xtreg y x1 x2, fe
y Coef. Std. Err. t
x1 2.450663 .1151893 21.28
x2 -5.677261 1.089288 -5.21
_cons -13.99025 1.777297 -7.87
xtserial y x1 x2
Wooldridge test for autocorrelation in panel data
H0: no first order autocorrelation
F( 1, 161) = 7.810
Prob > F = 0.0058
xttest3
Modified Wald test for groupwise heteroskedasticity
in fixed effect regression model
H0: sigma(i)^2 = sigma^2 for all i
chi2 (163) = 9.5e+05
Prob>chi2 = 0.0000
上述估計結果,代表有異方差與序列相關,因此進行以下修正,
xtreg y x1 x2, fe robust
Robust
y Coef. Std. Err. t
x1 2.450663 .366632 6.68
x2 -5.677261 1.857797 -3.06
_cons -13.99025 5.417574 -2.58
再檢測一次異方差,xttest3
Modified Wald test for groupwise heteroskedasticity
in fixed effect regression model
H0: sigma(i)^2 = sigma^2 for all i
chi2 (163) = 9.5e+05
Prob>chi2 = 0.0000
結果仍然是存在異方差,請問大家是否估計程序出了問題了呢?謝謝!


雷达卡


京公网安备 11010802022788号







