基于参照点依赖偏好的网络博弈学习动力学研究

随着信息通讯技术的发展，人们日渐清晰地认识到，各种经济管理活动均发生在一个相互连接的世界中。网络博弈学习理论是分析此类应用实践的有力工具，现有研究假定个体偏好遵循经典的期望效用理论。行为经济学实验研究发现，相较期望效用理论，前景理论可更好地刻画真实世界中个体决策行为。本项目将借鉴前景理论等，区分博弈收益和个体效用，引入“个体效用依赖于参照点”假定，沿着“从个人参照点到社会参照点再到个人和社会双重参照点”这条主线，逐步构建基于参照点依赖偏好的网络博弈学习动力学模型。然后，本项目将运用发展博弈学习理论，求解随机稳定均衡，估算收敛到随机稳定均衡速度，考察不同参照点依赖效用对网络博弈学习动力学长期行为的影响。通过引入“个体效用依赖于参照点”的假定，本项目不仅拓展和完善网络博弈论的基础研究，而且为技术创新扩散、供应链上下游协同等应用实践提供全新的分析框架，具有较大的应用价值。

With the progress of information and communication technologies, it is generally realized that all economic and management activities take place in an interconnected world. The theory of learning dynamics in network games has emerged as an important tool to analyze how local interactions affect large interconnected populations，in which the existing studies usually assume that individual preferences are determined by the expected utility theory. However, the literature of behavioral economics shows that in comparison with expected utility theory, prospect theory explains better individual decisions in the real world. With the help of prospect theory, this project will distinguish payoff from utility and assume that agents have reference-dependent preferences in order to better model agents’ decision making for learning dynamics in network games. This project will develop models of learning dynamics in network games with reference-dependent preferences, by progressing from individual to social reference points, then to both individual and social reference points. Then, by applying the theory of learning in games, this project will identify stochastically stable equilibria, evaluate the rate of convergence to stochastically stable equilibria, and explore the impact of various reference-dependent preferences on agents' long-run behavior. By introducing the assumption of reference-dependent preferences, this project not only extends the theoretical exploration of network game theory but also can be used to analyze the diffusion of new technologies and the supply chain coordination.