Robust estimation for linear panel data models

被引:14
作者
Beyaztas, Beste Hamiye [1 ]
Bandyopadhyay, Soutir [2 ]
机构
[1] Istanbul Medeniyet Univ, Dept Stat, Istanbul, Turkey
[2] Stat Colorado Sch Mines, Dept Appl Math, Golden, CO USA
基金
美国国家科学基金会;
关键词
fixed effects; least squares; panel data; random effects; robust estimation; weighted likelihood; MINIMUM HELLINGER DISTANCE; CROSS-SECTION; TIME-SERIES; REGRESSION; EQUATIONS; DEMAND;
D O I
10.1002/sim.8732
中图分类号
Q [生物科学];
学科分类号
07 ; 0710 ; 09 ;
摘要
In different fields of applications including, but not limited to, behavioral, environmental, medical sciences, and econometrics, the use of panel data regression models has become increasingly popular as a general framework for making meaningful statistical inferences. However, when the ordinary least squares (OLS) method is used to estimate the model parameters, presence of outliers may significantly alter the adequacy of such models by producing biased and inefficient estimates. In this work, we propose a new, weighted likelihood based robust estimation procedure for linear panel data models with fixed and random effects. The finite sample performances of the proposed estimators have been illustrated through an extensive simulation study as well as with an application to blood pressure dataset. Our thorough study demonstrates that the proposed estimators show significantly better performances over the traditional methods in the presence of outliers and produce competitive results to the OLS based estimates when no outliers are present in the dataset.
引用
收藏
页码:4421 / 4438
页数:18
相关论文
共 49 条
[1]   Robust stepwise regression [J].
Agostinelli, C .
JOURNAL OF APPLIED STATISTICS, 2002, 29 (06) :825-840
[2]   A one-step robust estimator for regression based on the weighted likelihood reweighting scheme [J].
Agostinelli, C ;
Markatou, M .
STATISTICS & PROBABILITY LETTERS, 1998, 37 (04) :341-350
[3]  
Agostinelli C, 2001, STAT SINICA, V11, P499
[4]  
Aitkin A., 1935, P RSE, V55, P42, DOI 10.1017/S0370164600014346
[5]  
Allison P. D., 2009, quantitative applications in the social sciences, V160
[6]   One-step robust estimation of fixed-effects panel data models [J].
Aquaro, M. ;
Cizek, P. .
COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2013, 57 (01) :536-548
[7]   THEORY AND METHODS OF PANEL DATA MODELS WITH INTERACTIVE EFFECTS [J].
Bai, Jushan ;
Li, Kunpeng .
ANNALS OF STATISTICS, 2014, 42 (01) :142-170
[8]   POOLING CROSS SECTION AND TIME SERIES DATA IN ESTIMATION OF A DYNAMIC MODEL - DEMAND FOR NATURAL GAS [J].
BALESTRA, P ;
NERLOVE, M .
ECONOMETRICA, 1966, 34 (03) :585-&
[9]  
Baltagi B.H., 2005, Econometric Analysis of Panel Data
[10]   SHORT AND LONG RUN EFFECTS IN POOLED MODELS [J].
BALTAGI, BH ;
GRIFFIN, JM .
INTERNATIONAL ECONOMIC REVIEW, 1984, 25 (03) :631-645