论文部分内容阅读
利用从国内66篇文献中提取的126个经验结果,使用Meta回归方法,探讨了经验研究在样本构成、模型设定、估计方法等方面的差异如何显著影响了寿险需求的经验结果。研究发现,样本中投资型寿险的占比越高,则得到的社会保障、利率、居民储蓄对寿险需求的正向影响越大。是否采用逐步回归进行变量选择、自变量数目、是否采用动态模型、是否控制不可观测异质性等研究设计特征也显著影响了部分因素的经验结果。此外发现,该领域的研究存在一定程度的发表偏倚现象。
By using 126 empirical results extracted from 66 domestic literature and using the Meta regression method, we discuss the empirical results of empirical studies on how the differences in sample composition, model setting and estimation methods significantly affect the life insurance demand. The study found that the higher the proportion of investment-based life insurance in the sample, the greater the positive impact of social security, interest rate and household savings on life insurance demand. Whether stepwise regression is used for variable selection, the number of independent variables, whether or not dynamic models are used, and control of unobservable heterogeneity also influence the empirical results of some factors significantly. In addition, it was found that there was a certain degree of publication bias in the research in this field.