. Cheng, S.; Zhang, F. Regulatory pressure and consumer environmental awareness in a green supply chain with retailer responsibility: A dynamic analysis. Manag. Decis. Econ. 2022, 43, 1133–1151. [Google Scholar] [CrossRef] [Scilit]
  59. Sinayi, M.; Rasti-Barzoki, M. A game theoretic approach for pricing, greening, and social welfare policies in a supply chain with government intervention. J. Clean. Prod. 2018, 196, 1443–1458. [Google Scholar] [CrossRef] [Scilit]
  60. Wang, W.; Hao, S.; He, W.; Mohamed, M.A. Carbon emission reduction decisions in construction supply chain based on differential game with government subsidies. Build. Environ. 2022, 222, 109149. [Google Scholar] [CrossRef] [Scilit]
  61. Ghosh, D.; Shah, J. Supply chain analysis under green sensitive consumer demand and cost sharing contract. Int. J. Prod. Econ. 2015, 164, 319–329. [Google Scholar] [CrossRef] [Scilit]
  62. Liu, P.; Yi, S.P. Pricing policies of green supply chain considering targeted advertising and product green degree in the big data environment. J. Clean. Prod. 2017, 164, 1614–1622. [Google Scholar] [CrossRef] [Scilit]
  63. Guo, D.; He, Y.; Wu, Y.; Xu, Q. Analysis of Supply Chain under Different Subsidy Policies of the Government. Sustainability 2016, 8, 1290. [Google Scholar] [CrossRef] [Scilit]
  64. Ghosh, D.; Shah, J. A comparative analysis of greening policies across supply chain structures. Int. J. Prod. Econ. 2012, 135, 568–583. [Google Scholar] [CrossRef] [Scilit]
  65. Qin, J.; Fu, H.; Wang, Z.; Xia, L. Financing and carbon emission reduction strategies of capital-constrained manufacturers in E-commerce supply chains. Int. J. Prod. Econ. 2021, 241, 108271. [Google Scholar] [CrossRef] [Scilit]
  66. Zhu, W.; He, Y. Green product design in supply chains under competition. Eur. J. Oper. Res. 2017, 258, 165–180. [Google Scholar] [CrossRef] [Scilit]
  67. Li, P.; Rao, C.; Goh, M.; Yang, Z. Pricing strategies and profit coordination under a double echelon green supply chain. J. Clean. Prod. 2021, 278, 123694. [Google Scholar] [CrossRef] [Scilit]
  68. Moon, I.; Jeong, Y.J.; Saha, S. Investment and coordination decisions in a supply chain of fresh agricultural products. Oper. Res-Ger. 2020, 20, 2307–2331. [Google Scholar] [CrossRef] [Scilit]
  69. Zhang, C.; Lan, H.; Lyu, Z. Pricing and Coordination Decisions in Fresh Agricultural Product Supply Chain with Government Subsidy. J. Syst. Sci. Syst. Eng. 2025, 1–20. [Google Scholar] [CrossRef] [Scilit]
  70. Guo, D.; Guo, Y.; Jiang, K. Governance and effects of public R&D subsidies: Evidence from China. Technovation 2018, 74, 18–31. [Google Scholar] [CrossRef] [Scilit]
  71. Coulomb, R.; Henriet, F.; Reitzmann, L. “Bad” Oil, “Worse” Oil, and Carbon Misallocation. Rev. Econ. Stud. 2025, rdaf018. [Google Scholar] [CrossRef] [Scilit]
  72. Ji, J.; Zhang, Z.; Yang, L. Carbon emission reduction decisions in the retail-/dual-channel supply chain with consumers’ preference. J. Clean. Prod. 2017, 141, 852–867. [Google Scholar] [CrossRef] [Scilit]
  73. Chen, X.; Li, J.; Tang, D.; Zhao, Z.; Boamah, V. Stackelberg game analysis of government subsidy policy in green product market. Environ. Dev. Sustain. 2024, 26, 13273–13302. [Google Scholar] [CrossRef] [Scilit]
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.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.