从众行为与创新扩散：基于信息成本角度的研究

On the one hand, we often observe herd behavior such as local governments' developing rising industries and investors' following suit in stock markets. On the other hand, we rarely see the diffusion of local government policy innovations. Inspired by the two classes of seemingly opposite phenomena, we investigate the impact of costly information acquisition on herd behavior and study the optimal policy intervention in social learning environments from the perspective of manipulating the cost of information acquisition. Our project bases itself on an economic intuition as follows. When perfectly observing predecessors' actions that are possibly identical, people may make an inference from predecessors' actions and then follow suit regardless of their own private information. Consequently, information held by the public can not be effectively disseminated and aggregated. When observing no predecessors' actions, especially in a scenario with a prohibitively high cost of information acquisition, people have to make choice decisions based only on their own private information and in turn there is no information dissemination at all. Formalizing the above intuition, subprojects 1 and 2 investigate theoretically and experimentally respectively the impact of costly information acquisition on the behavior and the welfare of bounded rational decision makers in social learning settings. Subproject 3 investigates the empirical evidence that costly information acquisition accounts for little diffusion of local government policy innovations. An answer to our research questions will not only make important theoretical contributions such as establishing a novel theory of herd behavior and the welfare analysis under the assumption of bounded rationality, but also have plenty of policy implications such as how to improve the diffusion of local government policy innovations.

地方政府跟风上马新兴产业、投资者股市跟风盲从等从众行为经常出现，但同时地方政府的大量政策创新却鲜少得到推广。针对这两类看似相反的现象，本课题探索信息成本如何影响从众行为的发生以及探索从众环境下以改变信息成本为主要手段的最优经济政策干预。其所依赖的经济学直觉在于：当观察他人行为的成本为零时，观察到多个他人的相同行为容易使得决策人忽略自有信息采取从众决定，从而造成分散在众人中的大部分信息难以传递；当该成本很高时，决策人会失去获取他人行为中所隐藏信息的机会，从而造成分散在众人中的信息无法传递。子课题1和2分别从理论和实验方面探索信息成本及其相关政策干预对有限理性人从众行为和福利水平的影响。子课题3从实证方面探索地方政府创新较少扩散的信息成本解释。本项目科学问题的解决将不仅带来包括建立有限理性人从众行为及其福利经济理论在内的重要理论突破，而且对促进地方政府创新扩散这一重大现实关切具有重要应用价值。

地方政府跟风上马新兴产业、投资者股市跟风盲从等从众行为经常出现，但同时地方政府的大量政策创新却鲜少得到推广。针对这两类看似相反的现象，本课题探索信息成本如何影响从众行为的发生以及探索从众环境下以改变信息成本为主要手段的最优经济政策干预。. 主要研究内容包括：个体决策者对直接信息和具有等同信息价值的间接信息的出价是否相同以及影响二者出价区别的原因；通过改变间接信息成本或者收取间接信息费用进行转移支付的干预机制提升从众环境下决策人整体的福利水平；公理化刻画具有有限容量的个体决策者的选择行为特征；通过信息设计机制提升博弈双方的信任程度、值得信任的程度以及整体的福利水平。. 重要研究结果包括：个体决策者对直接信息出价和间接信息的出价存在较大的个体异质性；对间接信息收取费用并将其平均分配给没有购买信息的决策人这种干预存在均衡状态并且可以提高决策人整体的福利水平；有限容量的个体选择模型存在公理化表达；理论和实验结果表明引入信息设计能够显著提升博弈双方的信任和促进市场交易。. 项目研究的科学意义包括：有助于理解从众行为实验研究文献中不同设计下互相矛盾的结果；对从众行为理论研究文献中的基本假设提出了挑战；填补了从众行为福利经济理论研究的空白；拓展了现有的两阶段个体选择模型；以及在信任研究文献和信息设计研究文献之间建立了联系。