《保险研究》20210903-《人寿保险市场多维不对称信息与逆向选择的识别——基于模型设定的检验和估计》（吴夕、周先波、甘犁）

［摘   要］保险市场中逆向选择的正确识别是判断市场是否有效、定量研究消费者福利损失的前提和基础。大量文献已发现,在多重不对称信息存在的情况下,传统正相关检验无法识别逆向选择,本文发现模型设定偏误可能是正相关检验失效的一个重要原因,通过应用稳健的Biprobit半非参数估计并配以Bootstrap检验,本文对正相关检验方法进行了优化改进,并将该方法应用至CHARLS数据。在控制了导致正向选择的不对称信息后,传统正相关检验仍无法识别我国寿险市场的逆向选择,而使用半非参数估计方法,则最终识别了寿险购买与事后风险的逆向选择效应。这一结果说明,虽然中国寿险市场的部分多维不对称信息一定程度缓解了市场信息偏差,但其中正向选择与逆向选择共存因素使市场整体仍具有较强的信息不对称性,导致市场失灵和社会福利下降的可能性上升。要使保险更好地发挥“稳定器”作用,保险市场的信息分类机制有待完善。

［关键词］逆向选择识别；多维不对称信息；半非参数估计

［基金项目］国家社科基金重大项目“增强消费对经济发展的基础性作用研究”(21ZDA036)；国家自然科学基金面上项目“政府减税的收入分配效应研究:来自增值税转型改革的证据”(71973157)和“不确定预期和家庭资产配置的联动性:影响机制与联立Tobit研究”(71773146)。

［作者简介］吴夕,首都经济贸易大学国际经济管理学院助理教授;周先波(通讯作者),中山大学岭南(大学)学院教授,研究方向:数量经济；甘犁,西南财经大学经济与管理研究院教授。

The Identification of Multi-dimension Asymmetric Information and Adverse Selection in China’s Life Insurance Market—Based on the testing and estimation of model settings

WU Xi,ZHOU-Xianbo,GAN Li

Abstract：The identification of adverse selection in insurance markets is the prerequisite and foundation for judging whether the insurance market is effective and conducting further consumer welfare analysis.A large amount of literature has shown that with the existence of multi-dimension asymmetric information,the classical positive correlation test cannot identify adverse selections.The paper found that model setting bias was another major reason for the failure of this method.It adopted the semi-non-parametric estimation of the Biprobit model,reinforced by Bootstrap testing,to optimize the classical positive correlation test,and applied this method to CHARLS data.After controlling the asymmetric information leading to positive selection,the classical positive correlation testing still failed to detect adverse selections in China’s life insurance market.The paper finally identified the adverse selection effects in life insurance purchase and ex post risks after employing the semi-non-parametric estimation method.This indicates that although the multi-dimension asymmetric information in China’s life insurance market alleviate the deviation of market information to some extent,the co-existence of positive selection and adverse selection make the market featured with quite strong information asymmetry on the whole.This increases the probability of market failure and social welfare loss.The information classification mechanism for the insurance market needs to be improved in order for the social insurance tool to better play its role of “social stabilizer”.

Key words：identification of adverse selection;multi-dimension asymmetric information;semi-non-parametric estimation