【作者】 章平；

【导师】 朱宪辰；

【作者基本信息】 南京理工大学 ， 管理科学与工程， 2008， 博士

【摘要】 在中国特定知识传统下,人们正经历从依赖科层化的垂直行政权力支配体系向市场化水平协调规则体系的社会转型,此时考察个体观念转变的知识更新过程显得尤为必要。通过对南京市住宅小区的两次抽样调查(30个小区878名业主、28个小区865名业主)发现,住宅小区共享资源治理转型困境的实质是治理主体自身决策知识限制导致的认知不协调。政府退出供给之后,绝大多数业主的认知状态是:趋于支配共享资源决策权利、同时回避与之对应、对等的供给共享资源的责任,缺乏市场条件下自我治理的决策知识。由此引出:如何刻画知识异质性个体在同一个社会环境中互动协调、调整信念和决策规则以习得新场景下决策知识的过程?围绕上述问题,本文结合动态过程和均衡选择两种分析范式,基于个体知识习得行为考察个体间协调互动的治理模式选择问题。全文研究工作如下:(1)将知识学习行为嵌入在社会经济转型过程中考察。住宅小区共享资源治理问题,在经典文献中属于公共物品和集体行动理论范畴。针对转型期经济的特殊性,本文率先从微观主体认知状态和学习行为角度考察这一特殊的公共物品的治理转型问题,研究个体认知状态调整、知识习得与制度转型的内在关联,强调个体信念调整、知识更新重于制度转型。整合当前个体层面的学习模型,用于探讨个体间协调互动、完成共享资源治理转型的过程。(2)系统研究重复协调博弈中的策略指导行为。将策略指导研究从两人3×3协调博弈拓展到3人3×3×3最小努力协调博弈,重点考察知识异质性与策略指导行为发生的关联,适应性学习、老练学习和策略指导行为的关系,并分析策略指导行为发生的激励和风险。(3)学习理论模型的实验表征。首次在博弈实验中引入知识异质性设置,并将参与同一博弈的同组被试双方知识状态设置差异,考察不同知识状态下的行为表现。将与被试博弈的另一方用固定选择纳什均衡行动作为策略指导的计算机代替,考察被策略指导的被试信念调整过程。将3人3×3×3最小努力协调博弈转化成形同2人3×3的记录形式,便于博弈学习模型的参数估计和拟合效果评价。用实验记录界定了EHOS模型的三类不同行动:往前推算一步的“Level 1”行动、往前推算两步的“Level2”行动和不做换位思考、随机决策的“Level 0”行动。实验研究结果表明:(1)在两人3×3纯策略纳什均衡博弈、3人3×3×3博弈,均存在显著的策略指导行为。存在策略指导行为的被试组收敛可能性显著高于不存在策略指导行为的被试组。知识异质被试组发生策略指导可能性显著低于知识同质的被试组。同一组参与者规模增大,使得策略指导对博弈均衡收敛的促进作用减弱。成功的指导者得到的支付显著比未收敛者多,但显著低于追随者(适应性学习者)。(2)在知识同质(AP)、固定匹配和知识异质、固定匹配两类重复博弈实验中,EHOS模型的拟合效果最优,即能够同时体现具备双方决策知识(AP)的个体信念形成和决策过程中“前瞻”和“后顾”特征,且该类个体更易于成为策略指导者。与强化模型、虚拟博弈学习和γ-加权信念比较,拟合效果显著改善。在知识同质(AP)、随机匹配和知识同质(AP)、与固定行动选择的计算机匹配两类重复博弈实验中,强化学习模型的拟和效果最好,这与所面临的场景有关,前者是高度不确定性,而后者是与确定行动选择的计算机互动。(3)对应现实问题,作为住宅小区共享资源局部治理制度的微观主体,业主的知识更新是决定该项制度变迁成败的关键所在。作为住宅小区共享资源外部主体的政府,能够做的是积极实现从管理到服务的职能转变,从制度上为共享资源自我治理扫清障碍,避免实施强化业主完全依赖政府的信念、阻碍学习新决策知识的政府行为。更多还原

【Abstract】 It is necessary to explore the updating process of individual knowledge, which through out the transition from hierarchical institutions of decision to horizontal coordination in market system, under traditional knowledge in a specific planning economy. Using the sampled investigation data from Nanjing residential communities, statistical results show that the essence of transition dilemma of Common Pool Resource governance in residential community is owners’ cognitive dissonance lead by the bound of decision knowledge. After the absence of state provision, most owners still prefer to gain the decision-making right of the commons while averse from the matching duty of providing. Then we present the problem on capturing the process of interactive coordination, adjusting of belief and decision-make rule and acquiring new knowledge.Applying paradigm of evolutionary dynamic process and equilibrium analysis, this paper explored the choice of governing model based on individual knowledge acquirement. The main work of this paper is as following.(1) The problem on common pool resource governance in residential community is treated as public good and collective action theory in classic literature. Considering the particularity of transition economy, we firstly introduced knowledge learning behavior into social economic transiting process, focusing on the inner relationship among individual cognitive state adjusting, acquisition of knowledge and institutional change. We also integrated existing individual learning models, using on individual coordination during CPR provision.(2) This paper systemically researched strategic teaching behavior during repeated coordination games. The research was extent from two persons 3×3×3 repeated coordination game to three persons minimum effort game, then the relation between knowledge heterogeneity and strategic teaching was investigated. In this paper we also paid attention to the motivation and risk of strategic teaching.(3) For the first time, this paper introduced the setting of knowledge heterogeneity in experimental game, that made one of partners know both the payoffs according to his choices and other partners’, while another only know his own payoffs according to his choices, then we can studied the performance under different knowledge states. Here we also demonstrated how subjects adjust their belief and decision make rule when we replaced their opponents by computers who chose fixed action from beginning to end. We distinguished Level 0 players who randomized and chose each strategy with probability 1/3, Level 1 players who chose the best response to his opponent’s last period’s action, and Level 2 players whose best response to last period’s outcome is the strategy which his opponent could be choosing according to Ehrblatt-Hyndman-Ozbay-Schotter model, using experimental data.The experiment results indicate that, (1) for two persons 3×3 and three persons 3×3×3 pure strategy games, there are strategic teaching behavior significantly. The matched subjects with teaching behavior converge more frequently than these without teaching behavior. The matched subjects with knowledge heterogeneity converge less frequently than these with knowledge homogeneity. When the scale of players expands, the role of teaching in facilitating convergence to unique pure strategy equilibria is decreasing. As their motive, successful teachers’ payoffs more than those players who do not converge, while less than followers significantly.(2) For repeated fixed matching experimental games with knowledge heterogeneity and knowledge homogeneity, EHOS model best fits these data. It captures both backward-looking and forward-looking characters in the process of individual belief formation and decision-making. Its Goodness-of-fit is also improved. For repeated random matching experimental games with knowledge heterogeneity and playing with computers who chose the same action during games, reinforcement model performs best, which is led by the extent of uncertainty of situations players facing.(3) Corresponding to issues come from transition of governance of common pool resource in residential community, its key factor is the knowledge updating process of house-owners, as they are the main inner partner of governing the common pool resource. As an outer partner, what the government should do is to transform its duty from management to service, clear away the obstacles of self-governing, and avoid to reinforce house-owners’ dependent on the government and discourage them to learning new decision knowledge.更多还原