面向经济复杂性的行为建模与计算实验及应用研究

The important theoretic contribution of behavior and experimental economics focuses on the broad existence of cognitive behavioral bias with significant effect. Basing on deepening multi-attribute analysis of behaviors including heterogeneity and interaction, and by means of observation or experiment and making the obtained datum from real behavior parameterization with cardinal, ordinal number variables styles as well as distributional vectors, then introducing and incorporating them into classical decision-making model, and breaking through the RA (Representative Agent) analysis framework, for constructing the integrated model with behavior-through and macro-micro link feature as the experimental platform; With the cross matching simulation between multi-scenario, multi-state and multiple dimensions of behavioral properties, and to explore the inner relationship between microscopic behaviors and macroscopic phenomena and their interacting mechanism, and how to affect inversely the individual decisions, and it is greatly benefit for revealing the essence of an increasingly complex economic and financial activities from the source. There will be some new contributions in this research project in the endogenous parameterization of key behavioral characteristics, the behavioral relationships between individual and group, and conducting experiments by human subjects combined with virtual subjects methods etc., for promoting the traditional game experiment and econometrics methods as well as empirical positives. It is realizable to visually and interactively evaluate and foresee the policy effect, and to deal well with the interacting roles between the market-determine and the change of government's function, to provide more powerful and reliable decision-support, and to blaze a new path of exploring the humanistic nature of complex economy with interdisciplinary. At last, it is helpful, taking salary incentive of heterogeneous employees, the relationship between investor's behaviors and the complicated volatility of stock market as illustrating examples, to show the validity of applying behavior modeling and computing experiment.

行为与实验经济学的重要贡献突显认知行为bias的广泛存在及显著影响；以深化分析异质和交互等多元行为属性为基础，借助观察和实验并采用基数类、序数类变量及分布向量等方法将真实行为参数化后引入经典决策行为模型，突破RA分析框架，构建以行为贯通的宏微观连接的一体化模型作为实验平台；通过多情景状态与多元行为属性交叉匹配的仿真，探索微观行为与宏观异象的内在联系与交变机理，以及又如何影响个体决策，有助于从根源上揭示日趋复杂的经济金融活动的本质；在关键行为特征内生参数化、个体与群体行为关系和人类被试与虚拟主体结合的实验方法等方面推进计量实证与博弈实验等传统方法；能互动、可视化地预现和评价政策效应，为处理好市场的决定性作用与政府职能转变的关系提供更加有力可靠的支撑；开辟跨学科探索复杂经济问题人文本质的新途径；并以异质员工薪酬激励、投资行为与股市波动之间的复杂联动关系为例，说明行为建模与计算实验的可行有效性。

本项目按课题申报计划书的设定内容和技术路线逐步展开，并结合实际进展尽可能地深化和拓展，通过聚焦行为、由微观到宏观地逐层透析，探索和促进形成通宏洞微地分析复杂经济金融问题的理论（体系）和方法过程，并在金融市场、人力资源管理和公共政策等领域取得了预期的尝试应用效果。在揭示和界定经济金融复杂性的本质特征、类型刻画的基础上，所研究的基本内容、可能的创新之处和主要步骤为：. （1）深化微观分析，分类研究多元行为属性的异质性主体的行为响应类别、与外界环境条件变化互动的决策模式；借助实验方法获取真实情景中关键行为特征参数和交互连通路径，揭示个体与群体决策的复杂行为关系；结合利用行为大数据和人工智能技术等，为更有针对性的深入研究中国经济金融环境中特有的复杂问题奠定更为坚实的基础。. （2）构建微观宏观一体化模型平台，具有多学科大跨度深度交叉融合、整体论与还原论结合的特点，实现人•机一体、虚--实交融，沿个体-组织-社会的I-O-S技术路径分层估算校准参数和互联关系，以此作为分析经济复杂性的理论架构、模型框架和实验平台，探讨行为与情景（人与自然、相互之间）、微观与宏观的内生互反馈机制，为揭示复杂现象的本质特征和内在机理提供方法技术基础。. （3）运用计算实验方法破解复杂奥秘，探索宏观复杂形态和现象及典型化事实的微观成因、诱变条件、传导机理、演化路径、加总模式、临界阈值、整体涌现、递进反馈等，具有各关键环节特征凸显、环环相扣、整体贯通的特点与优势，拓展和提升了计量经济学等传统的计量实证方法，为分析复杂经济金融乃至人文社会科学问题及政策分析应用等开辟新的有效途径。