摘要
对多个只含有个体效应的Panel数据模型,研究了模型中回归系数向量相等性的假设检验问题,提出了一种参数Bootstrap检验方法.有限样本的数值模拟研究结果表明,提出的检验方法具有良好.的检验功效,且受个体效应方差、误差方差、模型个数、回归系数维数的影响不明显.
In this paper, testing equality of regression coefficients in several one-way error component regression models is discussed. For the hypothesis testing problem, a parametric bootstrap is proposed. All the numerical results indicate that the PB test, regardless of the sample sizes, values of the variance components, number of the models, dimensions of regression coefficients, maintain the Type Ⅰ error rates very well.
引文
[1] Baltagi B H. Econometric Analysis of Panel Data[M]. New York:John Wiley&Sons, 2008.
[2] Rao J N K, Sutradhar B C, Yue K. Generalized least squares F-test in regression analysis with two-stage cluster samples[J]. Journal of the American Statistical Association, 1993, 88:1388-1391.
[3] Wu C F, Holt D, Halmes D J. The effect of two-stage sampling on F statistics[J]. Journal of the American Statistical Association, 1988, 83:150-159.
[4] Wang S G, Liski E P. Small sample properties of the two-way error component regression[J]. ACTA Mathematica Application Sinica, 1999, 3:287-296.
[5] Wang S G, Ma W Q. On exact tests of linear hypothesis in linear models with nested error structure[J]. Journal of Statistical Planning and Inference, 2002, 106:225-233.
[6] Yue L L, Shi J H, Song W X. A parametric bootstrap approach for two-way error component regression models[J], Communications in Statistics-Simulation and Computation, 2017, 46(5):3952-3961.
[7] Tian L L, Ma C X, Vexler A. A parametric bootstrap test for comparing heteroscedastic regression models[J]. Communications in Statistics-Simulation and Computation, 2009, 38(5):1026-1036.
[8] Krishnamoorthy K, Lu F. A parametric bootstrap solution to the MANOVA under heteroscedasticity[J]. Journal of Statistical Computation and Simulation, 2010, 80(8):873-887.
[9]王松桂,尹素菊.线性混合模型参数的一种新估计[J]·中国科学,2002, 32(5):434-443.
[10]王松桂,史建红,尹素菊等.线性模型引论[M].北京:科学出版社,2004.
[11]王松桂,范永辉.Panel模型中两步估计的优良性[J].应用概率统计,1998, 14(2):177-184.