我的情况是
1. FE 回归结果最后一行如下
F test that all u_i=0: F(3118, 5764) = 0.92 Prob > F = 0.9945
因此OLS 优于FE
2. BREUSCH PAGAN LM 检验结果如下
Breusch and Pagan Lagrangian multiplier test for random effects
consumption_lag[hhid,t] = Xb + u[hhid] + e[hhid,t]
Estimated results:
Var sd = sqrt(Var)
---------+-----------------------------
consump~g 1.42e+07 3770.974
e 7356042 2712.202
u 1442900 1201.208
Test: Var(u) = 0
chi2(1) = 375.39
Prob > chi2 = 0.0000
因此RE 由于OLS
3. 按4楼的建议进行HAUSMAN 检验, 结果如下
. hausman FE RE
Note: the rank of the differenced variance matrix (4) does not equal the number of
coefficients being tested (5); be sure this is what you expect, or there may be
problems computing the test. Examine the output of your estimators for anything
unexpected and possibly consider scaling your variables so that the coefficients
are on a similar scale.
---- Coefficients ----
(b) (B) (b-B) sqrt(diag(V_b-V_B))
FE RE Difference S.E.
income_lag .3848475 .4055527 -.0207053 .003818
income_ins~g .0661347 -.0134085 .0795432 .008158
insurance_~g -149.3701 -14.11473 -135.2554 45.22008
hhsize_lag 119.9776 120.7448 -.7671836 15.90935
hhage_lag 7.107392 5.392909 1.714483 2.090247
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(4) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 96.40
Prob>chi2 = 0.0000
H0 被拒绝, 应选FE.
但这是不是有点矛盾呢. 前两个实验结果显示 OLS 优于FE 以及 RE 优于OLS, 最后我却选了FE
多谢解答!