CAJ | 학술논문

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 xt （β＾2－β＾1）。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。
개소료면판수거모형판별적보취화일반방법，분석료면판수거모형중적고정효응적본질이급수거포함적시간서렬화횡절면적이질성문제。분석표명，시간고정효응시유선성약속적횡절면회귀，횡절면고정효응유유사결과；유우면판수거2개유도유시상관이질，이차저충이질성병불시완전불가관측적，인차경전모형중적고정효응제료진정불가관측적이질성＾ηt ，환유원본가이관측적 xt （β＾2－β＾1）。인차제출료개진적시간고정효응모형，병이용실례검험차모형。
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 con-tained 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 -^β( )1 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.