人类合作问题的经济学理论与实验研究

Cooperation in social dilemma problems has been a major research topic in many scientific disciplines. Various settings and modeling paradigms have been investigated to induce prevalence of cooperation that is infeasible in the baseline of random matching one-shot encounters without community information. The current proposal contributes to this literature with several novel approaches. .Project 1 is based on the “assortative matching” literature (Frank, 1988; Sobel and Wilson, 1998). Following Amann and Yang (1998) and Yang et al. (2007), we further design an experiment that illustrates the complementary effect of maintaining an AM system and the (state) promotion of moral role models. Motivated by Fukuyama (2011) and Zhou (2008), we further investigate its historical reflections in the Confucian doctrines and political practices of Chinese dynasties, which helps explain the “China growth miracle” and support the “Culture and Path confidence” proposals..Project 2 proposes a series of experiment of budget-balanced, endogenous reward setups to reverse the decay trend on cooperation in the public goods game, as alternatives to voluntary punishment, which may help solve the hot reward-punishment debate in the literature (Fehr and Gaechter, 2002; Rand et al., 2009; Guala, 2012)..Project 3 starts with experiments to test theory predictions in the voluntary separation setup of repeated PD. After observation of data, we construct new equilibrium strategies where players prefer within-partnership punishment instead of breaking up the partnership with the defector being more cooperative in the next period, for better data fit. We apply the new stats method of "finite-mixture models" for strategy fitting, following Breitmoser (2015)..Project 4 investigates the effect of higher-order information on reputation building and PD behavior. We construct the novel notion of “continuous reputation scores” that enables us to sort out subject behavior after viewing naturally available trading histories on the partner. This finding can fill the gap between theory and experimental analysis in the Indirect Reciprocity literature (Nowak and Sigmund, 2005), where theory claims the necessity of higher order information in construction of any coherent reputation sccore while empirical evidence has been hard to find.

群体合作问题为多种学科共同关心，主要研究如囚徒困境、公地悲剧、信任等社会困境问题。本项目在文献为解决社会困境问题开发出的多种理论范式其中4种中，分别提出新颖独特的研究进路。1.分别从理论实验及历史文献角度讨论物以类聚机制的前世今生，发现在文化体系中强调物以类聚行为，并通过政府旌表教化鼓励民众中的亲社会行为，可以有效解决合作问题，建构经济发展的基础。2.通过实验设计在文献中首先发现让自主内生奖励可以有效解决搭便车问题的机构框架，阐释‘条件合作’行为模型在其中的解释功能；可协助厘清文献中关于奖励和惩罚孰优孰劣的不休争论。3.在重复博弈囚徒困境框架中引入自由分手选择，从实验数据和理论创新角度双管齐下，论证现实中常见的‘关系出问题不分手’原则的合理性。并探讨福利最大解的参数区间。4. 通过拓展‘间接报复’进路中信誉分值的定义，回归分析实验数据找到高阶信息影响交易决策的证据和方式，填补重要学术空白。

2020诺奖得主米尔格罗姆认为合作及协调是经济及管理问题背后应关注的最根本概念。.本项目在三个方向拓展了关于合作问题的研究。.（一）从不同角度、引入不同机制结构研究内生解决公共品一类合作问题。我们在文献中首次通过内部自主奖励分配机制，实现了稳定逆转公地悲剧搭便车的结果。并且通过拓展条件合作模型以及基于此的仿真运算，推测其原动力是个体条件合作行为在我们的机制中得以有效发挥作用，如鱼得水。.（二）在间接对等施报(indirect reciprocity)问题中，通过拓展高阶声誉信号的常规定义，在允许观察多轮二阶信息的囚徒困境实验数据中，创新性拟合观察到高级声誉信号正向影响合作决策，远不止文献曾观察到的实验vs对照组比较效果。.（三）在物以类聚配对机制的实验中，通过引入少量机器合作者(dummies)，提高被试群合作行为初始比例，观察到显著的“德不孤必有邻”的合作提升效果。.本项目共发表论文7篇，其中包括1篇PNAS（综合类顶刊），1篇MS（管理top2），2篇GEB（经济A），1篇JEBO（经济B）。