依次加入变量进行双向固定效应回归，其余变量的系数和显著性数值没有变化是怎么回事

3971

收藏 2020-03-15

行业面板数据，豪斯曼检验是固定效应，则利用了双向固定效应回归。
一个一个加入变量回归，但是回归结果除了新加入的变量显著性和系数有变化，其余变量的系数和显著性与之前一模一样，没有变化。下图是两次回归结果，不同之处在于加入了lnmfdi 这个变量，但其余变量的系数和显著性数值两次回归居然一模一样，请各位大神看看，这样的结果是正常的吗？如果有问题，问题出在哪里呢？

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

全部回复

黃河泉

2020-3-15 18:02:59

1. 记得永远都应该修正标准误。2. 的确结果怪怪的，请 lnmfdi 与 lnreri (有改变) 有何关系？基本统计量与相关系数为何？3. 此外，你应该连时间固定效果之"所有" 系数都提供。

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

yc5005

2020-3-15 19:18:45

黃河泉发表于 2020-3-15 18:02
1. 记得永远都应该修正标准误。2. 的确结果怪怪的，请 lnmfdi 与 lnreri (有改变) 有何关系？基本统计量与 ...

感谢回复，mfdi是流入制造业的FDI；reri是实际汇率指数，二者的相关系数-0.2874；
前期分析觉得没什么问题，回归结果这一步感觉不对；
下面是固定时间效应回归的，与双向固定回归出现的问题一样
xtreg  tc lnfdi lnsval lnsex lnopen lnhc lnreri lnfca i.year
note: 2016.year omitted because of collinearity

Random-effects GLS regression                Number of obs    =       70
Group variable: sector                         Number of groups  =       7

R-sq:                                        Obs per group:
   within  = 0.7885                                        min =       10
   between = 0.9761                                        avg =    10.0
   overall = 0.9458                                        max =       10

                                             Wald chi2(15)    =    942.39
corr(u_i, X) = 0 (assumed)                   Prob > chi2    =    0.0000

------------------------------------------------------------------------------
      tc |    Coef. Std. Err.    z P>|z|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   lnfdi | .229015 .057219    4.00 0.000    .1168678 .3411623
   lnsval |  -1.298974 .2197859 -5.91 0.000 -1.729746 -.8682013
   lnsex | .9880101 .0606834 16.28 0.000    .8690728 1.106947
   lnopen |  -1.171934 .1098999 -10.66 0.000 -1.387334 -.9565342
      lnhc |  -.1923307 .2174353 -0.88 0.376 -.6184961 .2338347
   lnreri |  -.4442429 .7451448 -0.60 0.551    -1.9047 1.016214
   lnfca | .0237714 .0305539    0.78 0.437 -.0361131 .0836559
         |
      year |
   2008  | .0432641 .0603332    0.72 0.473 -.0749868    .161515
   2009  | .0417873 .0581541    0.72 0.472 -.0721927 .1557673
   2010  |    .08464 .0571903    1.48 0.139    -.027451 .1967309
   2011  | .1384309 .0616683    2.24 0.025    .0175633 .2592985
   2012  | .161334 .0686974    2.35 0.019    .0266895 .2959785
   2013  | .1778365 .0703985    2.53 0.012    .0398579    .315815
   2014  | .2092588 .0827374    2.53 0.011    .0470964 .3714211
   2015  | .1420621 .0688044    2.06 0.039    .007208 .2769162
   2016  |       0  (omitted)
         |
   _cons | 1.166234 1.593008    0.73 0.464 -1.956003 4.288472

xtreg  tc lnfdi lnmfdi lnsval lnsex lnopen lnhc lnreri lnfca i.year
note: 2015.year omitted because of collinearity
note: 2016.year omitted because of collinearity

Random-effects GLS regression                Number of obs    =       70
Group variable: sector                         Number of groups  =       7

R-sq:                                        Obs per group:
   within  = 0.7885                                        min =       10
   between = 0.9761                                        avg =    10.0
   overall = 0.9458                                        max =       10

                                             Wald chi2(15)    =    942.39
corr(u_i, X) = 0 (assumed)                   Prob > chi2    =    0.0000

------------------------------------------------------------------------------
      tc |    Coef. Std. Err.    z P>|z|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   lnfdi | .229015 .057219    4.00 0.000    .1168678 .3411623
   lnmfdi | 2.667561 1.291969    2.06 0.039    .1353474 5.199774
   lnsval |  -1.298974 .2197859 -5.91 0.000 -1.729746 -.8682013
   lnsex | .9880101 .0606834 16.28 0.000    .8690728 1.106947
   lnopen |  -1.171934 .1098999 -10.66 0.000 -1.387334 -.9565342
      lnhc |  -.1923307 .2174353 -0.88 0.376 -.6184961 .2338347
   lnreri | 1.161365 .7684396    1.51 0.131 -.3447488 2.667479
   lnfca | .0237714 .0305539    0.78 0.437 -.0361131 .0836559
         |
      year |
   2008  |  -.2178377 .1363356 -1.60 0.110 -.4850506 .0493752
   2009  |  -.1650804 .1081057 -1.53 0.127 -.3769638 .0468029
   2010  |  -.2401703 .1471202 -1.63 0.103 -.5285207    .04818
   2011  | -.281474 .1826144 -1.54 0.123 -.6393916 .0764436
   2012  | -.218524 .1579372 -1.38 0.166 -.5280753 .0910272
   2013  | -.134464 .1262179 -1.07 0.287 -.3818465 .1129185
   2014  | .0051898 .0827029    0.06 0.950 -.1569048 .1672845
   2015  |       0  (omitted)
   2016  |       0  (omitted)
         |
   _cons |  -9.152633 4.470429 -2.05 0.041 -17.91451 -.3907537

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

yc5005

2020-3-15 19:27:55

刚刚发的截图不完整，重发一下，完整的是：
xtreg  tc lnfdi lnsval lnsex lnopen lnhc lnreri lnfca i.year, fe
note: 2016.year omitted because of collinearity

Fixed-effects (within) regression             Number of obs    =       70
Group variable: sector                         Number of groups  =       7

R-sq:                                        Obs per group:
   within  = 0.8382                                        min =       10
   between = 0.7706                                        avg =    10.0
   overall = 0.7594                                        max =       10

                                             F(15,48)       =    16.58
corr(u_i, Xb)  = 0.4502                      Prob > F       =    0.0000

------------------------------------------------------------------------------
      tc |    Coef. Std. Err.    t P>|t|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   lnfdi | .2212469 .0734521    3.01 0.004    .0735615 .3689322
   lnsval |  -1.100336 .3438375 -3.20 0.002 -1.791667 -.4090041
   lnsex | .7807222 .076835 10.16 0.000    .6262351 .9352093
   lnopen |  -.8319578 .1236889 -6.73 0.000 -1.080651 -.5832645
      lnhc | .5407045 .2139182    2.53 0.015    .1105931 .9708159
   lnreri |  -1.387228 1.637498 -0.85 0.401 -4.679638 1.905183
   lnfca |  -.0438386 .0965434 -0.45 0.652 -.2379521 .1502749
         |
      year |
   2008  | .0481813 .0396919    1.21 0.231 -.0316246 .1279872
   2009  | .0440309 .0382081    1.15 0.255 -.0327917 .1208536
   2010  | .1208759 .053436    2.26 0.028    .0134355 .2283162
   2011  | .1796689 .0648751    2.77 0.008    .0492288    .310109
   2012  | .2223523 .0744507    2.99 0.004    .0726591 .3720454
   2013  | .1782195 .0683964    2.61 0.012    .0406994 .3157396
   2014  | .2206279 .0968469    2.28 0.027    .0259042 .4153516
   2015  | .1546934 .0511325    3.03 0.004    .0518846 .2575023
   2016  |       0  (omitted)
         |
   _cons | 1.876257 2.652185    0.71 0.483 -3.456319 7.208833
-------------+----------------------------------------------------------------
   sigma_u |  .25913844
   sigma_e |  .07817849
      rho |  .91657805 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(6, 48) = 13.54                   Prob > F = 0.0000

xtreg  tc lnfdi lnmfdi lnsval lnsex lnopen lnhc lnreri lnfca i.year, fe
note: 2015.year omitted because of collinearity
note: 2016.year omitted because of collinearity

Fixed-effects (within) regression             Number of obs    =       70
Group variable: sector                         Number of groups  =       7

R-sq:                                        Obs per group:
   within  = 0.8382                                        min =       10
   between = 0.7706                                        avg =    10.0
   overall = 0.7594                                        max =       10

                                             F(15,48)       =    16.58
corr(u_i, Xb)  = 0.4502                      Prob > F       =    0.0000

------------------------------------------------------------------------------
      tc |    Coef. Std. Err.    t P>|t|    [95% Conf. Interval]
-------------+----------------------------------------------------------------
   lnfdi | .2212469 .0734521    3.01 0.004    .0735615 .3689322
   lnmfdi | 2.904745 .9601372    3.03 0.004    .9742594    4.83523
   lnsval |  -1.100336 .3438375 -3.20 0.002 -1.791667 -.4090041
   lnsex | .7807222 .076835 10.16 0.000    .6262351 .9352093
   lnopen |  -.8319578 .1236889 -6.73 0.000 -1.080651 -.5832645
      lnhc | .5407045 .2139182    2.53 0.015    .1105931 .9708159
   lnreri | .3611417 1.422632    0.25 0.801 -2.499251 3.221534
   lnfca |  -.0438386 .0965434 -0.45 0.652 -.2379521 .1502749
         |
      year |
   2008  |  -.2361362 .097466 -2.42 0.019 -.4321046 -.0401677
   2009  |  -.1812302 .0767359 -2.36 0.022 -.3355181 -.0269424
   2010  |  -.2328147 .0992676 -2.35 0.023 -.4324055 -.0332238
   2011  |  -.2775715 .1237948 -2.24 0.030 -.5264776 -.0286654
   2012  |  -.1912805 .1071973 -1.78 0.081 -.4068151 .0242541
   2013  |  -.1618489 .0856662 -1.89 0.065 -.3340923 .0103944
   2014  |  -.0015856 .0746175 -0.02 0.983 -.1516141 .1484429
   2015  |       0  (omitted)
   2016  |       0  (omitted)
         |
   _cons |  -9.360103 3.199731 -2.93 0.005 -15.79359 -2.926613
-------------+----------------------------------------------------------------
   sigma_u |  .25913844
   sigma_e |  .07817849
      rho |  .91657805 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(6, 48) = 13.54                   Prob > F = 0.0000

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

Ariel999

2021-12-21 15:10:56

请问楼主解决这个问题了吗，我也遇到了同样的问题

扫码加我拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

扫码加我 拉你入群

扫码加我 拉你入群

扫码加我 拉你入群

扫码加我 拉你入群

扫码加我 拉你入群

分享

扫码加好友，拉您进群

扫码加我拉你入群

扫码加我拉你入群

扫码加我拉你入群

扫码加我拉你入群

扫码加我拉你入群