聚类稳健固定和随机的比较-经管之家官网！

数据有异方差性，分别跑了
稳健固定：xtreg y1 lnm3 w2 w4 p r lniv lngdp lnpop, fe vce(cluster province)
得到：
Fixed-effects (within) regression Number of obs = 248
Group variable: province Number of groups = 31
R-sq:within= 0.5218 Obs per group: min = 8
between = 0.0015 avg = 8.0
overall = 0.0036 max = 8
F(8,30) = 2.37
corr(u_i, Xb)= -0.9747 Prob > F = 0.0415
(Std. Err. adjusted for 31 clusters in province)
------------------------------------------------------------------------------
| Robust
y1 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnm3 | 33.05835 28.3696 1.17 0.253 -24.88009 90.9968
w2 |-49.09019 38.627 -1.27 0.214 -127.977 29.79666
w4 | 4.193893 4.958539 0.85 0.404 -5.932795 14.32058
p | 30.01157 19.26801 1.56 0.130 -9.338968 69.3621
r | .033537 .0611479 0.55 0.587 -.0913437 .1584177
lniv | -113.336 94.01337 -1.21 0.237 -305.3369 78.66495
lngdp | 57.88043 106.3124 0.54 0.590 -159.2385 274.9994
lnpop | 1313.12 812.8167 1.62 0.117 -346.873 2973.113
_cons | -10865.2 6517.589 -1.67 0.106 -24175.89 2445.496
-------------+----------------------------------------------------------------
sigma_u |1360.3404
sigma_e |107.28264
rho |.99381883 (fraction of variance due to u_i)
------------------------------------------------------------------------------
再跑随机：xtreg y1 lnm3 w2 w4 p r lniv lngdp lnpop, re vce(cluster province)
得到：
Random-effects GLS regression Number of obs = 248
Group variable: province Number of groups = 31
R-sq:within= 0.4204 Obs per group: min = 8
between = 0.5005 avg = 8.0
overall = 0.4840 max = 8
Random effects u_i ~ Gaussian Wald chi2(8) = 38.42
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
(Std. Err. adjusted for 31 clusters in province)
------------------------------------------------------------------------------
| Robust
y1 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnm3 | 65.81309 28.6891 2.29 0.022 9.583495 122.0427
w2 |-8.410657 34.3611 -0.24 0.807 -75.75717 58.93586
w4 | 11.00109 8.93954 1.23 0.218 -6.520083 28.52227
p | 26.83362 17.88128 1.50 0.133 -8.213045 61.88029
r |-.0238832 .0370084 -0.65 0.519 -.0964183 .048652
lniv | 42.90071 41.44648 1.04 0.301 -38.3329 124.1343
lngdp |-108.3871 105.0597 -1.03 0.302 -314.3003 97.52615
lnpop |-175.6441 92.38208 -1.90 0.057 -356.7096 5.421457
_cons | 1743.48 1097.621 1.59 0.112 -407.817 3894.777
-------------+----------------------------------------------------------------
sigma_u |110.41051
sigma_e |107.28264
rho |.51436525 (fraction of variance due to u_i)
------------------------------------------------------------------------------
我该怎么选择，用固定，还是随机呢

「经管之家」APP：经管人学习、答疑、交友，就上经管之家！
免流量费下载资料----在经管之家app可以下载论坛上的所有资源，并且不额外收取下载高峰期的论坛币。
涵盖所有经管领域的优秀内容----覆盖经济、管理、金融投资、计量统计、数据分析、国贸、财会等专业的学习宝库，各类资料应有尽有。
来自五湖四海的经管达人----已经有上千万的经管人来到这里，你可以找到任何学科方向、有共同话题的朋友。
经管之家（原人大经济论坛），跨越高校的围墙，带你走进经管知识的新世界。
扫描下方二维码下载并注册APP