xtgls urb tin tie ucs pcg,p(c)
Cross-sectional time-series FGLS regression
Coefficients: generalized least squares
Panels: heteroskedastic with cross-sectional correlation
Correlation: no autocorrelation
Estimated covariances = 21 Number of obs = 96
Estimated autocorrelations = 0 Number of groups = 6
Estimated coefficients = 5 Time periods = 16
Wald chi2(4) = 2087.28
Prob > chi2 = 0.0000
------------------------------------------------------------------------------
urb | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
tin | .3019681 .0529322 5.70 0.000 .1982229 .4057133
tie | .073784 .0065263 11.31 0.000 .0609927 .0865754
ucs | -.0541428 .0086998 -6.22 0.000 -.0711941 -.0370914
pcg | .1899971 .0069925 27.17 0.000 .176292 .2037023
_cons | .737825 .1720074 4.29 0.000 .4006966 1.074953
------------------------------------------------------------------------------
前面分析中发现面板数据存在自相关和异方差,所以选用了xtgls,有几个问题:
1、xtgls前要加xdata去除个体效应吗
2、上面这个结果有问题吗?原本有的变量是不显著的,执行xtgls后,全都是显著了,这是怎么回事呢?
3、面板数据选用随机模型或者固定模型,执行xtgls后,只是说明了用FGLS,没有说用什么模型
4、xtgls的结果中没有R-sq值,怎样看模型合适还是不合适呢?
谢谢