(1)tsset province year
(2)xtdes
(3)xi: xtreg lnincome lnmarket lngdpper lnpunished lnarea i.account,fe vce(cluster province)
i.account _Iaccount_0-1 (naturally coded; _Iaccount_0 omitted)
Fixed-effects (within) regression Number of obs = 396
Group variable: province Number of groups = 31
R-sq: within = 0.9110 Obs per group: min = 7
between = 0.8149 avg = 12.8
overall = 0.8514 max = 13
F(5,30) = 393.27
corr(u_i, Xb) = 0.0099 Prob > F = 0.0000
(Std. Err. adjusted for 31 clusters in province)
------------------------------------------------------------------------------
| Robust
lnincome | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnmarket | .482041 .1143536 4.22 0.000 .2484999 .7155822
lngdpper | 1.501299 .1172979 12.80 0.000 1.261744 1.740853
lnpunished | .0692428 .0912037 0.76 0.454 -.11702 .2555055
lnarea | .7239415 .0765667 9.46 0.000 .5675715 .8803116
_Iaccount_ .0811348 .1072505 0.76 0.455 -.1378999 .3001696
_cons | -5.88166 1.000968 -5.88 0.000 -7.92591 -3.83741
-------------+----------------------------------------------------------------
sigma_u | .64345609
sigma_e | .43293424
rho | .68837558 (fraction of variance due to u_i)
------------------------------------------------------------------------------
(4) xi: xtreg lnincome lnmarket lngdpper lnpunished lnarea i.account,re vce(cluster province)
i.account _Iaccount_0-1 (naturally coded; _Iaccount_0 omitted)
Random-effects GLS regression Number of obs = 396
Group variable: province Number of groups = 31
R-sq: within = 0.9089 Obs per group: min = 7
between = 0.8664 avg = 12.8
overall = 0.8770 max = 13
Wald chi2(5) = 1793.44
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
(Std. Err. adjusted for 31 clusters in province)
------------------------------------------------------------------------------
| Robust
lnincome | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnmarket | .5227059 .110567 4.73 0.000 .3059985 .7394132
lngdpper | 1.230361 .0949567 12.96 0.000 1.044249 1.416472
lnpunished | .0363822 .0884557 0.41 0.681 -.1369877 .2097521
lnarea | .8388615 .0678176 12.37 0.000 .7059414 .9717816
_Iaccount_1 | .1778101 .0981268 1.81 0.070 -.0145148 .370135
_cons | -4.262785 .9419984 -4.53 0.000 -6.109068 -2.416502
-------------+----------------------------------------------------------------
sigma_u | .37477717
sigma_e | .43293424
rho | .42836907 (fraction of variance due to u_i)
------------------------------------------------------------------------------
(5)estat ovtest
Ramsey RESET test using powers of the fitted values of lnincome
Ho: model has no omitted variables
F(3, 387) = 4.44
Prob > F = 0.0044
(6)estat hettest
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
Ho: Constant variance
Variables: fitted values of lnincome
chi2(1) = 12.36
Prob > chi2 = 0.0004
(7)qui xi: xtreg lnincome lnmarket lngdpper lnpunished lnarea i.account,fe
(8)est store fe
(9)qui xi: xtreg lnincome lnmarket lngdpper lnpunished lnarea i.account,re
(10)est store re
(11)hausman fe re,constant sigmamore
---- Coefficients ----
| (b) (B) (b-B) sqrt(diag(V_b-V_B))
| fe re Difference S.E.
-------------+----------------------------------------------------------------
lnmarket | .482041 .5227059 -.0406649 .02831
lngdpper | 1.501299 1.230361 .2709379 .066832
lnpunished | .0692428 .0363822 .0328606 .0112915
lnarea | .7239415 .8388615 -.11492 .0244186
_Iaccount_1 | .0811348 .1778101 -.0966753 .032294
_cons | -5.88166 -4.262785 -1.618875 .5430612
------------------------------------------------------------------------------
b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ha, efficient under Ho; obtained from xtreg
Test: Ho: difference in coefficients not systematic
chi2(6) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 38.32
Prob>chi2 = 0.0000
(V_b-V_B is not positive definite)
(去掉 constant 和sigmamore ,chi2值就为负的了)
(12)xi: xtgls lnincome lnmarket lngdpper lnpunished lnarea i.account,panels (hetero)
i.account _Iaccount_0-1 (naturally coded; _Iaccount_0 omitted)
Cross-sectional time-series FGLS regression
Coefficients: generalized least squares
Panels: heteroskedastic
Correlation: no autocorrelation
Estimated covariances = 31 Number of obs = 396
Estimated autocorrelations = 0 Number of groups = 31
Estimated coefficients = 6 Obs per group: min = 7
avg = 12.77419
max = 13
Wald chi2(5) = 7510.71
Prob > chi2 = 0.0000
------------------------------------------------------------------------------
lnincome | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnmarket | .4584226 .0540707 8.48 0.000 .352446 .5643991
lngdpper | .8573275 .0483444 17.73 0.000 .7625742 .9520808
lnpunished | -.1886504 .042592 -4.43 0.000 -.2721291 -.1051716
lnarea | 1.000523 .0247193 40.48 0.000 .9520746 1.048972
_Iaccount_1 | .2667665 .0550432 4.85 0.000 .1588839 .3746491
_cons | -2.265709 .3943075 -5.75 0.000 -3.038538 -1.492881
------------------------------------------------------------------------------
欢迎大牛们指出其中的问题和改进的建议,谢谢。