- library(car)
- library(plm)
- library(estimatr)
- Paneldata<-pdata.frame(Data,index=c("city","year"))
- ols1=lm_robust(lce~letp+lur,data=Paneldata)
- ols2=lm_robust(lce~letp+lur+ecs,data=Paneldata)
- ols3=lm_robust(lce~letp+lur+lecs,data=Paneldata)
- summary(ols1) #结果和stata相同
- summary(ols2) #加入非对数项后fdi in2 ecs后和stata出现差异
- summary(ols3) #加上对数后,结果与stata相同
lm_robust(formula = lce ~ letp + lur + ecs, data = Paneldata)
Standard error type: HC2
Coefficients:
Estimate Std. Error t value Pr(>|t|) CI Lower CI Upper DF
(Intercept) 4.054283 0.124363 32.600 9.140e-194 3.810415 4.298151 2435
letp 0.377539 0.012566 30.045 5.743e-169 0.352898 0.402180 2435
lur 0.249592 0.018549 13.456 7.261e-40 0.213219 0.285965 2435
ecs 0.005104 0.001295 3.942 8.309e-05 0.002565 0.007642 2435
Multiple R-squared: 0.5817 , Adjusted R-squared: 0.5812
F-statistic: 1348 on 3 and 2435 DF, p-value: < 2.2e-16
stata的回归结果
. reg lce letp lur ecs, r
Linear regression Number of obs = 2,439
F(3, 2435) = 1379.17
Prob > F = 0.0000
R-squared = 0.5811
Root MSE = .73403
------------------------------------------------------------------------------
| Robust
lce | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
letp | .3830517 .0125671 30.48 0.000 .3584084 .407695
lur | .2470695 .0187324 13.19 0.000 .2103364 .2838026
ecs | -.0044635 .0011671 -3.82 0.000 -.0067521 -.0021749
_cons | 4.859845 .1195206 40.66 0.000 4.625473 5.094218
------------------------------------------------------------------------------
其中ecs的正负都不一样,但在取对数后lecs的结果与Stata相同请问是什么问题导致?