陈强老师的《高级计量经济学及stata应用(第二版)》第261页中说道,
“LSDV法的stata命令为
reg y x1 x2 x3 i.id,r
其中,选择项”r"表示使用聚类稳健标准误;如果使用选择项“vce(cluster id)” 也能达到完全相同的效果。“
与我实验的结果相左:
书上用vce(cluster id)的部分结果:
. reg fatal beertax spircons unrate perinck i.state,vce(cluster state)
Linear regression Number of obs = 336
F( 3, 47) = .
Prob > F = .
R-squared = 0.9359
Root MSE = .15679
(Std. Err. adjusted for 48 clusters in state)
------------------------------------------------------------------------------
| Robust
fatal | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
beertax | -.4840728 .2395323 -2.02 0.049 -.9659495 -.002196
spircons | .8169652 .1373903 5.95 0.000 .5405716 1.093359
unrate | -.0290499 .0102108 -2.85 0.007 -.0495914 -.0085085
perinck | .1047103 .0368628 2.84 0.007 .0305519 .1788687
我用r的结果:
. reg fatal beertax spircons unrate perinck i.state,r
Linear regression Number of obs = 336
F( 51, 284) = 109.23
Prob > F = 0.0000
R-squared = 0.9359
Root MSE = .15679
------------------------------------------------------------------------------
| Robust
fatal | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
beertax | -.4840728 .1646223 -2.94 0.004 -.8081075 -.160038
spircons | .8169652 .1151614 7.09 0.000 .5902869 1.043643
unrate | -.0290499 .0083051 -3.50 0.001 -.0453973 -.0127025
perinck | .1047103 .0249763 4.19 0.000 .0555482 .1538724
显然标准误不同,并且右上方给出的F统计量的自由度也不一样。。。
请问怎么解释?
谢谢!


雷达卡





京公网安备 11010802022788号







