The evolutionary mechanism of haze collaborative governance: novel evidence from a tripartite evolutionary game model and a case study in China

The paper addresses how China’s central government and local governments behave in collaborative haze governance under the Central Environmental Protection Inspection (CEPI) system. It highlights principal–agent problems arising from fiscal decentralization, information asymmetry, and bounded rationality, which hinder effective policy implementation. The study asks: how do central and local governments act under CEPI, what are their tripartite evolutionary game strategies, and which factors shape these actions? It positions CEPI as a mandated collaborative governance mechanism that changes local incentives through top-down intervention. The authors argue that repeated interactions and bounded rationality make an evolutionary game approach suitable, and they aim to model the strategic dynamics and validate them with a CEPI case in the Beijing-Tianjin-Hebei and surrounding regions (BTHS).

The review covers three strands: (1) Collaborative environmental governance: Defined as multi-actor, multi-level coordination that can outperform single-center governance but depends on informal elements (trust, leadership, social capital) and participant characteristics; collaboration is not a panacea and mechanisms linking collaboration to outcomes include representation of environmental concerns, knowledge, deliberation, and networked conflict resolution and implementation. (2) Government strategies and dilemmas in haze governance: Local governments often prioritize GDP over environmental performance due to political incentives, frequent cadre rotations, and alliances with enterprises; principal–agent problems and information asymmetry cause non-cooperative enforcement dynamics, prompting central administrative control. (3) Evolution and effects of CEPI: Transition from Regional Environmental Protection Supervision centers (REPS) to CEPI increased authority and accountability, integrating public participation to reduce information asymmetry. Empirical work shows CEPI improves air quality and enforcement in the short term, with heterogeneous pollutant effects (e.g., PM2.5/PM10 down, O3 up) and concerns about high operating costs and long-term sustainability. The gaps identified include limited attention to CEPI within strategy research, need for evolutionary models accommodating bounded rationality, and lack of combined theoretical–case analysis.

The study develops a tripartite evolutionary game model (TEGM) among the central government and two local governments (A and B) under CEPI. Assumptions: local governments implement central Air Pollution Control Policies (APCP) either fully or incompletely; the central government inspects thoroughly or incompletely; both levels value economic growth and air quality; full inspection triggers penalties for non-compliance. Variables include implementation costs (C_A, C_B), inspection cost (C_c), environmental loss and economic benefit differentials (Q, G), performance evaluation weights (δ1 for environment, δ2 for economy), industrial transfer losses/benefits (R, T), cross-regional environmental loss (W), fiscal subsidies (M), penalties (F_A, F_B), central government pressure/reputation (F_c, H), and rent-seeking (B) with impact share λ. Payoff matrix is specified for all strategy profiles; expected payoffs for each player’s strategy and replicator dynamic equations are derived using a Malthusian dynamic framework. The Jacobian and eigenvalues at equilibrium points are analyzed to determine evolutionary stable strategies and conditions under which locals fully implement and the center inspects thoroughly. To validate the model, a qualitative case study is conducted on CEPI in the BTHS regions (Beijing, Tianjin, Hebei, Henan, Shandong, Shanxi, Inner Mongolia), using policy documents, official statistics (2006 vs 2017), and authoritative news to examine mechanisms: performance evaluation adjustments, implementation costs, clean government construction, industrial transfer trends, fiscal subsidies, inspection costs, accountability, and public participation. The CEPI operational process (preparation, stationed inspection, reporting, feedback, handover, rectification, archiving) and the handling of public petitions are documented.

- Main determinants of collaborative haze governance: performance evaluation system weights (δ1↑, δ2↓), policy implementation cost reductions, clean government construction (reducing λB), industrial transfer trends (lower R and T), fiscal subsidies (M↑), lower inspection costs (C_c↓), stronger environmental accountability (F_A, F_B↑), and greater public participation (heightened F_c and H effects). - Local governments’ full implementation conditions: Inequalities indicate full implementation is favored when rent-seeking gains and industrial transfer advantages are outweighed by subsidies, environmental weight in evaluations, and reduced costs: T_i − M_i − δ1 Q_i + δ2 G_i + λB + C_i < 0 (for i=A,B). Practical levers: increase environmental evaluation weight; reduce execution costs; curb corruption; reduce industrial transfer incentives; and expand fiscal subsidies. - Central government’s thorough inspection conditions: When at least one local does not fully implement, thorough inspection is favored if C_c − F_B − F_c < 0; C_c − F_A − F_c < 0; and C_c − F_A − F_B − F_c < 0. Levers: reduce inspection costs, raise actual penalties, and increase public participation to heighten reputational/pressure costs of weak inspection. - CEPI supports collaborative governance by reshaping incentives, improving information flows (public petitions), and strengthening accountability. - Case evidence (BTHS): • Environmental outcomes: Per capita SO2 emissions fell from 10.99–64.47 kg (2006) to 0.93–21.60 kg (2017); per capita soot and dust fell from 5.01–50.48 kg to 0.94–21.20 kg, consistent with stronger environmental evaluation and implementation. • Economic/structural change: GDP per capita rose from 1.32–5.07 ten-thousand yuan (2006) to 3.26–10.68 ten-thousand yuan (2017); secondary industry shares declined across all regions (e.g., Beijing from 27% to 19%; Tianjin 55% to 41%), reducing implementation costs and industrial transfer pressures. • Fiscal support: Central air pollution control funds pre-allocated to BTHS from 2020–2023 were 17.90, 10.78, 10.71, and 10.69 billion yuan, supporting local implementation. • Public participation: Over 80,000 public environmental concerns were resolved during CEPI Round 1; the petitions process reduced information asymmetry and increased transparency and pressure on local implementation. • Costs and accountability: CEPI incurs substantial administrative and human resource costs across multiple actors; institutional upgrades enhanced accountability and penalty capacity, aligning with model predictions that lower C_c and higher F increase central inspection likelihood. Overall, empirical patterns align with the TEGM’s predicted mechanisms: adjusting incentives and costs fosters full local implementation and thorough central inspection, advancing collaborative haze governance.

The findings address the research questions by showing how CEPI, as a mandated collaborative governance mechanism, reconfigures incentives in a tripartite central–local game. When performance evaluations value environmental outcomes more, local implementation costs are reduced or shared, and corruption rents are deterred, local governments converge toward full APCP implementation. Concurrently, reducing inspection costs, increasing penalties, and mobilizing public oversight drive the central government toward thorough inspection. The BTHS case corroborates these dynamics: improved air quality indicators, economic upgrading, robust public petition handling, and increased fiscal support collectively translated into stronger collaboration and enforcement. The interplay of horizontal (cross-regional industrial coordination) and vertical (central inspection and accountability) collaboration reduces externalities from cross-boundary pollution and mitigates principal–agent problems. These results underscore the importance of aligning performance metrics, providing fiscal incentives, institutionalizing accountability, and lowering administrative costs to sustain long-term collaborative haze governance.

The paper contributes a tripartite evolutionary game framework that explains central–local collaborative dynamics in haze governance under CEPI and validates it with evidence from the BTHS regions. It identifies key levers: performance evaluation weights, cost-sharing to lower implementation costs, anti-corruption/clean governance, industrial restructuring to curb transfer externalities, fiscal subsidies and ecological compensation, reduced inspection costs through cost–benefit evaluation frameworks, stronger accountability, and enhanced public participation. Policy recommendations include: optimizing green performance evaluations; sharing implementation costs and internalizing cross-regional externalities; strengthening anti-corruption and transparency; coordinating cross-regional industrial upgrading; allocating fiscal subsidies and transfer payments to compensate for externalities; normalizing supervision and increasing penalties for non-compliance; establishing integrated cost–benefit assessment frameworks to reduce CEPI costs; and broadening public and third-party oversight channels. These measures can enhance collaborative governance effectiveness and offer lessons for developing countries confronting cross-regional air pollution.

The study does not fully explore the evolutionary mechanism of different policy combinations due to space and feasibility constraints. It focuses primarily on provincial-level governments, without examining municipal or county-level dynamics. Future research should analyze policy combination effects and extend to lower tiers of government for deeper insights and generalizability.