Correlated Random Effects - Hausman Test
Pool: QQQQ
Test cross-section random effects
Test Summary Chi-Sq. Statistic Chi-Sq. d.f. Prob.
Cross-section random 7.245345 3 0.0645
Cross-section random effects test comparisons:
Variable Fixed Random Var(Diff.) Prob.
LOG(RAILWAY?) 0.836952 0.852806 0.007595 0.8557
LOG(RIVER?) 1.085986 0.159457 0.205159 0.0408
LOG(ROAD?) 1.261898 1.266622 0.000492 0.8313
Cross-section random effects test equation:
Dependent Variable: LOG(INOUT?)
Method: Panel Least Squares
Date: 04/18/16 Time: 17:44
Sample: 2005 2014
Included observations: 10
Cross-sections included: 8
Total pool (balanced) observations: 80
Variable Coefficient Std. Error t-Statistic Prob.
C -15.52331 3.641055 -4.263410 0.0001
LOG(RAILWAY?) 0.836952 0.309178 2.707021 0.0085
LOG(RIVER?) 1.085986 0.512786 2.117814 0.0378
LOG(ROAD?) 1.261898 0.190249 6.632873 0.0000
Effects Specification
Cross-section fixed (dummy variables)
R-squared 0.899850 Mean dependent var 13.33125
Adjusted R-squared 0.885336 S.D. dependent var 1.283176
S.E. of regression 0.434510 Akaike info criterion 1.297884
Sum squared resid 13.02712 Schwarz criterion 1.625413
Log likelihood -40.91536 Hannan-Quinn criter. 1.429200
F-statistic 61.99700 Durbin-Watson stat 0.842235
Prob(F-statistic) 0.000000
这是检验结果,P大于0.05,应该是不能拒绝原假设,可以使用随机模型,但是可以用固定模型么?
随机模型结果:
Dependent Variable: LOG(INOUT?)
Method: Pooled EGLS (Cross-section random effects)
Date: 04/18/16 Time: 17:57
Sample: 2005 2014
Included observations: 10
Cross-sections included: 8
Total pool (balanced) observations: 80
Swamy and Arora estimator of component variances
Cross sections without valid observations dropped
Variable Coefficient Std. Error t-Statistic Prob.
C -8.944554 2.368351 -3.776701 0.0003
LOG(RAILWAY?) 0.852806 0.296641 2.874876 0.0052
LOG(RIVER?) 0.159457 0.240396 0.663308 0.5091
LOG(ROAD?) 1.266622 0.188952 6.703415 0.0000
Random Effects (Cross)
CQ--C 1.119734
SC--C -0.241240
GZ--C -0.962989
YN--C -0.473777
SX--C -0.132017
GS--C -0.152410
QH--C -0.890927
NX--C 1.733625
Effects Specification
S.D. Rho
Cross-section random 0.924737 0.8191
Idiosyncratic random 0.434510 0.1809
Weighted Statistics
R-squared 0.572209 Mean dependent var 1.959341
Adjusted R-squared 0.555323 S.D. dependent var 0.669546
S.E. of regression 0.446481 Sum squared resid 15.15023
F-statistic 33.88563 Durbin-Watson stat 0.725608
Prob(F-statistic) 0.000000
Unweighted Statistics
R-squared 0.388217 Mean dependent var 13.33125
Sum squared resid 79.57875 Durbin-Watson stat 0.138142
固定模型结果:
Dependent Variable: LOG(INOUT?)
Method: Pooled Least Squares
Date: 04/18/16 Time: 17:58
Sample: 2005 2014
Included observations: 10
Cross-sections included: 8
Total pool (balanced) observations: 80
Cross sections without valid observations dropped
Variable Coefficient Std. Error t-Statistic Prob.
C -15.52331 3.641055 -4.263410 0.0001
LOG(RAILWAY?) 0.836952 0.309178 2.707021 0.0085
LOG(RIVER?) 1.085986 0.512786 2.117814 0.0378
LOG(ROAD?) 1.261898 0.190249 6.632873 0.0000
Fixed Effects (Cross)
CQ--C 0.146552
SC--C -2.079066
GZ--C -1.778154
YN--C -1.101244
SX--C 0.168929
GS--C 0.307718
QH--C 0.278118
NX--C 4.057147
Effects Specification
Cross-section fixed (dummy variables)
R-squared 0.899850 Mean dependent var 13.33125
Adjusted R-squared 0.885336 S.D. dependent var 1.283176
S.E. of regression 0.434510 Akaike info criterion 1.297884
Sum squared resid 13.02712 Schwarz criterion 1.625413
Log likelihood -40.91536 Hannan-Quinn criter. 1.429200
F-statistic 61.99700 Durbin-Watson stat 0.842235
Prob(F-statistic) 0.000000
感觉固定模型结果看起来要好的多,怎么办?求问!!