风险态度的稳定性——来自研究文献和微观数据库的检验

Stability of Risk Attitudes:Tests from Research Literature and Microdata

风险态度是稳定的吗?在保险经济学理论建模时,决策主体的风险态度常用风险厌恶系数来描述,这一系数多设定为常数,其中隐含的假设是,风险态度在不同时间和不同群体中保持稳定。本文分别采用手工整理的46篇测算中国和美国风险态度的文献数据库,以及CFPS和CHFS两个微观调查数据库,来重新检验这一假设。研究发现:第一,文献元分析显示,不同文献测算的风险态度差异明显,表现出时间维度、测算方法和群体三方面的异质性。第二,微观调查结果显示,在时间维度上,风险态度相关系数低于0.3;在方法维度上,不同测算方法下的风险态度之间的相关系数低于0.1;在群体维度上,风险态度在特定群体内和在群体之间均不稳定,而且在控制家庭财务决策者、家庭平均和家庭整体3个层面的7类特征后,风险态度依旧不稳定。进一步地,本文还发现:年龄、健康状况、就业状况会影响风险态度,这也是后者变化的根源。本文的经验证据显示,保险经济学理论建模设定风险态度为常数是一个强假设,需放松以更准确刻画现实和设计保险产品。此外,由于风险态度易受外生冲击的诱导,短期内消费者可能做出非理性行为,这就要求保险产品设计中要更加重视消费者保护。

Is risk attitude stable?In the theoretical modeling of insurance economics,the risk aversion coefficient is often used to describe the risk attitude of decision-makers,and this coefficient is mostly set as a constant.The implicit assumption of this setting is that risk attitudes remain stable at different time and among different groups.In this paper,we re-test this hypothesis by using 46 manually compiled literature databases for measuring risk attitudes in China and the United States,as well as two micro-survey databases,CFPS and CHFS.The literature meta-analysis shows that risk attitudes measured by different literature are significantly different,showing heterogeneity in three dimensions:time dimension,measurement method,and group.The micro-survey results show that the correlation coefficient of risk attitude is less than O.3 in the time dimension.The correlation coefficient between risk attitudes under different measurement methods is less than O.1.Moreover,risk attitudes are unstable within specific groups and between groups.After controlling seven characteristics of family financial decision-makers,family average,and family overall,the risk attitude is still unstable.We also find that age,health status,and employment status affect risk attitudes,which are also the root cause of changes in the latter.