摘要
介绍了面板数据模型判别的步骤和一般方法,分析了面板数据模型中的固定效应的本质以及数据包含的时间序列和横截面的异质性问题.分析表明,时间固定效应是有线性约束的横截面回归,横截面固定效应有类似结果;由于面板数据2个维度有时相关异质,而且这种异质性并不是完全不可观测的,因此经典模型中的固定效应除了真正不可观测的异质性ηt,还有原本可以观测的(β2-β1).因此提出了改进的时间固定效应模型,并利用实例检验此模型.
The paper gives a brief description of the steps and the general approaches to determine the panel data models, and analyzes the nature of their fixed effects and the heterogeneity of time series and cross - section conmined in the data. It concludes that the time - fixed effects are a cross - sectional regression of linear constraints, similar to cross - sectional fixed effects. Because the two dimensions of the panel data are heterogeneous sometimes and the heterogeneity is not completely unobserved, the fixed effects of the classical model reveal x1 (β2 -β1 )that could be observed besides the real unobserved heterogeneity ηt. It finally brings up an improved model of the time -fixed effects, and conducts experiments to validate its efficiency.
出处
《云南民族大学学报(自然科学版)》
CAS
2015年第6期486-491,共6页
Journal of Yunnan Minzu University:Natural Sciences Edition
基金
国家自然科学基金(61103183)
关键词
面板数据
固定效应
时间序列
横截面
异质性
panel data
fixed effects
time series
cross - section
heterogeneity