用面板数据回归模型时,在固定效应输出窗口中进行F检验后,F检验的下面出现了新的固定效应的回归结果,和之前的不一样,到底以哪个为准呢?
Dependent Variable: LNFDI?
Method: Pooled EGLS (Cross-section weights)
Date: 06/26/14 Time: 14:30
Sample (adjusted): 2002 2012
Included observations: 11
Cross-sections included: 20
Total pool (balanced) observations: 219
Linear estimation after one-step weighting matrix
Variable Coefficient Std. Error t-Statistic Prob.
C -37.04159418833498 18.54191704715581 -1.997721923473758 0.047143861963324
GNI? -0.08433880121466214 1.939785271004994 -0.04347842128472631 0.9653648825394445
LNGDP? 3.584812403559514 1.630801966041787 2.198189895650187 0.02911868493891859
T? -0.00109790716556217 0.1131317851208201 -0.009704674635776766 0.9922667618332505
NXN? 0.0250278310890828 0.1109550723723026 0.2255672548714449 0.8217754584268284
XN? -0.359494944863983 0.5101654749629212 -0.7046634131604281 0.4818648076585521
上面的是初始的固定效应模型,结果很不好,下面是F检验的结果:
Redundant Fixed Effects Tests
Pool: POOLUS
Test cross-section fixed effects
Effects Test st d.f. prob
Cross-section F 6.703182274122195 (19,194) 0.0000
接着有:
Cross-section fixed effects test equation:
Dependent Variable: LNFDI?
Method: Panel EGLS (Cross-section weights)
Date: 06/26/14 Time: 14:58
Sample (adjusted): 2002 2012
Included observations: 11
Cross-sections included: 20
Total pool (balanced) observations: 219
se pre-specified GLS weights
Variable Coefficient Std. Error t-Statistic Prob.
C 32.65318582532467 11.19128351001105 2.917733769867871 0.003903996916132306
GNI? 6.002668885562823 0.6815310875256581 8.807623005659055 0.0000
LNGDP? -2.382690006624018 0.9547065785458681 -2.495730164814762 0.01332851531930182
T? -0.6248017707566306 0.06860931092800126 -9.106661505641686 0.0000
NTPP? 0.1691123908247177 0.129860603006215 1.30226093911349 0.1942335947309213
TPP? -1.364010406037612 0.5549087809790651 -2.45808041392135 0.01476654114026601
Weighted Statistics
R-squared 0.6089429679902279 Mean dependent var 10.41938134065428
Adjusted R-squared 0.5997632254547872 S.D. dependent var 16.45962978303653
S.E. of regression 5.362046067941648 Sum squared resid 6124.077601397166
F-statistic 66.33551710619915 Durbin-Watson stat 1.459382280893585
Prob(F-statistic) 1.511151562977284e-41
两部分红色字体部分不一致,pooled EGLS和panle EGLS有什么区别呢?由于两部分的参数检验结果不一致,影响最终结果,所以迫切想知道到底该以那个结果为准,望高手指教。