Hi,
I have a problem with my Panel data.
I have an important variable "Env" that IS supposed to be significant (low p-value) but I noticed that there is quite a big difference between the random and fixed effect. Differences for the yearly dummy variables are (I believe) not a problem.
Env is the updated version data and Env2 is the "previous" version data. Both are on a scale of 0 to 100.
Env2 is significant in case of random effect and in case of fixed effect. I find it quite difficult to imagine that the slight differences between the two variables justifies that Env is not significant in the fixed effect case.
He's the "problematic" code
The "problematic" results
With Env2 (the outdated Env version)
Also something that I noticed, since I changed model with the updated data, a message about a problem related to the rank of the "differenced variance matrix" appeared. I did not have such a message before with the previous framework of "outdated" variables with Env2.
I have a problem with my Panel data.
I have an important variable "Env" that IS supposed to be significant (low p-value) but I noticed that there is quite a big difference between the random and fixed effect. Differences for the yearly dummy variables are (I believe) not a problem.
Env is the updated version data and Env2 is the "previous" version data. Both are on a scale of 0 to 100.
Env2 is significant in case of random effect and in case of fixed effect. I find it quite difficult to imagine that the slight differences between the two variables justifies that Env is not significant in the fixed effect case.
He's the "problematic" code
Code:
xtreg ROA Soc Env Y23 Y21 Y20 Y19 Y18 Y17 Y16 Y15 Y14 Y13, fe estimates store fixed xtreg ROA Soc Env Y23 Y21 Y20 Y19 Y18 Y17 Y16 Y15 Y14 Y13, re estimates store random hausman fixed random, sigmamore
Code:
. xtreg ROA Soc Env Y23 Y21 Y20 Y19 Y18 Y17 Y16 Y15 Y14 Y13, fe
Fixed-effects (within) regression Number of obs = 534
Group variable: company_ Number of groups = 89
R-squared: Obs per group:
Within = 0.0903 min = 1
Between = 0.0116 avg = 6.0
Overall = 0.0028 max = 10
F(12,433) = 3.58
corr(u_i, Xb) = -0.3980 Prob > F = 0.0000
------------------------------------------------------------------------------
ROA | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
Soc | -.1950134 .0420047 -4.64 0.000 -.277572 -.1124549
Env | .0555025 .037416 1.48 0.139 -.018037 .129042
Y23 | 2.487787 1.645604 1.51 0.131 -.7465792 5.722152
Y21 | 2.30636 .9572937 2.41 0.016 .4248395 4.18788
Y20 | -.589413 1.001768 -0.59 0.557 -2.558346 1.37952
Y19 | .3879365 1.141829 0.34 0.734 -1.856279 2.632152
Y18 | 2.255327 1.277438 1.77 0.078 -.255424 4.766078
Y17 | 1.223604 1.395741 0.88 0.381 -1.519666 3.966875
Y16 | .6098173 1.505278 0.41 0.686 -2.348742 3.568377
Y15 | -.7039242 1.54012 -0.46 0.648 -3.730966 2.323117
Y14 | -.4803801 1.61304 -0.30 0.766 -3.650743 2.689982
Y13 | -.2086145 1.877213 -0.11 0.912 -3.898198 3.480969
_cons | 13.53602 2.58561 5.24 0.000 8.454115 18.61793
-------------+----------------------------------------------------------------
sigma_u | 12.144308
sigma_e | 6.2508888
rho | .79055495 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(88, 433) = 7.54 Prob > F = 0.0000
. estimates store fixed
. xtreg ROA Soc Env Y23 Y21 Y20 Y19 Y18 Y17 Y16 Y15 Y14 Y13, re
Random-effects GLS regression Number of obs = 534
Group variable: company_ Number of groups = 89
R-squared: Obs per group:
Within = 0.0786 min = 1
Between = 0.0125 avg = 6.0
Overall = 0.0284 max = 10
Wald chi2(12) = 37.66
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0002
------------------------------------------------------------------------------
ROA | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
Soc | -.1443823 .0368393 -3.92 0.000 -.216586 -.0721785
Env | .0937023 .0334626 2.80 0.005 .0281168 .1592878
Y23 | 2.700422 1.64934 1.64 0.102 -.5322246 5.933068
Y21 | 2.652062 .960349 2.76 0.006 .769813 4.534312
Y20 | .2326752 .9933623 0.23 0.815 -1.714279 2.17963
Y19 | 1.658275 1.112127 1.49 0.136 -.5214548 3.838005
Y18 | 3.929181 1.22406 3.21 0.001 1.530067 6.328295
Y17 | 2.894581 1.349148 2.15 0.032 .2502986 5.538864
Y16 | 2.43456 1.446509 1.68 0.092 -.4005449 5.269666
Y15 | 1.233097 1.474467 0.84 0.403 -1.656805 4.122999
Y14 | 1.816727 1.519329 1.20 0.232 -1.161102 4.794557
Y13 | 2.215353 1.788924 1.24 0.216 -1.290873 5.721579
_cons | 6.133317 2.175926 2.82 0.005 1.868579 10.39805
-------------+----------------------------------------------------------------
sigma_u | 10.100888
sigma_e | 6.2508888
rho | .72308163 (fraction of variance due to u_i)
------------------------------------------------------------------------------
. estimates store random
. hausman fixed random, sigmamore
Note: the rank of the differenced variance matrix (11) does not equal the number of coefficients being tested (12); 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))
| fixed random Difference Std. err.
-------------+----------------------------------------------------------------
Soc | -.1950134 -.1443823 -.0506312 .0208556
Env | .0555025 .0937023 -.0381998 .0173834
Y23 | 2.487787 2.700422 -.2126351 .1737672
Y21 | 2.30636 2.652062 -.3457026 .0923275
Y20 | -.589413 .2326752 -.8220883 .1803371
Y19 | .3879365 1.658275 -1.270339 .2956484
Y18 | 2.255327 3.929181 -1.673854 .3989196
Y17 | 1.223604 2.894581 -1.670977 .39808
Y16 | .6098173 2.43456 -1.824743 .4572045
Y15 | -.7039242 1.233097 -1.937021 .4849187
Y14 | -.4803801 1.816727 -2.297108 .5782523
Y13 | -.2086145 2.215353 -2.423968 .6156221
------------------------------------------------------------------------------
b = Consistent under H0 and Ha; obtained from xtreg.
B = Inconsistent under Ha, efficient under H0; obtained from xtreg.
Test of H0: Difference in coefficients not systematic
chi2(11) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 26.73
Prob > chi2 = 0.0050
(V_b-V_B is not positive definite)
Code:
. xtreg ROA Soc Env2 Y23 Y21 Y20 Y19 Y18 Y17 Y16 Y15 Y14 Y13, fe
Fixed-effects (within) regression Number of obs = 509
Group variable: company_ Number of groups = 86
R-squared: Obs per group:
Within = 0.1215 min = 1
Between = 0.0002 avg = 5.9
Overall = 0.0126 max = 10
F(12,411) = 4.74
corr(u_i, Xb) = -0.1587 Prob > F = 0.0000
------------------------------------------------------------------------------
ROA | Coefficient Std. err. t P>|t| [95% conf. interval]
-------------+----------------------------------------------------------------
Soc | .0596746 .0257027 2.32 0.021 .0091494 .1101997
Env2 | -.0759133 .0201294 -3.77 0.000 -.1154827 -.0363439
Y23 | 1.714019 .9464357 1.81 0.071 -.1464398 3.574477
Y21 | .724673 .5702726 1.27 0.205 -.3963419 1.845688
Y20 | -1.065076 .5963272 -1.79 0.075 -2.237308 .1071562
Y19 | .9091219 .6810253 1.33 0.183 -.4296055 2.247849
Y18 | 2.49774 .747119 3.34 0.001 1.029089 3.966392
Y17 | 1.110816 .8117859 1.37 0.172 -.4849544 2.706586
Y16 | 1.461078 .8683571 1.68 0.093 -.245897 3.168054
Y15 | .1247239 .8978083 0.14 0.890 -1.640145 1.889593
Y14 | 1.0386 .9380706 1.11 0.269 -.8054144 2.882615
Y13 | 1.747345 1.109523 1.57 0.116 -.4337018 3.928392
_cons | 6.742863 1.587189 4.25 0.000 3.622843 9.862884
-------------+----------------------------------------------------------------
sigma_u | 8.5014054
sigma_e | 3.5629637
rho | .85059528 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(85, 411) = 14.08 Prob > F = 0.0000
. estimates store fixed
. xtreg ROA Soc Env2 Y23 Y21 Y20 Y19 Y18 Y17 Y16 Y15 Y14 Y13, re
Random-effects GLS regression Number of obs = 509
Group variable: company_ Number of groups = 86
R-squared: Obs per group:
Within = 0.1193 min = 1
Between = 0.0068 avg = 5.9
Overall = 0.0230 max = 10
Wald chi2(12) = 55.41
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
------------------------------------------------------------------------------
ROA | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
Soc | .0626816 .0231504 2.71 0.007 .0173076 .1080556
Env2 | -.0600636 .0191302 -3.14 0.002 -.097558 -.0225692
Y23 | 1.771838 .9485326 1.87 0.062 -.0872514 3.630928
Y21 | .9011142 .5714768 1.58 0.115 -.2189598 2.021188
Y20 | -.8542544 .5940272 -1.44 0.150 -2.018526 .3100174
Y19 | 1.231007 .6671197 1.85 0.065 -.0765231 2.538538
Y18 | 2.931363 .7259445 4.04 0.000 1.508538 4.354188
Y17 | 1.562656 .7944344 1.97 0.049 .0055932 3.119719
Y16 | 1.880968 .8433974 2.23 0.026 .2279397 3.533997
Y15 | .6088329 .8679521 0.70 0.483 -1.092322 2.309988
Y14 | 1.57318 .8958133 1.76 0.079 -.1825821 3.328941
Y13 | 2.304711 1.069608 2.15 0.031 .2083169 4.401105
_cons | 4.824359 1.466826 3.29 0.001 1.949433 7.699286
-------------+----------------------------------------------------------------
sigma_u | 7.6336945
sigma_e | 3.5629637
rho | .82112064 (fraction of variance due to u_i)
------------------------------------------------------------------------------
. estimates store random
. hausman fixed random, sigmamore
Note: the rank of the differenced variance matrix (11) does not equal the number of coefficients being tested (12); 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))
| fixed random Difference Std. err.
-------------+----------------------------------------------------------------
Soc | .0596746 .0626816 -.003007 .011512
Env2 | -.0759133 -.0600636 -.0158497 .006636
Y23 | 1.714019 1.771838 -.0578196 .0815824
Y21 | .724673 .9011142 -.1764412 .049841
Y20 | -1.065076 -.8542544 -.2108213 .0834126
Y19 | .9091219 1.231007 -.3218855 .1557256
Y18 | 2.49774 2.931363 -.4336225 .194461
Y17 | 1.110816 1.562656 -.4518402 .1889185
Y16 | 1.461078 1.880968 -.4198898 .2273164
Y15 | .1247239 .6088329 -.484109 .2495674
Y14 | 1.0386 1.57318 -.5345792 .2965421
Y13 | 1.747345 2.304711 -.5573657 .3187259
------------------------------------------------------------------------------
b = Consistent under H0 and Ha; obtained from xtreg.
B = Inconsistent under Ha, efficient under H0; obtained from xtreg.
Test of H0: Difference in coefficients not systematic
chi2(11) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 22.09
Prob > chi2 = 0.0237
(V_b-V_B is not positive definite)
Comment