A two-tier coordination mechanism for pollution control under water quality differences: a case study of the Yangtze-to-Hanjiang River Water Diversion Project Authors: Junhua Zhang, Yang Yang, Yaohong Yang, Jing Dai, Ran Jing Published: 2026 ABSTRACT The implementation of the Yangtze-to-Hanjiang River Water Diversion Project has caused the water resources to be replaced in the middle and lower reaches of the Han River, bringing about the problem of “engineering intervention-type” water pollution control. The central government, which leads the national water network construction, bears the responsibility for coordinating water pollution control and, as a beneficiary, shares a certain proportion of pollution control costs for the water supply area, polluting enterprises, and water diversion enterprises in water pollution control. Considering differences in water quality and the central government-led nature of the national water network, this study employs differential game theory, with the dynamic change of pollutant elimination as the state variable and the pollution control effort levels of the three main stakeholders as the control variables, constructs a differential game model involving the central government, the water supply area, polluting enterprises, and water diversion enterprises. Numerical simulation results show that, compared to decentralized decision-making, the two-tier cost-sharing mechanism can increase total pollutant elimination by approximately 45% and overall governance benefits by 30%, outperforming the centralized control mode. The main conclusions include: (1) There is an interactive relationship between the cost-sharing mechanisms of the central government and the coordination mechanism within the basin, and the two-tier cost-sharing coordination mechanism can bring the highest governance benefits and the most efficient pollution treatment effect. (2) When the loss coefficient increases, the efforts of the WSA and PE are improved under the three modes of Decision C, Decision S, and Decision SR. (3) When the amount of pollutants eliminated per unit of pollution control effort increases, the efforts of the WSA and PE show an increasing trend under the four different modes, and the total amount of pollutants eliminated in the basin also shows an increasing trend. This study provides a dynamic mechanism design framework for national water network governance and offers timely policy references for the governance of large-scale inter-basin water transfer systems. Introduction Globally, climate change and rapid economic development have significantly increased uncertainty in water security, posing severe challenges to regional ecological stability and sustainable development1. These challenges are particularly prominent in trans-boundary and engineered water systems, where climatic risks (e.g., extreme droughts, uneven precipitation) not only intensify water scarcity but also reshape the spatial distribution of pollutants2. Against this broader international context, China is facing similar pressures. The construction of a national water network is a major strategic plan for China. By establishing a water infrastructure network and enabling cross-basin water resource allocation, this project aims to address the uneven spatial and temporal distribution of the country’s water resources and enhance the ability to ensure national water security3. However, as a highly integrated socio-hydrological system, the national water network simultaneously alters hydrological connectivity, resource flows, and pollutant transport processes, thereby introducing new complexities beyond traditional water management frameworks. A representative example is the South-to-North Water Diversion Project (SNWDP) and its follow-up Yangtze-to-Hanjiang River Water Diversion Project (YHRWDP), which together form a typical engineered water network characterized by large-scale, long-distance, and policy-driven water transfers. While these projects significantly improve water quantity allocation and supply reliability, they also introduce water quality heterogeneity across connected river systems. In particular, the substitution of higher-quality water sources with relatively lower-quality transferred water may induce asymmetric redistribution of pollutants in downstream regions. This case points to a fundamental question: can engineered water networks simultaneously improve resource allocation efficiency and maintain balanced pollution distribution, or do they inherently generate new spatial inequalities in environmental burden? Due to differences in natural conditions, distribution of pollution sources, economic development, and utilization of water resources in different regions, water quality heterogeneity is inherently embedded in river basins. When such heterogeneous systems become increasingly interconnected through engineered water networks, pollutant transport is no longer confined within natural hydrological boundaries but instead follows reconfigured flow paths. As a result, water transfers may induce cross-regional redistribution of pollutants, potentially leading to localized deterioration of water quality and ecological conditions. Therefore, while improving water resource utilization efficiency, ensuring effective water pollution prevention and control and maintaining water quality are crucial challenges that must be addressed in the construction of the national water network. Building on the above mechanism, the key challenge lies in how to design effective governance strategies under conditions where water quality heterogeneity and engineered connectivity jointly reshape pollution distribution. In such contexts, traditional pollution control frameworks, which are typically based on naturally bounded river systems and single-level governance structures, may no longer be sufficient. Instead, pollution control in engineered water networks involves multiple actors operating at different administrative levels, with potentially misaligned incentives and responsibilities. This raises a critical research problem: how can coordination mechanisms be designed to align the interests of central authorities and basin-level stakeholders, thereby achieving efficient and equitable pollution control in the presence of water quality heterogeneity? The main contributions are as follows: First, it extends the analysis of water pollution control from naturally connected river systems to engineered water networks, where water transfers actively reshape hydrological and pollution linkages. Second, it conceptualizes water quality heterogeneity as a key mechanism driving pollution redistribution and governance challenges in cross-basin systems. Third, it develops a two-tier coordination framework that integrates central government incentives with intra-basin interactions, providing both theoretical insights and policy implications for pollution control in national water network construction. The remainder of this paper is organized as follows. Section “Literature review” reviews the literature on water pollution control, coordination mechanisms, and analytical approaches, with a particular focus on governance issues in engineered water networks. Section “Methology” presents the study area and develops the model framework, including key assumptions and the formulation of the two-tier coordination mechanism under different decision-making scenarios. Section “Numerical analysis” provides the numerical analysis, examining equilibrium outcomes, total benefits, and pollutant elimination under alternative mechanisms, as well as the effects of key parameters such as the loss coefficient and pollutant removal efficiency. Section “Discussion” discusses the underlying mechanisms and policy implications for cross-regional water pollution control in engineered water networks. Finally, Section “Conclusions” concludes the paper with the main findings and directions for future research. Literature review Coordination mechanism for water pollution control Water pollution control is a long-term and complex process, and cooperative control is an important environmental management measure. While conventional engineering measures such as source control, process governance, and end-of-pipe treatment remain the mainstream, nature-based solutions (NbS), such as constructed wetlands, have been increasingly explored for water pollution control and mitigation4,5. In addition, monitoring and prediction systems can effectively identify pollution sources, simulate pollutant transport processes, and provide technical support for collaborative governance6,7. However, the effectiveness of these approaches is often influenced by environmental variability and system complexity, which makes coordination among multiple stakeholders essential. Existing studies mainly focus on the cooperative management of different stakeholders8,9, which can be divided into two categories based on governance subjects and regional scope. The first category is intra-regional collaborative governance, where all stakeholders within the same region share responsibility for pollution control through coordination and cooperation. For instance, tripartite cooperation among local governments, enterprises, and farmers to control the rural water environment10, cooperation among central government, local government, and polluting enterprises to promote pollution control in the Yellow River basin11, and collaboration among local governments, enterprises, and the public to explore the interaction mechanisms between environmental regulation, public participation, and enterprise environmental behavior12,13. The second category is trans-boundary synergistic governance, in which multiple regions work together to address water pollution issues through cooperation and coordination. For example, cooperation among upstream governments, downstream governments and central governments14, cooperation between left-bank and right-bank regions belonging to different jurisdictions15, and cooperation among upstream and downstream countries of transnational rivers16. Water pollution control in basin usually requires significant capital investment, and the resulting revenues will be shared by all parties involved. Therefore, it is essential to form a unified governance concept17 and design an appropriate cooperation and coordination mechanism to balance the interests of all stakeholders. In the existing research on water pollution control and resource allocation, several key directions have laid the foundation for this study: First, regarding pollution load reallocation in river systems, some scholars have proposed a bankruptcy method to realize fair distribution of pollution loads among different regions, providing a new idea for resolving interest conflicts18. Second, for reservoir eutrophication in trans-boundary rivers, relevant studies have focused on exploring targeted control measures to mitigate water quality degradation caused by nutrient accumulation19. Third, in terms of reservoir eutrophication modeling and water quality improvement, researchers have established various simulation models to analyze nutrient cycling processes and proposed corresponding optimization strategies20–22. Fourth, for interactive reservoir-irrigation planning considering environmental issues, multi-objective optimization models combined with parallel processes techniques have been adopted to balance irrigation demands and ecological protection, improving the rationality of water resource allocation23. Fifth, aiming at waste load reallocation conflicts in river systems, conflict resolution methods have been developed to coordinate the interests of different stakeholders and promote cooperative governance24. However, these studies are primarily developed for natural river systems and do not adequately address pollution problems induced by large-scale water diversion projects in engineered water networks, which constitute the key focus of this study. In terms of pollution management mechanisms, current research results can be divided into four main categories. The first is the ecological compensation mechanism based on the internalization of the spatial spillover of public goods. It balances the economic burden among regions and promotes cooperation by compensating the governing parties whose interests are harmed. For example, Yi et al. analyzed trans-boundary pollution management in three regions in the upper, middle and lower reaches of a watershed, showing that ecological compensation effectively improves social welfare and promotes pollution reduction25. The second is the reward and punishment mechanism for intervention by environmental regulatory bodies. For example, Zhu et al. analyzed the interaction between upstream and downstream governments and central government intervention in a watershed, showing that central government regulation can accelerate upstream and downstream governments in reaching a stable state and improve cooperation efficiency26. Li et al. showed that in the ecological compensation in the basin, the optimal strategy of social expectations cannot be achieved only by the autonomous choice of upstream and downstream governments, and the incentive and constraint mechanism through central government intervention must be increased in order to maximize the benefits27. The third is market-based cost-sharing mechanisms. For example, Ding et al. analyzed the optimal ecological conservation efforts, cost-sharing rates, and optimal benefits of all participants under the three mechanisms of no-cost sharing, cost sharing, and collaborative cooperation among upstream and downstream governments and enterprises, and discussed the optimal decision-making of each party and its impact on their own revenues and the revenues of the whole system28. The fourth is the revenue-sharing mechanism, which incentives stakeholders to jointly participate in pollution management by rationally distributing the benefits brought by environmental management, and this incentive coordination mechanism can enhance the motivation for cooperation. For example, Sheng and Webber demonstrated that the revenue-sharing mechanism under the incentive coordination scenario can make the participants’ interests more compatible with the improvement of water quality29. These cooperative mechanisms are primarily single-level governance mechanisms and have limitations in addressing long-term and complex pollution management needs. For example, the ecological compensation mechanism can enhance the profit of the participants in the long-term, but the profit of the participants is rather lower than the level in the baseline scenario in the initial stage29. Overall, existing studies have provided important insights into stakeholder coordination, incentive design, and pollution control strategies. However, three key limitations remain. First, most studies focus on natural river basins and overlook the engineered characteristics of large-scale water diversion systems. Second, the majority of coordination mechanisms are designed within a single-level governance framework, lacking integration across hierarchical institutional structures. Third, current research rarely considers pollutant redistribution and governance complexity induced by “engineering intervention-type” water transfer projects. These limitations fundamentally constrain the explanatory power of existing frameworks when applied to engineered water network systems, particularly in capturing the complexity of large-scale water diversion contexts. Analysis method of coordination mechanism of water pollution control Pollution problems exhibit clear long-term and dynamic characteristics in practice, and it is often difficult to reach a specific equilibrium due to information asymmetry and the limited rationality of participants. As an efficient solution tool, differential games extend game theory into the time dimension, allowing for a comprehensive consideration of the impact of pollution accumulation on the regional environment, and are therefore commonly used to study pollution control problems. However, most studies focus on natural river basins rather than engineered water networks, lack integration of vertical governmental interactions, and fail to fully address pollutant redistribution induced by water diversion projects30,31. More importantly, existing modeling approaches rarely capture the coupling between hydrological alteration caused by engineering interventions and strategic interactions among multiple stakeholders, which significantly limits their ability to explain the coupled dynamics of hydrological alteration and strategic interactions in complex water network systems. This gap makes it difficult for existing models to provide effective support for “engineering intervention-type” water pollution control, which is the core motivation for this study. Existing studies can be broadly categorized into three strands. The first strand is to analyze based on a deterministic differential game model. For example, Liu et al. used a differential game model to explore the environmental payment for services (EPS) strategies in trans-boundary air pollution control, especially analyzing the optimal strategies under the three scenarios of no EPS, dynamic EPS, and fixed-fee EPS32. Jiang et al. constructed a differential game model to analyze the cross-border pollution control solutions between compensating and compensated regions by considering random disturbances33. The second strand extends these models by incorporating stochastic disturbances, learning effects, or dynamic adjustment processes to better reflect real-world uncertainties. Some studies have also combined differential games with other theories, such as “learning-by-doing” theory, optimal control theory and prospect theory, to enhance the practicality of the model and the effectiveness of the solutions. For example, Chen et al. discussed two ecological compensation mechanisms (proportional compensation and investment compensation) and dynamic optimal strategies under emission reduction policies based on the stochastic differential game of trans-boundary pollution reduction under the learning by doing mechanism34. The third strand integrates differential games with complementary approaches to capture behavioral heterogeneity and complex system interactions. Yuan et al. analyzed the trans jurisdictional river water pollution problem based on prospect theory, evolutionary game and system dynamics35. Lu et al. constructed a differential game model of trans-boundary river basin pollution among upstream, midstream and downstream based on sewage permit trading and learning-by-doing, and used optimal control theory to explore the optimal pollution discharge level and pollution reduction level under various regional joint strategies36. While these studies provide valuable insights into dynamic decision-making and policy design, they are primarily developed for naturally connected systems and do not explicitly consider the structural and hydrological changes induced by engineered water networks. As a result, they are insufficient to explain the coupled processes of pollutant redistribution and multi-level strategic interactions in “engineering intervention-type” water pollution control scenarios, which constitutes the central focus of this study. Pollution control issues from the perspective of water networks The water network consists of ‘natural + artificial’ lake and river systems within and across basins37. Current research focuses on water pollution control in natural basins, involving multi-scale and multi-faceted cooperation between upstream and downstream regions38, left and right banks15, main and tributary rivers, and across countries. For example, Yang et al. extended the collaborative governance relationship between upstream and downstream to the competitive relationship between two tributaries and the main river, which is more complex but more in line with the actual situation of the basin39,40. Based on the Belt and Road Initiative, Wang et al. integrated the ecological compensation mechanism into cross-border water resources management and explored the interaction mechanism between upstream and downstream countries and the regulatory committee jointly funded by the coastal countries41. From the perspective of water network systems, however, existing studies still largely focus on pollution problems in natural river network systems rather than engineered water networks42,43. For example, Wang et al. proposed a method to quantify pollution sources in river networks to improve the accuracy and pertinence of water pollution management plans44. Chen et al. analyzed the relationship between agricultural non-point source pollution in plain river networks and rural transformation and development45. Song et al. analyzed a comprehensive governance plan for mitigating pollutant emissions in coastal cities with complex river networks46. Lai et al. studied the impact of urban non-point source pollution on river network water environment quality47. However, these studies rarely consider the coupling between natural hydrological processes and engineered water diversion systems, and thus fail to adequately capture how such coupling reshapes pollutant transport pathways and governance structures. With the construction of the national water network, increasing attention has been paid to resolving conflicts of interest between source and receiving regions, and most researchers have focused on improving the water quality of water supply areas or solving the problem of unbalanced development between water supply areas and water receiving areas through cost-sharing, ecological compensation, and reward and punishment mechanisms48. For example, Li et al. explored the role of ecological compensation mechanism in coordinating pollution control behaviors of water supply cities and water receiving cities around the problem of cross-border water pollution control of the Heihe River water diversion project49. Zhou et al., taking the middle route of the SNWDP as an example, established a horizontal ecological compensation standard based on cost-sharing and revenue-sharing among the water source area, water receiving area, and downstream area. This approach effectively improved the development imbalance between the water source and water receiving areas and provided a feasible method for determining the project interest balance mechanism and calculating the compensation standard50. In the context of the SNWDP, existing studies have made important contributions in three main aspects. First, they have advanced ecological compensation frameworks to balance interests between source and receiving regions26,51,52. Second, they have explored coordination mechanisms, including cost-sharing arrangements and incentive-based strategies, to improve coordination efficiency and mitigate regional disparities53,54. Third, they have developed quantitative approaches for compensation standard setting, benefit assessment, and resource allocation optimization55–57. However, these studies are primarily based on static allocation perspectives and focus more on benefit distribution than on dynamic pollution interactions. They also tend to rely on single-level governance structures and do not fully capture the multi-level coordination characteristics of national water networks. More importantly, the issue of pollutant redistribution driven by water quality differences under large-scale water diversion remains insufficiently explored. With the implementation of a large number of water diversion projects, the national water network has been gradually improved, and the supply and demand relationship and interest conflicts have become increasingly complex. Regarding the “engineering intervention” water pollution control issues caused by water diversion projects, relevant research remains insufficient and requires further exploration. Aiming at the problem of water pollution prevention and control in the middle and lower reaches of the Han River caused by water resource displacement, this paper designs a two-tier collaborative mechanism that couples the central government cost-sharing mechanism with the intra-basin cost-sharing mechanism. It also compares and analyzes the governance effects of the water-supply area (WSA), polluting enterprises (PE) and water diversion enterprises (WDE) in the basin under different decision scenarios, including decentralized decision-making (N), centralized decision-making (C), cost-sharing mechanism (S) and cost-sharing & revenue-sharing mechanism (SR). Compared with existing research, the main differences are as follows: First, unlike existing studies that primarily focus on static allocation models or single-level governance structures, this paper develops a dynamic four-party differential game model to analyze pollutant redistribution induced by large-scale inter-basin water diversion projects under water quality heterogeneity. Second, this study proposes a novel two-tier coordination framework that integrates the central government’s cost-sharing mechanism with intra-basin coordination mechanisms, extending traditional basin governance models to a multi-level institutional structure consistent with the national water network system. Third, through equilibrium analysis and numerical simulations, this study quantitatively evaluates the governance performance of different coordination mechanisms and reveals their parameter sensitivity, thereby providing dynamic mechanism design insights for cross-basin pollution control. Methology Study area The SNWDP is the backbone (outline) of the national water network, initially constructing a water resources allocation pattern of “four horizontal and three vertical, north–south deployment, east–west mutual aid”, which is of great significance to further improve the main skeleton and artery of the national water network58. Among them, the first phase of the Central Route Project was commissioned in December 2014, and as of January 2024, the cumulative water transfer has exceeded 68 billion m3, which has strongly guaranteed the water security of the receiving area. However, the Danjiangkou Reservoir as a water source has the problems of uneven water quantity, large inter-annual variations, and an increase in the number of dry years and consecutive dry years, which have a greater impact on the amount of water transferred to the north. At the same time, the northern water-receiving areas have a more urgent demand for increased water transfers due to economic and social development. Therefore, to maximize the benefits of the SNWDP and improve the reliability of the project, the YHRWDP is implemented, as shown in Fig. 1, which will replenish water from the Three Gorges Reservoir to the middle and lower reaches of the Han River, reduce the amount of water discharged from the Danjiangkou Reservoir to the Han River, and replenished 2.49 billion m3 of water to the Middle Route of SNWDP through replacement, improving the capacity of stabilizing the supply of water to the receiving areas59.Fig. 1The geographical location of the Middle Route of SNWDP and YHRWDP. (This figure is derived from the National Platform for Common GeoSpatial In-formation Services). The geographical location of the Middle Route of SNWDP and YHRWDP. (This figure is derived from the National Platform for Common GeoSpatial In-formation Services). According to the ‘Monthly Report of Water Environment Quality in Danjiangkou City’, the water quality of Danjiangkou Reservoir remained at the Class II standard for an extended period from September 2022 to September 2023, indicating a relatively good water quality condition. In comparison, according to the data published by the Yangtze River Hydrology Bureau, the water quality of the Three Gorges Reservoir area of the Yangtze River mainly remained at the Class III standard. The water quality of Danjiangkou Reservoir is significantly better than that of the Three Gorges Reservoir. However, with the implementation of the YHRWDP, for the middle and lower reaches of the Han River, the natural high-quality water (Class II) from the Danjiang River is being replaced by Class III water transferred from the mainstream of the Yangtze River, which brings new challenges on the river basin management of the middle and lower reaches of the Han River. On the one hand, enterprises in the region that draw water from the river need to strengthen water treatment to meet the water quality requirements of users. On the other hand, with the increase in pollutant discharges brought about by economic development, enterprises discharging sewage are also under pressure to raise their wastewater treatment standards in order to comply with ecological and environmental protection requirements. These differences in water quality have brought a dual challenge to the region’s water pollution control efforts. Therefore, it is necessary to research and solve this “engineering intervention type” water pollution control problem, which can not only solve the water pollution prevention and control problems in the middle and lower reaches of the Han River, but also provide references for solving the pollution prevention and control problems caused by water quality differences in the process of national water network construction. Model framework and assumptions The central government, which leads the construction of the national water network, assumes the responsibility of coordinating water pollution control and, as a beneficiary, shares a certain proportion of pollution control costs for the WSA, PE, and WDE. For the parties within the basin, four decision scenarios of decentralized decision-making (N), centralized decision-making (C), cost-sharing mechanism (S) and cost-sharing & revenue-sharing (SR) are considered, and the equilibrium outcomes of the two-tier coordination mechanism resulting from the combination of the central government’s cost-sharing mechanism and different coordination mechanisms within the basin are analyzed. This framework is designed to reflect the governance structure and water allocation characteristics of the middle and lower reaches of the Han River under inter-basin water diversion, where strong interactions exist between water supply areas and receiving areas. The definition of stakeholders and cost-sharing arrangements is therefore aligned with the actual institutional and hydrological conditions of the study area. The analysis framework of the Two-Tier Coordination Mechanismis shown in Fig. 2.Fig. 2Analysis framework of the Two-Tier Coordination Mechanism. Analysis framework of the Two-Tier Coordination Mechanism. Assumption 1 Assumed that the cost of water quality protection of WSA is in the form of quadratic function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{{r_1}}}{2}{(S_{1}^{i})^2}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$i \in \left\{ {N,C,S,SR} \right\}$$\end{document}, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${r_1}$$\end{document} represents the cost coefficient of water quality protection. The effort degree of water quality protection of WSA is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{i}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$0 0$$\end{document} represents the initial welfare status of the river basin, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1} > 0$$\end{document} represents the impact coefficient of the pollution control work of WSA on social welfare, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2} > 0$$\end{document} represents the impact coefficient of the emission reduction work of PE on social welfare, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi > 0$$\end{document} represents the impact coefficient of pollutant removal in the watershed on social welfare. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${w_0}>0$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta _1}>0$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\beta _2}>0$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi >0$$\end{document} Assumption 8 The cost-sharing ratios provided by the central government to the WSA, PE, and WDE are \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi _{1}^{i}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi _{2}^{i}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi _{3}^{i}$$\end{document}, respectively, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$i \in \left\{ {N,C,S,SR} \right\}$$\end{document}. Assumption 9 The coefficients of the social welfare effect of the river basin on the income of WSA, PE, WDE, and the central government are \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _1}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _2}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _3}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _4}$$\end{document}, respectively, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _1}>0$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _2}>0$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _3}>0$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\pi _4}>0$$\end{document}. Assumption 10 Drawing on the supply chain, we consider that the market demand for water resources is affected by the upstream water quality and the emission reduction behavior of local enterprises. Assumed that the market demand for water resources is linearly related to the efforts of the water supply WSA \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{i}$$\end{document}, the efforts of the PE \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{2}^{i}$$\end{document}, and the efforts of the WDE \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${Z^i}$$\end{document}, ignoring the impact of water prices on water demand, the water demand can be expressed as:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$U=u+{c_1}S_{1}^{i}+{c_2}S_{2}^{i}+{c_3}{Z^i}+\sigma q{(t)_3}$$\end{document} Among them, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$u$$\end{document} is the fixed demand for water resources, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c_{1}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c_{2}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c_{3}$$\end{document} are the user’s sensitivity coefficients to the efforts of the WSA, PE, and WDE. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma$$\end{document} is the user’s experience of water quality, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$q(t)_{3}$$\end{document} is the pollutant treatment capacity of WDE, which can be expressed as:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop q\limits^{.} (t)_{3} = \eta_{3} Z^{i} - \theta q(t)_{3}$$\end{document} Among them, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{3}$$\end{document} is the pollution elimination of WDE paying unit pollution control efforts. Assumption 11 The participating entities have the same discount rate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho$$\end{document}, with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho >0$$\end{document}. Model formulation Decentralized decision-making(N) Based on the cost-sharing mechanism implemented by the central government, decentralized decision-making within the basin is considered. At this time, the members of the basin aim to maximize their own interests, forming a Nash game, and all entities in the basin are in an equal position. The decision-making order is: the central government decides the cost-sharing ratio for the WSA, PE, and WDE, and then the WSA, PE, and WDE decide their own optimal strategy. The revenue function of WSA is:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\int_{0}^{t} {[H_{0} } + m_{1} \eta_{1} S_{1}^{N} + \pi_{1} (w_{0} + \beta_{1} S_{1}^{N} + \beta_{2} S_{2}^{N} + \varphi q(t)) - (1 - \xi_{1}^{N} )\frac{1}{2}r_{1} (S_{1}^{N} )^{2} ]e^{ - \rho t} dt$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$H_{0}$$\end{document} is the cost compensation from the central government to WSA based on the amount of water transferred, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{1}$$\end{document} is the environmental benefit that WSA brings to the central government and itself by combating pollution, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{1} > 0$$\end{document}. The revenue function of PE is:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[J + kS_{2}^{N} + \pi_{2} } (w_{0} + \beta_{1} S_{1}^{N} + \beta_{2} S_{2}^{N} + \varphi q(t)) - (1 - \xi_{2}^{N} )\frac{1}{2}r_{2} (S_{2}^{N} )^{2} - (q_{E} - \eta_{2} S_{2}^{N} )\omega \hfill \\ + n_{1} \eta_{1} S_{1}^{N} - L(q_{s} - \eta_{1} S_{1}^{N} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$J$$\end{document} is the benefit gained by the production activities of PE. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k$$\end{document} represents the reputation benefit obtained through the emission reduction efforts of PE. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{1}$$\end{document} represents the benefit due to WSA’s pollution control to relieve pollution control pressure of PE \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{1} < r_{2}$$\end{document}. The revenue function of WDE is:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[P(u + c_{1} S_{1}^{N} + c_{2} S_{2}^{N} + c_{3} Z^{N} + \sigma q(t)_{3} } ) + \pi_{3} (w_{0} + \beta_{1} S_{1}^{N} + \beta_{2} S_{2}^{N} + \varphi q(t)) - \hfill \\ (1 - \xi_{3}^{N} )\frac{1}{2}r_{3} (Z^{N} )^{2} + n_{2} \eta_{1} S_{1}^{N} + n_{3} \eta_{2} S_{2}^{N} - L(q_{s} - \eta_{1} S_{1}^{N} - \eta_{2} S_{2}^{N} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$P$$\end{document} is the price of water resources, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{2}$$\end{document} represents the benefit due to WSA’s pollution control to relieve water treatment pressure of WDE, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{3}$$\end{document} represents the benefit due to PE efforts of reducing emissions to relieve water treatment pressure of WDE. The revenue function of the central government is:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[G - H_{0} + m_{1} } \eta_{1} S_{1}^{N} - m_{2} (q_{S} - \eta_{1} S_{1}^{N} ) + m_{3} \eta_{2} \left( {S_{2}^{N} - S_{0} } \right) + \pi_{4} (w_{0} + \beta_{1} S_{1}^{N} + \hfill \\ \beta_{2} S_{2}^{N} + \varphi q(t)) + m_{4} Z^{N} - \frac{1}{2}\xi_{1}^{N} r_{1} (S_{1}^{N} )^{2} - \frac{1}{2}\xi_{2}^{N} r_{2} (S_{2}^{N} )^{2} - \frac{1}{2}\xi_{3}^{N} r_{3} (Z^{N} )^{2} ]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G$$\end{document} is the benefit brought to the central government by the increased northward transfer of water from the SNWDP, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{1}$$\end{document} is the environmental benefit brought to the central government by WSA to control pollution, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{2}$$\end{document} is the environmental loss brought to the central government by the pollutants of the YHRWDP, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{3}$$\end{document} is the environmental benefit brought to the central government by the increase in emissions reductions of PE, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{4}$$\end{document} is the social benefit brought to central government of integrated water resources treatment of WDE, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{0}$$\end{document} is the initial emission reduction effort level of the PE when they are not affected by water quality differences. The optimal equilibrium solution is solved according to the inverse solution method, the results are as follows: The optimal effort level of the WSA of the YHRWDP is:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{N*} = \frac{{m_{1} \eta_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}{{(1 - \xi_{1}^{N*} )r_{1} }}$$\end{document} The optimal effort level of the PE is:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{2}^{N*} = \frac{{k + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}{{(1 - \xi_{2}^{N*} )r_{2} }}$$\end{document} The optimal effort level of the WDE is:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z^{N*} = \frac{{Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta {}_{3}}}{{(1 - \xi_{3}^{N*} )r_{3} }}$$\end{document} The optimal cost-sharing ratio of the central government for each party is12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{1}^{N*} = \frac{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) - m_{1} \eta_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) + m_{1} \eta_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document}13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{2}^{N*} = \frac{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) - k - \pi_{2} \beta_{2} - \eta_{2} \omega - \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) + k + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}$$\end{document}14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{3}^{N*} = \frac{{2m_{4} - Pc_{3} - \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}{{2m_{4} + Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}$$\end{document} Under this equilibrium condition, the optimal revenues of the WSA, PE, WDE, and the central government are respectively:15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{1}^{N*} = \frac{{\pi_{1} \varphi }}{\theta + \rho }q(t)^{N*} + \frac{1}{\rho }\left[ \begin{gathered} H_{0} + (m_{1} \eta_{1} + \pi_{1} \beta_{1} - (1 - \xi_{1}^{N*} )\frac{1}{2}r_{1} S_{1}^{N*} + \hfill \\ \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{N*} + (\pi_{1} \beta_{2} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{N*} + \pi_{1} w_{0} \hfill \\ \end{gathered} \right]$$\end{document}16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{2}^{N*} = \frac{{\pi_{2} \varphi }}{\theta + \rho }q(t)^{N*} + \frac{1}{\rho }\left[ \begin{gathered} J + (\pi_{2} \beta_{1} + n_{1} \eta_{1} + L\eta_{1} + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{N*} + (k + \hfill \\ \pi_{2} \beta_{2} - (1 - \xi_{2}^{N*} )\frac{1}{2}r_{2} S_{2}^{N*} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{N*} + \hfill \\ \pi_{2} w_{0} - q_{E} \omega - Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{3}^{N*} = \frac{{\pi_{3} \varphi }}{\theta + \rho }q(t)^{N*} + \frac{P\sigma }{{\theta + \rho }}q(t)_{3}^{N*} + \frac{1}{\rho }\left[ \begin{gathered} Pu + (Pc_{1} + \pi_{3} \beta_{1} + n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{N*} + \hfill \\ (Pc_{2} + \pi_{3} \beta_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{N*} + \hfill \\ (Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} - (1 - \xi_{3}^{N*} )\frac{1}{2}r_{3} Z^{N*} )Z^{N*} + \hfill \\ \pi_{3} w_{0} - Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{4}^{N*} = \frac{{\pi_{4} \varphi }}{\theta + \rho }q(t)^{N*} + \frac{1}{\rho }\left[ \begin{gathered} G - H_{0} + (m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} - \frac{1}{2}\xi_{1}^{N*} r_{1} S_{1}^{N*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{N*} + (m_{3} \eta_{2} + \pi_{4} \beta_{2} - \frac{1}{2}\xi_{2}^{N*} r_{2} S_{2}^{N*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{N*} + (m_{4} - \frac{1}{2}\xi_{3}^{N*} r_{3} Z^{N*} )Z^{N*} - m_{2} q_{S} - \hfill \\ m_{3} \eta_{2} S_{0} + \pi_{4} w_{0} \hfill \\ \end{gathered} \right]$$\end{document} Centralized decision-making (C) Assuming that under the constraints of the central government, all entities in the basin have established a long-term cooperative relationship, meaning all entities in the basin act as a coordinated and controlled unit. In this case, WSA, PE, and WDE make decisions with the goal of maximizing the overall economic growth of the basin. The game sequence involves the central government first determining the optimal cost-sharing ratio, after which basin members maximize the overall profit based on the central government’s cost-sharing ratio. The decision-making objective functions for the basin as a whole and the central government are as follows: The overall revenue function of the basin is:19\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[H_{0} } + m_{1} \eta_{1} S_{1}^{C} + (\pi_{1} + \pi_{2} + \pi_{3} )(w_{0} + \beta_{1} S_{1}^{C} + \beta_{2} S_{2}^{C} + \varphi q(t)) - \hfill \\ (1 - \xi_{1}^{C} )\frac{1}{2}r_{1} (S_{1}^{C} )^{2} + J + kS_{2}^{C} + - (1 - \xi_{2}^{C} )\frac{1}{2}r_{2} (S_{2}^{C} )^{2} - (q_{E} - \eta_{2} S_{2}^{C} )\omega + \hfill \\ n_{1} \eta_{1} S_{1}^{C} - L(q_{s} - \eta_{1} S_{1}^{C} ) + P(u + c_{1} S_{1}^{C} + c_{2} S_{2}^{C} + c_{3} Z^{C} + \sigma q_{3} ) - \hfill \\ (1 - \xi_{3}^{C} )\frac{1}{2}r_{3} (Z^{C} )^{2} + n_{2} \eta_{1} S_{1}^{C} + n_{3} \eta_{2} S_{2}^{C} - L(q_{s} - \eta_{1} S_{1}^{C} - \eta_{2} S_{2}^{C} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document} The revenue function of the central government is:20\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[G - H_{0} + m_{1} } \eta_{1} S_{1}^{C} - m_{2} (q_{S} - \eta_{1} S_{1}^{C} ) + m_{3} \eta_{2} \left( {S_{2}^{C} - S_{0} } \right) + \pi_{4} (w_{0} + \beta_{1} S_{1}^{C} + \hfill \\ \beta_{2} S_{2}^{C} + \varphi q(t)) + m_{4} Z^{C} - \frac{1}{2}\xi_{1}^{C} r_{1} (S_{1}^{C} )^{2} - \frac{1}{2}\xi_{2}^{C} r_{2} (S_{2}^{C} )^{2} - \frac{1}{2}\xi_{3}^{C} r_{3} (Z^{C} )^{2} ]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document} The optimal equilibrium solution is solved according to the inverse solution method, the results as follows: The optimal effort level of the WSA of the YHRWDP is:21\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{C*} = \frac{{m_{1} \eta_{1} + (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{1} + n_{1} \eta_{1} + Pc_{1} + n_{2} \eta_{1} + 2L\eta_{1} + \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{1} }}{{(1 - \xi_{1}^{C*} )r_{1} }}$$\end{document} The optimal effort level of the PE is:22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{2}^{C*} = \frac{{k + (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{2} + \eta_{2} \omega + Pc_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{(\pi_{1} + \pi_{2} + \pi_{2} )\varphi }}{\theta + \rho }\eta_{2} }}{{(1 - \xi_{2}^{C*} )r_{2} }}$$\end{document} The optimal effort level of the WDE is:23\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z^{C*} = \frac{{Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}{{(1 - \xi_{3}^{C*} )r_{3} }}$$\end{document} The optimal cost-sharing ratio of the central government for each party is24\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{1}^{C*} = \frac{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) - m_{1} \eta_{1} - (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{1} - n_{1} \eta_{1} - Pc_{1} - n_{2} \eta_{1} - 2L\eta_{1} - \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{1} }}{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) + m_{1} \eta_{1} + (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{1} + n_{1} \eta_{1} + Pc_{1} + n_{2} \eta_{1} + 2L\eta_{1} + \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{1} }}$$\end{document}25\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{2}^{C*} = \frac{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) - k - (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{2} - \eta_{2} \omega - Pc_{2} - n_{3} \eta_{2} - L\eta_{2} - \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{2} }}{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) + k + (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{2} + \eta_{2} \omega + Pc_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{2} }}$$\end{document}26\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{3}^{C*} = \frac{{2m_{4} - Pc_{3} - \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}{{2m_{4} + Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}$$\end{document} The optimal returns for the basin as a whole and the central government are:27\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{1}^{C*} = \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }q(t)^{C*} + \frac{P\sigma }{{\theta + \rho }}q(t)_{3}^{C*} + \frac{1}{\rho }\left[ \begin{gathered} H_{0} + (m_{1} \eta_{1} + (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{1} - (1 - \xi_{1}^{C*} )\frac{1}{2}r_{1} S_{1}^{C*} + n_{1} \eta_{1} + \hfill \\ L\eta_{1} + Pc_{1} + n_{2} \eta_{1} + L\eta_{1} + \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{1} )S_{1}^{C*} + \hfill \\ (k + (\pi_{1} + \pi_{2} + \pi_{3} )\beta_{2} - (1 - \xi_{2}^{C*} )\frac{1}{2}r_{2} S_{2}^{C*} + \eta_{2} \omega + Pc_{2} + \hfill \\ n_{3} \eta_{2} + L\eta_{2} + \frac{{(\pi_{1} + \pi_{2} + \pi_{3} )\varphi }}{\theta + \rho }\eta_{2} )S_{2}^{C*} + (Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} - \hfill \\ (1 - \xi_{3}^{C*} )\frac{1}{2}r_{3} Z^{C*} )Z^{C*} + (\pi_{1} + \pi_{2} + \pi_{3} )w_{0} + J - q_{E} \omega + \hfill \\ Pu - 2Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}28\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{2}^{C*} = \frac{{\pi_{4} \varphi }}{\theta + \rho }q(t)^{C*} + \frac{1}{\rho }\left[ \begin{gathered} G - H_{0} + (m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} - \frac{1}{2}\xi_{1}^{C*} r_{1} S_{1}^{C*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{C*} + (m_{3} \eta_{2} + \pi_{4} \beta_{2} - \frac{1}{2}\xi_{2}^{C*} r_{2} S_{2}^{C*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{C*} + (m_{4} - \frac{1}{2}\xi_{3}^{C*} r_{3} Z^{C*} )Z^{C*} - \hfill \\ m_{2} q_{S} - m_{3} \eta_{2} S_{0} + \pi_{4} w_{0} \hfill \\ \end{gathered} \right]$$\end{document} Cost-sharing mechanism (S) Based on the cost-sharing mechanism of the central government, the intra-basin cost-sharing mechanism is also considered. At this time, the PE and WDE in the middle and lower reaches of the Han River, as the leading parties, share a portion of the cost for the WSA to increase their enthusiasm for pollution control. The WDE also share a portion of the emission reduction costs for the PE to encourage them to reduce emissions. At this stage, the various entities in the basin form a Stackelberg leader–follower game model. The game sequence is as follows: The central government first decides the cost-sharing ratio for each entity in the basin, then the WDE determine their optimal effort level and the cost-sharing ratio they provide to the WSA and PE. Next, the PE determine their optimal effort level and the cost-sharing ratio which they provide to the WSA based on the cost-sharing ratios from the central government and the WDE. Finally, the WSA determine their optimal effort level based on the cost-sharing ratios from the central government, PE, and WDE. The revenue function of WSA is:29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\int_{0}^{t} {[H_{0} } + m_{1} \eta_{1} S_{1}^{S} + \pi_{1} (w_{0} + \beta_{1} S_{1}^{S} + \beta_{2} S_{2}^{S} + \varphi q(t)) - (1 - \xi_{1}^{S} - \xi_{5}^{S} - \xi_{6}^{S} )\frac{1}{2}r_{1} (S_{1}^{S} )^{2} ]e^{ - \rho t} dt$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{5}$$\end{document} is the cost-sharing ratio provided by the WDE for the WSA, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{6}$$\end{document} is the cost-sharing ratio provided by the PE for the WSA. The revenue function of PE is:30\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[J + kS_{2}^{S} + \pi_{2} } (w_{0} + \beta_{1} S_{1}^{S} + \beta_{2} S_{2}^{S} + \varphi q(t)) - (1 - \xi_{2}^{S} - \xi_{4}^{S} )\frac{1}{2}r_{2} (S_{2}^{S} )^{2} - \hfill \\ \frac{1}{2}\xi_{6}^{S} r_{1} (S_{1}^{S} )^{2} - (q_{E} - \eta_{2} S_{2}^{S} )\omega + n_{1} \eta_{1} S_{1}^{S} - L(q_{s} - \eta_{1} S_{1}^{S} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{4}$$\end{document} is the cost-sharing ratio provided by the WDE to the PE. The revenue function of WDE is:31\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[P(u + c_{1} S_{1}^{S} + c_{2} S_{2}^{S} + c_{3} Z^{S} + \sigma q(t)_{3} } ) + \pi_{3} (w_{0} + \beta_{1} S_{1}^{S} + \beta_{2} S_{2}^{S} + \varphi q(t)) - \hfill \\ (1 - \xi_{3}^{S} )\frac{1}{2}r_{3} (Z^{S} )^{2} - \frac{1}{2}\xi_{4}^{S} r_{2} (S_{2}^{S} )^{2} - \frac{1}{2}\xi_{5}^{S} r_{1} (S_{1}^{S} )^{2} + n_{2} \eta_{1} S_{1}^{S} + n_{3} \eta_{2} S_{2}^{S} - \hfill \\ L(q_{s} - \eta_{1} S_{1}^{S} - \eta_{2} S_{2}^{S} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document} The revenue function of the central government is:32\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[G - H_{0} + m_{1} } \eta_{1} S_{1}^{S} - m_{2} (q_{S} - \eta_{1} S_{1}^{S} ) + m_{3} \eta_{2} \left( {S_{2}^{S} - S_{0} } \right) + \pi_{4} (w_{0} + \beta_{1} S_{1}^{S} + \hfill \\ \beta_{2} S_{2}^{S} + \varphi q(t)) + m_{4} Z^{S} - \frac{1}{2}\xi_{1}^{S} r_{1} (S_{1}^{S} )^{2} - \frac{1}{2}\xi_{2}^{S} r_{2} (S_{2}^{S} )^{2} - \frac{1}{2}\xi_{3}^{S} r_{3} (Z^{S} )^{2} ]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document} The optimal equilibrium solution is solved according to the inverse solution method, the results as follows: The optimal effort level of the WSA of the YHRWDP is:33\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{S*} = \frac{{m_{1} \eta_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}{{(1 - \xi_{1}^{S*} - \xi_{5}^{S*} - \xi_{6}^{S*} )r_{1} }}$$\end{document} The optimal effort level of the PE is:34\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{2}^{S*} = \frac{{k + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}{{(1 - \xi_{2}^{S*} - \xi_{4}^{S*} )r_{2} }}$$\end{document} The optimal effort level of the WDE is:35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z^{S*} = \frac{{Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta {}_{3}}}{{(1 - \xi_{3}^{S*} )r_{3} }}$$\end{document} The optimal cost-sharing ratio of the central government for each party is36\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{1}^{S*} = \frac{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) - m_{1} \eta_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) + m_{1} \eta_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document}37\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{2}^{S*} = \frac{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) - k - \pi_{2} \beta_{2} - \eta_{2} \omega - \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) + k + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}$$\end{document}38\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{3}^{S*} = \frac{{2m_{4} - Pc_{3} - \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}{{2m_{4} + Pc_{3} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} }}$$\end{document} The optimal cost-sharing ratio provided by WDE for PE is39\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{4}^{S*} = \frac{{(1 - \xi_{2}^{S*} )[2(Pc_{2} + \pi_{3} \beta_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} ) - k - \pi_{2} \beta_{2} - \eta_{2} \omega - \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} ]}}{{2(Pc_{2} + \pi_{3} \beta_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} ) + k + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}$$\end{document} The optimal cost-sharing ratio provided by WDE for WSA is40\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{5}^{S*} = \frac{{(1 - \xi_{1}^{S*} )[2(Pc_{1} + \pi_{3} \beta_{1} + n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} ) - m_{1} \eta_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} ]}}{{2(Pc_{1} + \pi_{3} \beta_{1} + n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} ) + m_{1} \eta_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document} The optimal cost sharing ratio provided by PE for WSA is41\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{6}^{S*} = \frac{{(1 - \xi_{1}^{S*} - \xi_{5}^{S*} )[2(\pi_{2} \beta_{1} + n_{1} \eta_{1} + L\eta_{1} + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{1} ) - m_{1} \eta_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} ]}}{{2(\pi_{2} \beta_{1} + n_{1} \eta_{1} + L\eta_{1} + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{1} ) + m_{1} \eta_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document} The optimal returns of WSA, PE, WDE and the central government are respectively:42\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{1}^{S*} = \frac{{\pi_{1} \varphi }}{\theta + \rho }q(t)^{S*} + \frac{1}{\rho }\left[ \begin{gathered} H_{0} + (m_{1} \eta_{1} + \pi_{1} \beta_{1} - (1 - \xi_{1}^{S*} - \xi_{5}^{S*} - \xi_{6}^{S*} )\frac{1}{2}r_{1} S_{1}^{S*} + \hfill \\ \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{S*} + (\pi_{1} \beta_{2} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{S*} + \pi_{1} w_{0} \hfill \\ \end{gathered} \right]$$\end{document}43\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{2}^{S*} = \frac{{\pi_{2} \varphi }}{\theta + \rho }q(t)^{S*} + \frac{1}{\rho }\left[ \begin{gathered} J + (\pi_{2} \beta_{1} - \frac{1}{2}\xi_{6}^{S*} r_{1} S_{1}^{S*} + n_{1} \eta_{1} + L\eta_{1} + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{S*} + \hfill \\ (k + \pi_{2} \beta_{2} - (1 - \xi_{2}^{S*} - \xi_{4}^{S*} )\frac{1}{2}r_{2} S_{2}^{S*} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{S*} + \hfill \\ \pi_{2} w_{0} - q_{E} \omega - Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}44\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{3}^{S*} = \frac{{\pi_{3} \varphi }}{\theta + \rho }q(t)^{S*} + \frac{P\sigma }{{\theta + \rho }}q(t)_{3}^{S*} + \frac{1}{\rho }\left[ \begin{gathered} Pu + (Pc_{1} + \pi_{3} \beta_{1} - \frac{1}{2}\xi_{5}^{S*} r_{1} S_{1}^{S*} + n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{S*} + \hfill \\ (Pc_{2} + \pi_{3} \beta_{2} - \frac{1}{2}\xi_{4}^{S*} r_{2} S_{2}^{S*} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{S*} + \hfill \\ (Pc_{3} - (1 - \xi_{3}^{S*} )\frac{1}{2}r_{3} Z^{S*} + \frac{P\sigma }{{\theta + \rho }}\eta_{3} )Z^{S*} + \pi_{3} w_{0} - Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{4}^{S*} = \frac{{\pi_{4} \varphi }}{\theta + \rho }q(t)^{S*} + \frac{1}{\rho }\left[ \begin{gathered} G - H_{0} + (m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} - \frac{1}{2}\xi_{1}^{S*} r_{1} S_{1}^{S*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{S*} + (m_{3} \eta_{2} + \pi_{4} \beta_{2} - \frac{1}{2}\xi_{2}^{S*} r_{2} S_{2}^{S*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{S*} + (m_{4} - \frac{1}{2}\xi_{3}^{S*} r_{3} Z^{S*} )Z^{S*} - \hfill \\ m_{2} q_{S} - m_{3} \eta_{2} S_{0} + \pi_{4} w_{0} \hfill \\ \end{gathered} \right]$$\end{document} Cost-sharing & revenue-sharing(SR) Based on the cost-sharing mechanism within the river basin, the WDE will share their revenue to the WSA and PE in a certain proportion to encourage them to increase their efforts in pollution control and emission reduction since the demand for water resources is related to the efforts of the WSA and PE. The game order at this time is consistent with that when the cost-sharing mechanism is adopted in Sect. 3.4. The revenue function of WSA is:46\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[H_{0} } + m_{1} \eta_{1} S_{1}^{SR} + \lambda_{1} P(u + c_{1} S_{1}^{SR} + c_{2} S_{2}^{SR} + c_{3} Z^{SR} + \sigma q(t)_{3} ) + \hfill \\ \pi_{1} (w_{0} + \beta_{1} S_{1}^{SR} + \beta_{2} S_{2}^{SR} + \varphi q(t)) - (1 - \xi_{1}^{SR} - \xi_{5}^{SR} - \xi_{6}^{SR} )\frac{1}{2}r_{1} (S_{1}^{SR} )^{2} ]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda_{1}$$\end{document} is the revenue-sharing ratio provided by the WDE to the WSA. The revenue function of PE is:47\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[J + kS_{2}^{SR} + \lambda_{2} P(u + c_{1} S_{1}^{SR} + c_{2} S_{2}^{SR} + c_{3} Z^{SR} + \sigma q(t)_{3} ) + \pi_{2} } (w_{0} + \hfill \\ \beta_{1} S_{1}^{SR} + \beta_{2} S_{2}^{SR} + \varphi q(t)) - (1 - \xi_{2}^{SR} - \xi_{4}^{SR} )\frac{1}{2}r_{2} (S_{2}^{SR} )^{2} - \frac{1}{2}\xi_{6}^{SR} r_{1} (S_{1}^{SR} )^{2} - \hfill \\ (q_{E} - \eta_{2} S_{2}^{SR} )\omega + n_{1} \eta_{1} S_{1}^{SR} - L(q_{s} - \eta_{1} S_{1}^{SR} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda_{2}$$\end{document} is the revenue-sharing ratio provided by the WDE to the PE. The revenue function of WDE is:48\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[P(1 - \lambda_{1} - \lambda_{2} )(u + c_{1} S_{1}^{SR} + c_{2} S_{2}^{SR} + c_{3} Z^{SR} + \sigma q(t)_{3} )} + \pi_{3} (w_{0} + \beta_{1} S_{1}^{SR} + \hfill \\ \beta_{2} S_{2}^{SR} + \varphi q(t)) - (1 - \xi_{3}^{SR} )\frac{1}{2}r_{3} (Z^{SR} )^{2} - \frac{1}{2}\xi_{4}^{SR} r_{2} (S_{2}^{SR} )^{2} - \frac{1}{2}\xi_{5}^{SR} r_{1} (S_{1}^{SR} )^{2} + \hfill \\ n_{2} \eta_{1} S_{1}^{SR} + n_{3} \eta_{2} S_{2}^{SR} - L(q_{s} - \eta_{1} S_{1}^{SR} - \eta_{2} S_{2}^{SR} )]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document} The revenue function of the central government is:49\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} \int_{0}^{t} {[G - H_{0} + m_{1} } \eta_{1} S_{1}^{SR} - m_{2} (q_{S} - \eta_{1} S_{1}^{SR} ) + m_{3} \eta_{2} \left( {S_{2}^{SR} - S_{0} } \right) + \hfill \\ \pi_{4} (w_{0} + \beta_{1} S_{1}^{SR} + \beta_{2} S_{2}^{SR} + \varphi q(t)) + m_{4} Z^{SR} - \frac{1}{2}\xi_{1}^{SR} r_{1} (S_{1}^{SR} )^{2} - \hfill \\ \frac{1}{2}\xi_{2}^{SR} r_{2} (S_{2}^{SR} )^{2} - \frac{1}{2}\xi_{3}^{SR} r_{3} (Z^{SR} )^{2} ]e^{ - \rho t} dt \hfill \\ \end{gathered}$$\end{document} The optimal equilibrium solution is solved according to the inverse solution method, the results as follows: The optimal effort level of the WSA is:50\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{SR*} = \frac{{m_{1} \eta_{1} + \lambda_{1} Pc_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}{{(1 - \xi_{1}^{SR*} - \xi_{5}^{SR*} - \xi_{6}^{SR*} )r_{1} }}$$\end{document} The optimal effort level of the PE is:51\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{2}^{SR*} = \frac{{k + \lambda_{2} Pc_{2} + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}{{(1 - \xi_{2}^{SR*} - \xi_{4}^{SR*} )r_{2} }}$$\end{document} The optimal effort level of the WDE is:52\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z^{SR*} = \frac{{(1 - \lambda_{1} - \lambda_{2} )Pc_{3} + \frac{{(1 - \lambda_{1} - \lambda_{2} )P\sigma }}{\theta + \rho }\eta {}_{3}}}{{(1 - \xi_{3}^{SR*} )r_{3} }}$$\end{document} The optimal cost-sharing ratio of the central government for each party is53\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{1}^{SR*} = \frac{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) - m_{1} \eta_{1} - \lambda_{1} Pc_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}{{2(m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} ) + m_{1} \eta_{1} + \lambda_{1} Pc_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document}54\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{2}^{SR*} = \frac{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) - k - \lambda_{2} Pc_{2} - \pi_{2} \beta_{2} - \eta_{2} \omega - \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}{{2(m_{3} \eta_{2} + \pi_{4} \beta_{2} + \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} ) + k + \lambda_{2} Pc_{2} + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}$$\end{document}55\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{3}^{SR*} = \frac{{2m_{4} - (1 - \lambda_{1} - \lambda_{2} )Pc_{3} - \frac{{(1 - \lambda_{1} - \lambda_{2} )P\sigma }}{\theta + \rho }\eta_{3} }}{{2m_{4} + (1 - \lambda_{1} - \lambda_{2} )Pc_{3} + \frac{{(1 - \lambda_{1} - \lambda_{2} )P\sigma }}{\theta + \rho }\eta_{3} }}$$\end{document} The optimal cost-sharing ratio provided by WDE for PE is56\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{4}^{SR*} = \frac{{(1 - \xi_{2}^{SR*} )\left\{ {2\left[ {P(1 - \lambda_{1} - \lambda_{2} )c_{2} + \pi_{3} \beta_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} } \right] - k - \lambda_{2} Pc_{2} - \pi_{2} \beta_{2} - \eta_{2} \omega - \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} } \right\}}}{{2\left[ {P(1 - \lambda_{1} - \lambda_{2} )c_{2} + \pi_{3} \beta_{2} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} } \right] + k + \lambda_{2} Pc_{2} + \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} }}$$\end{document} The optimal cost-sharing ratio provided by WDE for WSA is57\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{5}^{SR*} = \frac{{(1 - \xi_{1}^{SR*} )\left\{ {2\left[ {P(1 - \lambda_{1} - \lambda_{2} )c_{1} + \pi_{3} \beta_{1} + n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} } \right] - m_{1} \eta_{1} - \lambda_{1} Pc_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} } \right\}}}{{2\left[ {P(1 - \lambda_{1} - \lambda_{2} )c_{1} + \pi_{3} \beta_{1} + n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} } \right] + m_{1} \eta_{1} + \lambda_{1} Pc_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document} The optimal cost-sharing ratio provided by PE for WSA is58\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{6}^{SR*} = \frac{{\left( {1 - \xi_{1}^{SR*} - \xi_{5}^{SR*} } \right)\left[ {2\left( {\lambda_{2} Pc_{1} + \pi_{2} \beta_{1} + n_{1} \eta_{1} + L\eta_{1} + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta } \right)_{1} - m_{1} \eta_{1} - \lambda_{1} Pc_{1} - \pi_{1} \beta_{1} - \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} } \right]}}{{2\left( {\lambda_{2} Pc_{1} + \pi_{2} \beta_{1} + n_{1} \eta_{1} + L\eta_{1} + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{1} } \right) + m_{1} \eta_{1} + \lambda_{1} Pc_{1} + \pi_{1} \beta_{1} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} }}$$\end{document} The optimal returns of WSA, PE, WDE and the central government are respectively:59\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{1}^{SR*} = \frac{{\pi_{1} \varphi }}{\theta + \rho }q(t)^{SR*} + \frac{{\lambda_{1} P\sigma }}{\theta + \rho }q(t)_{3}^{SR*} + \frac{1}{\rho }\left[ \begin{gathered} H_{0} + (m_{1} \eta_{1} + \lambda_{1} Pc_{1} + \pi_{1} \beta_{1} - (1 - \xi_{1}^{SR*} - \xi_{5}^{SR*} - \xi_{6}^{SR*} )\frac{1}{2}r_{1} S_{1}^{SR*} \hfill \\ + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{SR*} + (\lambda_{1} Pc_{2} + \pi_{1} \beta_{2} + \frac{{\pi_{1} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{SR*} + \hfill \\ (\lambda_{1} Pc_{3} + \frac{{\lambda_{1} P\sigma }}{\theta + \rho }\eta_{3} )Z^{SR*} + \lambda_{1} Pu + \pi_{1} w_{0} \hfill \\ \end{gathered} \right]$$\end{document}60\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{2}^{SR*} = \frac{{\pi_{2} \varphi }}{\theta + \rho }q(t)^{SR*} + \frac{{\lambda_{2} P\sigma }}{\theta + \rho }q(t)_{3}^{SR*} + \frac{1}{\rho }\left[ \begin{gathered} J + (\lambda_{2} Pc_{1} + \pi_{2} \beta_{1} - \frac{1}{2}\xi_{6}^{SR*} r_{1} S_{1}^{SR*} + n_{1} \eta_{1} + L\eta_{1} + \hfill \\ \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{SR*} + (k + \lambda_{2} Pc_{2} - (1 - \xi_{2}^{SR*} - \xi_{4}^{SR*} )\frac{1}{2}r_{2} S_{2}^{SR*} + \hfill \\ \pi_{2} \beta_{2} + \eta_{2} \omega + \frac{{\pi_{2} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{SR*} + (\lambda_{2} Pc_{3} + \frac{{\lambda_{2} P\sigma }}{\theta + \rho }\eta_{3} )Z^{SR*} \hfill \\ + \lambda_{2} Pu + \pi_{2} w_{0} - q_{E} \omega - Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}61\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{3}^{SR*} = \frac{{\pi_{3} \varphi }}{\theta + \rho }q(t)^{SR*} + \frac{{(1 - \lambda_{1} - \lambda_{2} )P\sigma }}{\theta + \rho }q(t)_{3}^{SR*} + \frac{1}{\rho }\left[ \begin{gathered} (1 - \lambda_{1} - \lambda_{2} )Pu + (P(1 - \lambda_{1} - \lambda_{2} )c_{1} + \pi_{3} \beta_{1} - \frac{1}{2}\xi_{5}^{SR*} r_{1} S_{1}^{SR*} + \hfill \\ n_{2} \eta_{1} + L\eta_{1} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{SR*} + (P(1 - \lambda_{1} - \lambda_{2} )c_{2} + \pi_{3} \beta_{2} - \hfill \\ \frac{1}{2}\xi_{4}^{SR*} r_{2} S_{2}^{SR*} + n_{3} \eta_{2} + L\eta_{2} + \frac{{\pi_{3} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{SR*} + (P(1 - \lambda_{1} - \lambda_{2} )c_{3} - \hfill \\ (1 - \xi_{3}^{SR*} )\frac{1}{2}r_{3} Z^{SR*} + \frac{{(1 - \lambda_{1} - \lambda_{2} )P\sigma }}{\theta + \rho }\eta_{3} )Z^{SR*} + \pi_{3} w_{0} - Lq_{s} \hfill \\ \end{gathered} \right]$$\end{document}62\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_{4}^{SR*} = \frac{{\pi_{4} \varphi }}{\theta + \rho }q(t)^{SR*} + \frac{1}{\rho }\left[ \begin{gathered} G - H_{0} + (m_{1} \eta_{1} + m_{2} \eta_{1} + \pi_{4} \beta_{1} - \frac{1}{2}\xi_{1}^{SR*} r_{1} S_{1}^{SR*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{1} )S_{1}^{SR*} + (m_{3} \eta_{2} + \pi_{4} \beta_{2} - \frac{1}{2}\xi_{2}^{SR*} r_{2} S_{2}^{SR*} + \hfill \\ \frac{{\pi_{4} \varphi }}{\theta + \rho }\eta_{2} )S_{2}^{SR*} + (m_{4} - \frac{1}{2}\xi_{3}^{SR*} r_{3} Z^{SR*} )Z^{SR*} - \hfill \\ m_{2} q_{S} - m_{3} \eta_{2} S_{0} + \pi_{4} w_{0} \hfill \\ \end{gathered} \right]$$\end{document} Comparison of discussion under the four decision scenarios The following is a comparative analysis of the equilibrium results. Based on the game equilibrium solutions of the four mechanisms mentioned above, we obtain: Corollary 1: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{1}^{N*} = \xi_{1}^{S*} > \xi_{1}^{SR*} > \xi_{1}^{C*}$$\end{document}; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{2}^{N*} = \xi_{2}^{S*} > \xi_{2}^{SR*} > \xi_{2}^{C*}$$\end{document}; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\xi_{3}^{SR*} > \xi_{3}^{N*} = \xi_{3}^{C*} = \xi_{3}^{S*}$$\end{document}. Corollary 1 shows that the cost-sharing ratio of the central government for the three parties in the basin is consistent when making Decision N and Decision S. The reason is that, when the basin chooses Decision N, each subject aims to maximize its own interests, so the central government will provide a higher cost-sharing ratio to encourage these subjects to actively participate in pollution control. When the cost-sharing mechanism is introduced in the basin, this mechanism provides incentives by reducing the costs of WSA and PE, rather than directly driving them to spontaneously increase their efforts in the form of benefits. In this case, the central government can still maintain a good balance between costs and benefits on the basis of the original cost-sharing ratio, so as to ensure that all parties can effectively participate in governance and achieve overall optimization of pollution control. For the WSA and PE, the cost-sharing ratio provided by the central government is greater under Decision N and Decision S than under Decision SR, and is the lowest under Decision C. The reason is that, under Decision C, they are more motivated to increase their efforts when entities within the basin consider the overall interests of the basin. Therefore, even if the central government reduces the cost-sharing ratio, it can still achieve better outcomes. Similarly, in the case of Decision SR, the WDE provides a portion of revenue-sharing for the WSA and PE on the basis of cost-sharing. This encourages the WSA and PE to increase their efforts to obtain higher benefits, resulting in a lower optimal cost-sharing ratio from the central government compared to Decision N and Decision S. For WDE, under the scenarios of Decisions N, Decisions S, and Decisions C, it fails to consider the impact of their governance efforts on WSA and PE, so they are relatively less affected by these decisions, and the central government’s cost-sharing ratio for WDE remains unchanged. However, WDE will allocate part of their revenue to WSA and PE after the introduction of the revenue-sharing mechanism, resulting in a decrease in their governance efforts. Then, the central government will correspondingly increase the cost-sharing ratio for the WDE to encourage them to enhance their water pollution control efforts and improve the overall effectiveness of governance. Corollary 2: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{1}^{S*} > S_{1}^{N*}$$\end{document}; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{2}^{S*} > S_{2}^{N*}$$\end{document};\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z^{N*} = Z^{C*} = Z^{S*}$$\end{document}. Corollary 2 shows that, under Decision S, the efforts of WSA and PE are greater than those under Decision N. The reason is that the cost-sharing ratio provided by the central government for the three parties in the basin remains unchanged under the both decision scenarios. However, under Decision S, a cost-sharing mechanism is added within the basin, resulting in more effective coordination and stimulating greater participation and efforts from all parties. The effort degree of WDE in Decision N, Decision C, and Decision S remains unchanged because their pollutant elimination directly impacts users’ experience but fails to significantly affect the basin’s overall water quality. Although the impact of pollution control efforts and the amount of pollutant removal on water demand is considered, the broader effects on the WSA, PE, and overall social welfare are not. This results in a relatively small impact on the WDE under these three decision scenarios, leaving their effort level unchanged. In subsequent numerical simulations, the focus will be on analyzing the WSA and PE. Numerical analysis To analyze the decision-making behavior of multiple stakeholders, numerical simulations were performed. The 2022 data were used since the 2023 data have not yet been released. In addition, some parameters refer to the data in 2021 as the 2022 data lack information on water production and supply. According to the Statistical Yearbook of the cities in the middle and lower reaches of the Han River, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$J$$\end{document} takes the sum of the total industrial production value of these cities in 2021, which is 1831.79277 billion yuan, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$q_{E}$$\end{document} takes the industrial wastewater discharge in the middle and lower reaches of the Han River in 2021, which is 136.913793 tons. According to the monitoring reports on water quality of Danjiangkou Reservoir and Three Gorges Reservoir, it was calculated that \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$q_{s}$$\end{document} was 38,017 tons. According to the “2022 Statistical Yearbook of Yichang City”, based on the proportion of the total industrial output value of Yichang City and the middle and lower reaches of the Han River to the total industrial output value of the two places, as well as the difference in output value of industrial enterprises in these two regions, the value \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{1} = 0.2$$\end{document} is assigned. And thus we assign \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{2} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n_{3} = 0.8$$\end{document} on the basis of the difference between the industrial output value of the middle and lower reaches of the Han River area after removing the water production and supply industry and the total industrial output value of the city of Yichang. According to the Law of the People’s Republic of China on Environmental Protection Tax, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega$$\end{document} is set to 1.4 yuan/m3. According to the “2022 Water Resources Bulletin of Hubei Province”, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$u$$\end{document} is taken as 9.216 billion m3 of water consumption in the middle and lower reaches of the Han River in 2021. According to the Water Fee Standard of Hubei Province, considering the proportion of water consumption for different purposes, the water resource price \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$P$$\end{document} is 1.81 yuan/m3. The initial social welfare of the basin is taken as the sum of the gross industrial production of both the water supply area and the receiving area. According to calculation, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$w_{0}$$\end{document} = 2314.03935 billion yuan. According to the 2022 “Soil and Water Conservation Bulletin of Yichang City”, the unit investment cost of maintaining water volume is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c$$\end{document} = 0.25264789 yuan/m3. Therefore, the cost compensation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$H_{0}$$\end{document} based on water volume is calculated to be 629 million yuan. Referring to the study of Sheng and Webber29, take \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\theta = 0.2$$\end{document},\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho = 0.2$$\end{document}. Referring to Chen et al.60, take \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c_{1} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c_{2} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c_{3} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma = 0.6$$\end{document}. Referring to Yang et al.39, take \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1} = 0.4$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2} = 0.5$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{3} = 0.5$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{1} = 0.9$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{2} = 0.9$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{3} = 0.9$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$m_{4} = 0.9$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S_{0} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k = 0.2$$\end{document}. Referring to Ding et al.28, take \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{1} = 0.5$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta_{2} = 0.5$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi = 0.5$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{1} = 3$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{2} = 4$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r_{3} = 5$$\end{document}. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pi_{1} = 0.4$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pi_{2} = 0.3$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pi_{3} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pi_{4} = 0.9$$\end{document}. Referring to Lai et al.61, take \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda_{1} = 0.2$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda_{2} = 0.2$$\end{document}. Game equilibrium solutions under different decision scenarios According to the above values, the game equilibrium solutions under different decision scenarios are shown in Fig. 3.Fig. 3Comparison of equilibrium solutions under different decision scenarios. Comparison of equilibrium solutions under different decision scenarios. Figure 3 shows that the efforts of the WSA and the PE are the lowest under Decision N. Specifically, the effort level of WSA (S₁) is 0.5906 under Decision N, which increases to 0.8907 under Decision S, representing an increase of approximately 50.8%. Similarly, the effort level of PE (S₂) rises from 0.3580 to 0.5039, corresponding to an increase of 40.8%. Under Decision S, the efforts of the WSA and the PE increase significantly, indicating that the cost-sharing mechanism can effectively motivate WSA and PE to participate in pollution control. However, when a revenue-sharing mechanism is further introduced on the basis of cost-sharing, the level of effort actually decreases. For example, under Decision SR, the effort level of WSA is 0.7844, which is 11.9% lower than that under Decision S, while the effort of PE decreases from 0.5039 to 0.4462, a reduction of 11.5%. In contrast, the effect in Decision C is not as good as a pure cost-sharing mechanism. Although Decision C improves the effort levels to 0.7562 (WSA) and 0.4942 (PE), these values remain 15.1% and 1.9% lower than those under Decision S, respectively. The central government’s proportion of cost-sharing among the three parties is higher when Decision N and Decision S are adopted, while it is lower under other decision-making mechanisms. This suggests an interaction between the central government’s cost-sharing strategy and the governance mechanisms within the basin. When making Decision N, since the three parties in the basin aim to maximize their own interests, the central government encourages all parties to increase their efforts by providing a higher cost-sharing ratio. However, as each entity acts independently, although this provides operational flexibility, it often becomes difficult to achieve effective pollution control due to dispersed resources and inconsistent goals. Under Decision S, a cost-sharing mechanism is introduced to provide incentives by reducing costs, without directly increasing the benefits of the WSA and PE. The implementation of the basin cost-sharing mechanism does not affect the implementation of the central government cost-sharing mechanism. The central government can still maintain a good balance between costs and benefits based on the original cost-sharing ratio. This two-tier cost-sharing coordination mechanism can promote the efforts of WSA and PE in the basin by reasonably sharing the governance costs, and enhance the sustainability of the mechanism and the enthusiasm of participants. Under Decision C, the WSA and PE consider the overall benefits of the entire basin and proactively increase their efforts. This spontaneous improvement allows the central government to reduce its cost-sharing ratio while still achieving good governance results. However, due to the reduced cost-sharing ratio from the central government, the effect is not as strong as the two-tier cost-sharing coordination mechanism. Under Decision SR, a revenue-sharing mechanism is introduced in addition to the basin cost-sharing mechanism. The WDE distributes part of its revenue to the WSA and PE. This direct economic connection encourages the WSA and PE to increase their investments to boost revenue, while the WDE reduces its own efforts due to revenue-sharing. In order to maintain governance efficiency and incentive balance, the central government will adjust the cost-sharing ratio, reduce the cost-sharing ratio for WSA and PE, and increase the cost-sharing ratio for WDE to adapt to this new revenue distribution model. The addition of revenue-sharing creates a secondary layer of economic linkage that reinforces, but does not surpass, the core cost-sharing incentives, leading to a moderate reduction in overall effort levels compared to Decision S. To summarize the four decision scenarios: the central government’s cost-sharing proportion is higher under Decision N and Decision S, and lower under Decision C and Decision SR. This difference reveals a significant interaction between the central government’s cost-sharing strategy and the intra-basin governance mechanisms. Decision S consistently achieves the highest effort levels across all stakeholders, exceeding Decision N by 40%–50% and outperforming Decision SR and Decision C by approximately 10%–20%. Under decentralized governance, higher fiscal input is required to stabilize governance effort; whereas under coordinated governance mechanisms, improved internal incentives reduce the need for excessive central intervention. The two-tier cost-sharing mechanism (Decision S) is the most effective model for aligning stakeholder efforts, balancing central and local incentives, and minimizing the need for excessive fiscal intervention. Total benefits and pollutant elimination under different decisions Based on the above values, the total benefits and total pollutant elimination in the basin under the four decision scenarios are analyzed. The results are shown in Figs. 4 and 5.Fig. 4Total benefits in the basin under different decisions.Fig. 5Total pollutant elimination in the basin under different decisions. Total benefits in the basin under different decisions. Total pollutant elimination in the basin under different decisions. As can be seen from Figs. 4 and 5, it reveals that, under Decision S, the basin achieves the highest governance benefits and the most efficient pollution treatment effects. Specifically, the total benefits under Decision S converge to approximately 2.0500 × 105 (100 million yuan), which is higher than those under Decision SR (approximately 2.04998 × 105), Decision C (approximately 2.04997 × 105), and Decision N (approximately 2.04995 × 105). In relative terms, Decision S improves total benefits by about 0.002%–0.003% compared with the other scenarios. Although the absolute difference appears small, it reflects a consistent and stable advantage under equilibrium conditions. Similarly, for pollutant elimination, Decision S reaches approximately 3.20 (10 thousand tons), which exceeds Decision SR (about 2.85), Decision C (about 2.88), and Decision N (about 2.20). This corresponds to an increase of approximately 45.5% compared to Decision N, and about 11%–13% compared to Decision C and Decision SR. In theory, it should be able to promote collaboration among participants, optimize resource allocation, and thus improve governance effectiveness when a revenue-sharing contract is added on this basis or Decision C is adopted. In reality, due to the participation of the central government, the central government’s adjustment of the cost-sharing ratio for the three entities in the basin has caused these strategies to fail to achieve the expected governance benefits, and are not as effective as a separate cost-sharing mechanism. From a dynamic perspective, all four scenarios exhibit a rapid increase in total benefits and pollutant elimination in the early stage (t = 0–10), followed by gradual convergence after t ≈ 15, indicating that the system reaches a stable equilibrium. However, the convergence level under Decision S is consistently higher than that of the other scenarios throughout the entire time horizon, demonstrating not only a higher equilibrium outcome but also a faster convergence speed. This shows that the cost-sharing mechanism adopted by the central government has an interactive relationship with the mechanism in the basin. This two-tier cost-sharing coordination mechanism complicates decision-making and produces different effects from a single-level mechanism. This finding demonstrates that the interaction between central fiscal policy and local coordination mechanisms is a key determinant of success, and that a well-designed two-tier cost-sharing system can outperform both centralized and hybrid models. In addition, due to the lack of unified coordination and goal consistency when making Decision N, each participant tends to act according to their own interests, resulting in inefficient governance measures. Quantitatively, the total benefits under Decision N stabilize at approximately 2.04995 × 105, which is about 0.002% lower than that under Decision S, while pollutant elimination remains at only 2.20, significantly lower than the other scenarios. As shown in both figures, the curves for Decision N are consistently the lowest, with total benefits and pollutant elimination stabilizing at levels far below the other three scenarios. Although this model provides a high degree of decision-making flexibility, it is the lowest among several models in terms of pollutant governance effects, and it is difficult to effectively control pollution. This aligns with the “tragedy of the commons” theory, where individual rationality leads to collective inefficiency, and further validates the necessity of coordinated governance mechanisms in water resource management. Effect of loss coefficient The influence of loss coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on effort level Set \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L \in$$\end{document}[0,0.3], we analyze the impact of loss coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the efforts of WSA and PE. The results are shown in Figs. 6 and 7.Fig. 6Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the effort level of WSA.Fig. 7Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the effort level of PE. Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the effort level of WSA. Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the effort level of PE. Figures 6 and 7 show that, when the unit pollutant loss coefficient increases, the efforts of the WSA and the PE under the three scenarios of Decision C, Decision S, and Decision SR all increase, and remain unchanged under Decision N. Specifically, when the loss coefficient increases from 0 to 0.3, the effort level of WSA under Decision S rises from approximately 0.73 to 0.97, corresponding to an increase of 32.9%, while under Decision SR it increases from 0.65 to 0.8, and under Decision C only from 0.72 to 0.75. In contrast, under Decision N, the effort level remains constant at approximately 0.59, indicating zero responsiveness. This trend reflects that the parties pay more attention to pollution control under the three scenarios of Decision C, Decision S, and Decision SR when the loss coefficient increases. This indicates that these coordinated governance mechanisms enable stakeholders to dynamically adjust their strategies in response to rising environmental risks, whereas the decentralized Decision N lacks such adaptive capacity. When making Decision C, the WSA and PE’s effort levels rise moderately with L, For example, the effort level of PE increases from 0.485 to 0.498, representing a change of only 2.7%, indicating relatively low sensitivity. As shown by the blue dashed lines in both figures, the WSA and PE will take more active measures to control pollution in order to safeguard the overall interests of the river basin and reduce the negative impact caused by differences in water quality. This reflects that centralized decision-making aligns individual incentives with basin-wide goals, prompting proactive adaptation to increasing losses. When making Decision S and Decision SR, the effort levels under Decision S rise most sharply, outpacing both Decision SR and Decision C, while Decision SR shows a moderate increase, PE and WDE are more proactive in increasing the cost-sharing ratio, providing economic support to WSA to encourage them to increase their efforts in water pollution control to reduce the losses caused by water quality differences. The WDE also increases their cost-sharing for the PE, encouraging them to intensify emission reduction efforts. And when the revenue-sharing mechanism exists, the direct connection of economic interests prompts WSA and PE to pay more attention to the revenue of WDE, thereby increasing their own efforts and reducing the negative impact of water quality differences on WDE. This highlights that the two-tier cost-sharing mechanism creates stronger incentives for adaptation, as the combined central and basin-level cost-sharing arrangements directly link effort levels to risk mitigation, while the addition of revenue-sharing introduces a secondary layer of economic linkage that reinforces, but does not surpass, the core cost-sharing incentives. When making Decision N, the amount of pollutant elimination remains unchanged even if the loss coefficient increases. The effort levels of both WSA and PE remain fixed at approximately 0.59 and 0.36, respectively, regardless of changes in L, indicating a complete lack of sensitivity. It means that the response capacity within the basin is limited when the external environment changes, and it is impossible to cope with the challenges brought by the changes by adjusting strategies. This validates the “tragedy of the commons” dynamic in decentralized governance, where individual self-interest prevents adaptive responses to rising environmental risks, leading to persistent underinvestment in pollution control. The two-tier cost-sharing mechanism demonstrates the strongest ability to align stakeholder efforts with increasing environmental risks, outperforming centralized, hybrid, and decentralized models. This underscores that well-designed multi-level coordination is critical for building resilient water pollution governance systems. The influence of loss coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the total pollutant elimination Set \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L \in$$\end{document}[0,0.3], at t = 1, we analyze the impact of loss coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the total pollutant elimination in the basin. The results are shown in Fig. 8.Fig. 8Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the total pollutant elimination in the basin. Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L$$\end{document} on the total pollutant elimination in the basin. As can be seen from Fig. 8, the total amount of pollutant elimination in the basin has shown significant improvement under the three modes of Decision C, Decision S, and Decision SR. Specifically, when the loss coefficient increases from 0 to 0.3, pollutant elimination under Decision S increases from approximately 0.50 to 0.63, representing an increase of about 26.0%. Under Decision SR, it rises from about 0.45 to 0.55, corresponding to an increase of 22.2%, while under Decision C, the increase is much smaller, from 0.50 to 0.52, representing only a 4.0% improvement. This improvement stems from the ability of these coordinated governance models to align stakeholder incentives with environmental risk mitigation. As rising loss coefficients create stronger pressure to reduce pollution-related damages, these mechanisms effectively translate this pressure into enhanced governance efforts. The reason is that these models can effectively improve the governance efforts of all parties through enhanced coordination and resource allocation, thereby achieving more efficient removal of pollutants. However, the amount of pollutant elimination remains unchanged under the scenarios of Deciding N. The characteristic of decentralized decision-making is that all parties make independent decisions and lack coordination. As a result, treatment strategies and effects are not adjusted accordingly when external factors such as loss coefficients change, preventing further improvement in pollutant elimination. Impact of pollutant elimination per unit of effort Effect on effort Taking \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1} \in$$\end{document}[0.1,0.6], we analyze the impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1}$$\end{document} on the pollution control efforts of WSA, and the results are shown in Fig. 9. Taking \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2} \in$$\end{document}[0.1,0.6], we analyze the impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2}$$\end{document} on the pollution control efforts of the PE, and the results are shown in Fig. 10.Fig. 9Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1}$$\end{document} on the effort level of WSA.Fig. 10Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2}$$\end{document} on the effort level of PE. Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1}$$\end{document} on the effort level of WSA. Impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2}$$\end{document} on the effort level of PE. As shown in Figs. 9 and 10, the efforts of the WSA and PE show an increasing trend in the four different modes when the amount of pollutant elimination per unit of pollution control effort increases. Specifically, when η₁ increases from 0.1 to 0.6, the effort level of WSA under Decision S rises from approximately 0.55 to 0.98, representing an increase of about 78.2%. Under Decision SR, it increases from 0.41 to 0.88, while under Decision C, it rises from 0.32 to 0.87. In contrast, under Decision N, the increase is from 0.23 to 0.68, corresponding to an increase of 195.7%. Similarly, for PE, when η₂ increases from 0.1 to 0.6, the effort level under Decision S increases from approximately 0.29 to 0.65, while under Decision C it rises from 0.27 to 0.65, under Decision SR from 0.24 to 0.59, and under Decision N from 0.18 to 0.48. The reason is that, when the amount of pollutant elimination per unit of pollution control effort increases, it means that the same investment can achieve a higher pollution treatment effect. Such efficiency improvements usually reduce the marginal cost of treating pollutants, making it easier for WSA and PE to implement more active pollution control measures. This positive relationship between efficiency and effort underscores a key insight: as pollution control becomes more cost-effective, stakeholders are incentivized to increase their investment, as the marginal benefit of each additional unit of effort rises. This creates a virtuous cycle of improved efficiency and enhanced governance. The influence on the total pollutant elimination Taking \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1} \in$$\end{document}[0.1,0.6], \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2} \in$$\end{document}[0.1,0.6], we analyze the impact of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2}$$\end{document} on the total pollutant elimination in the basin. The results are shown in Figs. 11 and 12.Fig. 11Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1}$$\end{document} on the total pollutant elimination in the basin.Fig. 12Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2}$$\end{document} on the total pollutant elimination in the basin. Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{1}$$\end{document} on the total pollutant elimination in the basin. Effect of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta_{2}$$\end{document} on the total pollutant elimination in the basin. Figures 11 and 12 show that the total pollutant elimination in the basin increases under the four scenarios of Decision N, Decision C, Decision S, and Decision SR as the amount of pollutant elimination per unit of pollution control effort increases. Specifically, when η₁ increases from 0.1 to 0.6, pollutant elimination under Decision S rises from approximately 0.23 to 0.72 (10 thousand tons), representing an increase of about 213.0%. Under Decision C, it increases from 0.21 to 0.65 , while under Decision SR, it rises from 0.20 to 0.64. In contrast, under Decision N, the increase is from 0.15 to 0.50, corresponding to an increase of 233.3%, but the absolute level remains significantly lower than that of the coordinated scenarios. Similarly, when η₂ increases from 0.1 to 0.6, pollutant elimination under Decision S increases from approximately 0.43 to 0.75, while under Decision C it rises from 0.37 to 0.6, under Decision SR from 0.38 to 0.6, and under Decision N from 0.29 to 0.53. This indicates that, regardless of the governance model adopted, improvements in the efficiency of pollution control efforts lead to better pollutant elimination effects, ultimately increasing the total amount of pollutant elimination in the basin. In terms of the degree of increase, Decision S shows the largest improvement, while Decision N shows the smallest. As the marginal productivity of pollution control efforts rises, stakeholders are incentivized to scale up their investments, leading to a collective improvement in environmental outcomes. The well-designed multi-level coordination is critical for maximizing the returns on pollution control investments, outperforming both centralized and decentralized alternatives in adapting to and capitalizing on efficiency gains. Discussion Mechanism interpretation The main conclusions are as follows:The two-tier cost-sharing coordination mechanism, namely the model of Decision S, is the best mechanism for “engineering intervention-type” water pollution control. In the basin, the level of effort from all parties is lowest under Decision N. When adopting Decision S, the effort levels of all parties significantly increase after introducing a cost-sharing mechanism within the basin, indicating that the cost-sharing mechanism effectively incentives participation in pollution control. However, the effort levels actually decrease when a revenue-sharing mechanism is further introduced on the basis of cost-sharing. The outcome of Decision C is also less effective compared to the pure cost-sharing mechanism. The introduction of a cost-sharing mechanism within the basin, which constitutes a two-tier cost-sharing coordination mechanism, can bring the highest governance benefits and the most efficient pollution treatment effect.There is an interactive relationship between the central government’s cost-sharing mechanism and the coordination mechanism within the basin. The two-tier coordination mechanism complicates decision-making and produces different effects compared to a single-tier mechanism. The two-tier mechanism promotes the joint and active efforts of all participants to achieve best pollution control results.Under the scenarios of Decision C, Decision S, and Decision SR, the attention paid by all parties to pollution control rises when the loss coefficient increases, leading to increased total pollutant elimination in the basin and greater efforts by the WSA and PE. However, under Decision N, the efforts of all parties and the total pollutant elimination remain unchanged even if the loss coefficient increases. It means that the response capacity within the basin is limited when the external environment changes, and it is impossible to adjust strategies to cope with the challenges brought by the changes. There is a lack of effective response mechanisms.When the amount of pollutant elimination per unit of pollution control effort increases, the same investment can achieve better pollution control results. Such efficiency improvements typically reduce the marginal cost of treating pollutants, enabling the WSA and PE to implement more proactive pollution control measures. Under the four scenarios, the efforts of the WSA and PE increase, and the total amount of pollutant elimination in the basin also shows an upward trend. This pattern suggests the presence of implicit thresholds in the coordination mechanism, beyond which changes in external conditions (e.g., loss coefficients or technological efficiency) lead to noticeable shifts in stakeholder behavior and overall system performance. The two-tier cost-sharing coordination mechanism, namely the model of Decision S, is the best mechanism for “engineering intervention-type” water pollution control. In the basin, the level of effort from all parties is lowest under Decision N. When adopting Decision S, the effort levels of all parties significantly increase after introducing a cost-sharing mechanism within the basin, indicating that the cost-sharing mechanism effectively incentives participation in pollution control. However, the effort levels actually decrease when a revenue-sharing mechanism is further introduced on the basis of cost-sharing. The outcome of Decision C is also less effective compared to the pure cost-sharing mechanism. The introduction of a cost-sharing mechanism within the basin, which constitutes a two-tier cost-sharing coordination mechanism, can bring the highest governance benefits and the most efficient pollution treatment effect. The two-tier cost-sharing coordination mechanism, namely the model of Decision S, is the best mechanism for “engineering intervention-type” water pollution control. In the basin, the level of effort from all parties is lowest under Decision N. When adopting Decision S, the effort levels of all parties significantly increase after introducing a cost-sharing mechanism within the basin, indicating that the cost-sharing mechanism effectively incentives participation in pollution control. However, the effort levels actually decrease when a revenue-sharing mechanism is further introduced on the basis of cost-sharing. The outcome of Decision C is also less effective compared to the pure cost-sharing mechanism. The introduction of a cost-sharing mechanism within the basin, which constitutes a two-tier cost-sharing coordination mechanism, can bring the highest governance benefits and the most efficient pollution treatment effect. There is an interactive relationship between the central government’s cost-sharing mechanism and the coordination mechanism within the basin. The two-tier coordination mechanism complicates decision-making and produces different effects compared to a single-tier mechanism. The two-tier mechanism promotes the joint and active efforts of all participants to achieve best pollution control results. There is an interactive relationship between the central government’s cost-sharing mechanism and the coordination mechanism within the basin. The two-tier coordination mechanism complicates decision-making and produces different effects compared to a single-tier mechanism. The two-tier mechanism promotes the joint and active efforts of all participants to achieve best pollution control results. Under the scenarios of Decision C, Decision S, and Decision SR, the attention paid by all parties to pollution control rises when the loss coefficient increases, leading to increased total pollutant elimination in the basin and greater efforts by the WSA and PE. However, under Decision N, the efforts of all parties and the total pollutant elimination remain unchanged even if the loss coefficient increases. It means that the response capacity within the basin is limited when the external environment changes, and it is impossible to adjust strategies to cope with the challenges brought by the changes. There is a lack of effective response mechanisms. Under the scenarios of Decision C, Decision S, and Decision SR, the attention paid by all parties to pollution control rises when the loss coefficient increases, leading to increased total pollutant elimination in the basin and greater efforts by the WSA and PE. However, under Decision N, the efforts of all parties and the total pollutant elimination remain unchanged even if the loss coefficient increases. It means that the response capacity within the basin is limited when the external environment changes, and it is impossible to adjust strategies to cope with the challenges brought by the changes. There is a lack of effective response mechanisms. When the amount of pollutant elimination per unit of pollution control effort increases, the same investment can achieve better pollution control results. Such efficiency improvements typically reduce the marginal cost of treating pollutants, enabling the WSA and PE to implement more proactive pollution control measures. Under the four scenarios, the efforts of the WSA and PE increase, and the total amount of pollutant elimination in the basin also shows an upward trend. This pattern suggests the presence of implicit thresholds in the coordination mechanism, beyond which changes in external conditions (e.g., loss coefficients or technological efficiency) lead to noticeable shifts in stakeholder behavior and overall system performance. When the amount of pollutant elimination per unit of pollution control effort increases, the same investment can achieve better pollution control results. Such efficiency improvements typically reduce the marginal cost of treating pollutants, enabling the WSA and PE to implement more proactive pollution control measures. Under the four scenarios, the efforts of the WSA and PE increase, and the total amount of pollutant elimination in the basin also shows an upward trend. This pattern suggests the presence of implicit thresholds in the coordination mechanism, beyond which changes in external conditions (e.g., loss coefficients or technological efficiency) lead to noticeable shifts in stakeholder behavior and overall system performance. The findings are broadly consistent with existing studies emphasizing the importance of coordination mechanisms in water pollution control, where cooperative arrangements and cost-sharing schemes are shown to improve environmental outcomes. However, most previous research has focused on single-level governance structures or static policy settings. In contrast, this study highlights the role of multi-level coordination under engineering-driven water transfer conditions, where the interaction between central intervention and intra-basin coordination significantly influences stakeholder behavior. By incorporating dynamic strategic interactions within a differential game framework, the results provide a more integrated perspective and extend the current literature on water pollution governance. Policy implications Building on the above results, this section discusses their practical implications rather than restating the main findings. In large cross-basin water diversion systems such as the South-to-North Water Diversion Project, water quality disparities can generate externalities that are difficult to fully address under decentralized governance alone. The two-tier cost-sharing coordination mechanism achieves higher pollution control efficiency than purely decentralized or centralized structures. From a policy perspective, the central government should act as both financial supporter and institutional coordinator. The optimal cost-sharing ratio should be dynamically adjusted based on the basin governance structure. Central intervention can be moderately reduced under cost-sharing mechanisms, while stronger fiscal incentives are needed under decentralized decision-making. Sensitivity analysis shows that coordinated governance becomes more effective as water quality-related risks increase, suggesting the need for stronger regulation and targeted subsidies in high-risk contexts. Improving pollutant elimination efficiency via technological innovation enhances all governance modes, thereby supporting sustained investment in pollution control and digital monitoring systems. These results also point to how the proposed mechanism may be implemented in practice. The two-tier coordination framework provides an analytical basis for adaptive multi-level governance in engineering-related watershed pollution control, facilitating the sustainable implementation of national water network strategies. For this framework to operate effectively, it needs to be embedded in a more explicit operational process. The central government should take the lead in establishing a formal two-tier coordination platform that integrates basin authorities, provincial governments, and major stakeholders, which helps reduce fragmentation in decision-making and improves coordination across regions. Within this institutional framework, cost-sharing ratios between central and local governments, as well as among upstream and downstream regions, should be negotiated and determined based on pollution responsibilities, water diversion intensity, ecological risk levels, and fiscal capacity. The cost-sharing scheme should remain adaptive rather than fixed, allowing for adjustments in response to changing hydrological and socio-economic conditions. Unified accounting and verification standards are also necessary. All participating parties should agree on expenditure boundaries, accounting methods, cost calculation procedures, water quality indicators, and data reporting requirements. Standardized monitoring protocols and third-party verification mechanisms should be introduced to ensure transparency and comparability of data, while also limiting disputes and lowering coordination costs. A phased implementation and settlement mechanism can further support this process. Pollution control expenditures and water quality performance should be assessed periodically, such as on a quarterly or annual basis, with financial transfers or compensation payments adjusted according to verified outcomes. This type of iterative evaluation allows policy parameters, including cost-sharing ratios and subsidy intensity, to evolve in line with performance and environmental conditions. Several observations from the results are worth noting for policy design. A pure cost-sharing approach tends to perform better than mixed mechanisms in engineering-driven pollution control contexts, since the introduction of revenue-sharing elements may weaken incentives. The interaction between central intervention and basin-level coordination is not substitutive but complementary, and multi-level governance can significantly improve overall performance. Without an effective coordination mechanism, the system shows limited responsiveness to external environmental changes, which underscores the importance of maintaining institutional flexibility under uncertainty. Improvements in pollutant elimination efficiency also play a central role, as they enhance governance outcomes across all decision scenarios. By combining institutional coordination, standardized accounting, and periodic performance-based settlement, the two-tier cost-sharing mechanism can function as a stable and operational governance tool rather than a purely theoretical arrangement. This practice-oriented framework provides a replicable management model for large-scale cross-basin water pollution control and supports the long-term, stable implementation of the national water network strategies. In practice, this implies that pollution control efforts should place greater emphasis on upstream regions, where water quality risks are more likely to propagate through the transfer system. Targeted ecological compensation and differentiated cost-sharing arrangements can help address regional disparities in both pollution responsibility and fiscal capacity, while strengthened monitoring and verification systems—particularly those supported by digital technologies—can improve transparency and enforcement. These considerations illustrate how the proposed coordination mechanism can be translated into concrete actions for water quality enhancement and pollution mitigation. Conclusions Based on the strategic background of national water network construction, large-scale inter-basin water transfer projects fundamentally reshape hydrological structures and spatial pollution distribution patterns, giving rise to “engineering intervention-type” water pollution problems. Existing studies on pollution allocation and basin governance mainly focus on natural river systems or single-tier coordination frameworks, with limited attention to pollution redistribution induced by engineered water transfer systems and vertical governmental interactions. In contrast, this study constructs a two-tier collaborative governance mechanism integrating central government macro-regulation with intra-basin coordination under heterogeneous water quality conditions. By embedding cost-sharing and coordination mechanisms into a unified differential game framework, this research extends pollution control theory to engineered water network systems and contributes to cross-level environmental governance analysis. These findings highlight the institutional importance of multi-level coordination in engineered water network systems. Compared with traditional centralized or fully decentralized governance models, properly designed cross-level cost-sharing arrangements provide stronger and more stable incentives for sustainable pollution control in complex engineered environments. From a practical water resources management perspective, improving water quality in national water networks requires establishing stable cross-regional cost-sharing mechanisms through intergovernmental fiscal transfer design and performance-based pollution reduction agreements, strengthening central coordination capacity, and combining institutional innovation with technological upgrading in pollution treatment. Such integrated governance approaches are essential for enhancing long-term environmental resilience and ensuring effective pollution mitigation under changing environmental conditions. Despite these contributions, several limitations should be acknowledged. The model relies on stylized parameter assumptions and aggregated pollution indicators rather than detailed hydrodynamic or nutrient-process simulations. Data constraints limit the incorporation of uncertainty, spatial heterogeneity, and dynamic ecological feedbacks. Moreover, stakeholder behavior is modeled under rationality assumptions, which may not fully capture real institutional complexity. Future research should address several urgent technical gaps, including integrating governance models with dynamic water quality simulations to better link institutional design with ecological responses, incorporating uncertainty and climate variability into multi-level coordination frameworks, and expanding stakeholder structures to reflect evolving governance complexity in large-scale engineered water systems. In addition, future research could place greater emphasis on specific case studies of typical inter-basin water transfer projects, in order to further examine the practical applicability of the proposed governance mechanism under real-world conditions. Such efforts would also help to identify region-specific pollution control challenges during actual engineering operations and support the development of more targeted policy recommendations for water network management.