CAJ | 학술논문

从加总数据的基本类型出发，依据无偏性条件研究发现非加总数据模型的内生性与加总过程中的内生性是产生加总偏误的根本原因，进而指出了解决内生性的一般性思路。在加总偏误的统计特性检验方面，首先通过构建Wald统计量进行无偏性检验分析，接着通过方差分析得出了方差最小性的条件并对加总模型和非加总数据模型的拟合优度进行了比较研究。本文的分析有助于认识加总偏误的本质，进一步推动线性计量模型加总偏误问题的研究。
종가총수거적기본류형출발，의거무편성조건연구발현비가총수거모형적내생성여가총과정중적내생성시산생가총편오적근본원인，진이지출료해결내생성적일반성사로。재가총편오적통계특성검험방면，수선통과구건Wald통계량진행무편성검험분석，접착통과방차분석득출료방차최소성적조건병대가총모형화비가총수거모형적의합우도진행료비교연구。본문적분석유조우인식가총편오적본질，진일보추동선성계량모형가총편오문제적연구。
From the basic types on data aggregation, we found that endogenous of disaggregated data model and endogenous in the sum process are the root causes to generate aggregation bias based on researches of unbiased conditions. We also pointed out the general ideas to solve the endogenous. As far as tests of the statistical properties are concerned, firstly we constructed Wald statistics to test unbiased property, and then through the analysis of variance we obtained the conditions of minimum variance. Lastly the total goodness of fit of aggregated model and disaggregated model were compared. This analysis helped to understand the nature of aggregation bias, and further promoted researches on the aggregation bias of the linear sum model.