摘要
中介效应分析的发展使得现有研究不再只是局限于变量之间的相互影响而是转向讨论这种影响的内在机制。基于两种中介效应分析方法即系数差异法和系数乘积法,本文系统地演绎了两种方法的分析步骤和内在差异,并讨论此两种方法在结论和应用上的一致性。此外,基于中国健康与营养调查(CHNS)数据,本文运用上述中介效应分析方法探讨农民个人收入与健康的因果关系,并基于人均农业税构建工具变量组解决个人收入在健康回归中的内生性问题。本文验证营养摄取和膳食偏好在农民个人收入与健康关系中的中介效应,借此详细阐述系数差异法与系数乘积法在实证研究中的具体应用。结果显示,两种方法所得结论一致:农民个人收入与健康的关系中约有12. 02%是通过个人收入影响营养摄取和膳食偏好进而影响农民健康,其中营养摄取的中介效应较大(解释率为10. 36%)。同时,个人收入可以独立的作用于健康,并可能通过其他不可观测的机制影响健康。
Under the framework of the mediating effect analysis,the existing literature goes beyond analyzing the mutual influence of variables but the internal mechanisms of these effects.Two kinds of mediation effects analysis methods including coefficient difference and coefficient product methods are documented.This paper systematically introduces the procedures and differences of the two methods,and tends to prove the consistency of them.Using the China Health and Nutrition Survey(CHNS)data,this study analyzes the relationship between farmers’income and health,as well as its mediation effects by using the two methods mentioned above.To address the endogeneity problem of income in the estimations,this study rests on the instrument variables(IV)to validate the mediation effects of nutritional intake and dietary preference on the relationship between farmers’income and health.The results show that the two decomposition methods draw the same conclusions as follows:first,approximately 12.02%of farmers’income effect on health is through the influences of nutrient intake and dietary preference,and nutritional intakes take the largest mediating effect(10.36%);secondly,individual income could affect health independently,and this effect might be through other unobservable mechanisms.
作者
李会
王晓兵
任彦军
LI Hui;WANG Xiaobing;REN Yanjun
出处
《农业技术经济》
CSSCI
北大核心
2019年第9期58-72,共15页
Journal of Agrotechnical Economics
基金
国家自然科学基金面上项目“区域间不同经营主体适度经营规模及其影响研究”(编号:71673008)
国家自然科学基金政策研究重点项目“新时期中国农业与农村发展战略与政策研究”(编号:71742002)
关键词
中介效应
系数差异法
系数乘积法
Mediation effect
Coefficient difference method
Coefficient multiplication method