节点文献
环境污染监督博弈的动态性分析与控制策略
Dynamic Analysis and Control Strategy of the Supervision Game in Environmental Pollution
【作者】 蔡玲如;
【导师】 王红卫;
【作者基本信息】 华中科技大学 , 系统工程, 2010, 博士
【摘要】 政府对生产排污企业的监管机制是降低环境污染失控引起的风险和落实各项环境政策的关键,环境污染问题的动态性和复杂性使得对它的预测和控制经常失效,因此,对环境污染监管机制的研究刻不容缓。博弈论为解决环境问题中各种冲突关系提供了一种有效的理论工具。然而,目前利用传统博弈论分析环境问题主要关注博弈均衡的求解问题,忽略了有限理性条件下动态博弈过程的不确定性;近年来基于有限理性的演化博弈研究主要讨论演化稳定策略的存在性,缺少对该问题相关控制策略的分析;而针对环境污染中多人博弈问题,现有文献主要从企业“联合排污”的合作角度分析,忽略了企业之间普遍存在的“竞争”关系。鉴于此,本文尝试分析政府与生产排污企业之间监督博弈的动态演化过程,利用惩罚策略对博弈过程的波动性进行优化控制,抑制企业的超标排污行为。本文以政府与生产排污企业之间的监督博弈演化过程为研究对象,以优化惩罚策略为主要控制手段,围绕博弈的动态性分析和稳定性控制展开研究。文中利用博弈论对政府与生产排污企业之间长期的动态监督关系进行建模,将动态系统分析工具与系统动力学计算机仿真手段相结合,分析惩罚策略等环境政策对博弈过程和均衡的影响,揭示博弈过程的动态性,并尝试给出惩罚策略优化结构模式。具体来说,本文首先建立政府与生产排污企业之间混合战略博弈的系统动力学仿真模型,在假设博弈参与者策略变化率与期望获益成正比的条件下,分别分析了初始模型和信息延迟模型。在初始模型中假设信息获取不存在延迟,则博弈均衡需要经过多次博弈才能达到;在信息延迟模型中,针对博弈均衡的不可达提出了双重惩罚策略。通过稳定性分析指出了模型的均衡状态为临界稳定,任何微小的扰动都可能造成博弈均衡的不可达,从而揭示博弈过程动态性研究的重要性。接着,考虑有限理性条件下政府与生产排污企业之间的监督博弈动态演化过程,针对一般支付矩阵下不存在演化稳定策略的情况,提出了罚款额度与污染程度相关的动态惩罚策略,并通过理论分析和仿真证明了动态惩罚策略条件下演化稳定均衡的存在性。随后,进一步考虑政府对竞争条件下两个企业的监督博弈动态演化过程。通过对多人不对称演化博弈过程的仿真分析,指出罚款额度的惩罚系数k的大小与博弈过程的动态性密切相关。一般惩罚策略能有效抑制环境污染,而动态惩罚策略则在博弈过程波动性控制方面具有明显效果。以控制环境污染和抑制博弈过程波动性为目标,本文最后优化惩罚策略模式结构。分析了带停产整顿期(SOP)和不带停产整顿期(NSOP)的两种惩罚机制下政府与生产排污企业的微分博弈模型,指出了优化的惩罚策略模式不仅与企业超标排污行为相关,而且与政府环境监管部门的策略选择相关。
【Abstract】 The supervision mechanism of environmental regulation is critical to reduce the risk of the uncontrollable pollution and to ensure the performance of environment policies. The dynamics and complexity of the environmental pollution commonly invalidates the prediction and control strategies in real life. Therefore, it is essential to investigate the environmental supervision mechanism. Game theory is an effective methodology to resolve the conflicts on the environmental resources. However, the related ressarch works up to date mainly focus on finding the equilibrium states using traditional game theory, regardless of the uncertainties induced by the bounded rationality in the dynamic gaming procedure. The recent studies on the application of evolutionary game mostly pay attention to the existence of the Evolutionary Stable Strategy, in spite of the discussion about the control strategies. Researches on multi-person game focused on the "cooperative pollution" firms, neglecting the fact that the competitions between firms are more common. Thus, the objective of this thesis is to investigate the dynamic procedure of supervisory game and to optimize the penalty mechanism to control the over-pollution behavior of firms.This thesis investigates the dynamics analysis of evolutionary procedure of supervisory game and the optimization of punishment strategy as a control method. The long term dynamic supervisory relationship between government and polluting firms is modelled using game theory. The dynamic systems analysis methodology combined with System Dynamics (SD) are used to study the effects of environment policies, especially the penalty mechanism, on the Nash equilibrium and the complex dynamic gaming procedure. An optimized punishment strategy is proposed in a differential game framework.In particular, this thesis firstly develops a System Dynamic model for mixed-strategy game between the government and the polluting firm, without or with the consideration of information delay. For the case without information delay, it takes both game players a long time to reach the Nash equilibrium. For the case with information delay, a double penalty is proposed to deal with the unreachable Nash equilibrium. As a critical stable state, the Nash equilibrium might be unreachable by any small perturbation. So the dynamic analysis of paths to the equilibrium state is more significant than the Nash equilibrium itself.Subsequently, considering the bounded rational of decision-makers, the dynamic evolutionary procedure of two-player asymmetric game model is investigated. A dynamic penalty is suggested to deal with the problem that there does not exit evolutionary stable strategy (ESS) under the condition of static payoff matrix. Theory analysis and computer simulation validate the dynamic penalty.This thesis further considers the competitive relationships between the polluting firms. The dynamic procedure of multi-person asymmetric evolutionary game model is developed. The penalty factor k, which indicates the limit of fine, is closely related with the volatility of the dynamic evolutionary procedure. The static penalty has a better effect on the restrain of environmental pollution, while the dynamic one can stabilize the fluctuation during the evolutionary game procedure.Lastly, it is considered, within the framework of a differential game, that the enforcement of regulator using an appropriate penalty can force a polluting firm to act in a socially optimal way. Two different penalty mechanisms, one with the suspension of production (SOP) and the other without SOP (NSOP), are discussed. A completely restraining penalty (CRP), which relates to both players’ strategies, is identified to reduce the probability of the firm’s pollution behaviors.