Dear community,
I have encountered an issue with xtdcce2 (xtdcce2 4 - 08.02.2023). I have a balanced panel dataset with N=116 and T=22, 2552 observations in total. Somehow, only 30 obs are used in the regression. I get the following output from running a simple command:
************************************************** ************************************************** **************
. xtdcce2 ln_CO2 ln_N ln_x ln_x_2 ln_x_3 ln_epcff urban_pop trade_share treat, nocross reportc
(Dynamic) Common Correlated Effects Estimator - Mean Group
Panel Variable (i): cid Number of obs = 30
Time Variable (t): year Number of groups = 116
Degrees of freedom per group: Obs per group (T) = 0
without cross-sectional averages = -8.7397176
with cross-sectional averages = -8.7397176
Number of F(1025058.375, 1508)=170757.84
cross-sectional lags none Prob > F = 0.00
variables in mean group regression = 1025058.4 R-squared = -55.20
variables partialled out = 0 R-squared (MG) = 1.00
Root MSE = 0.04
CD Statistic = 1.70
p-value = 0.0890
ln_CO2 Coef. Std. Err. z P>z [95% Conf. Interval]
Mean Group:
ln_N 1.092561 .4976889 2.20 0.028 .1171085 2.068013
ln_x 2.801099 2.275498 1.23 0.218 -1.658795 7.260993
ln_x_2 -.3599327 .4323254 -0.83 0.405 -1.207275 .4874095
ln_x_3 .0174324 .0234778 0.74 0.458 -.0285832 .063448
ln_epcff .4504235 .0372784 12.08 0.000 .3773592 .5234879
urban_pop -.0114534 .0243626 -0.47 0.638 -.0592032 .0362963
trade_share 1.098167 .6406565 1.71 0.087 -.1574965 2.353831
treat -.0006374 .0016545 -0.39 0.700 -.0038801 .0026054
_cons -4.391728 3.390281 -1.30 0.195 -11.03656 2.253101
Mean Group Variables: ln_N ln_x ln_x_2 ln_x_3 ln_epcff urban_pop trade_share treat _cons
Warning:
Collinearities detected. One or more variables are dropped and set to zero.
Use estat ebistructure to display more details.
************************************************** ************************************************** **************
//Here's a reghdfe command and its output for reference.
************************************************** ************************************************** **************
. reghdfe ln_CO2 ln_N ln_x ln_x_2 ln_x_3 ln_epcff urban_pop trade_share treat, absorb(cid year) vce(cluster cid)
(MWFE estimator converged in 2 iterations)
HDFE Linear regression Number of obs = 2,552
Absorbing 2 HDFE groups F( 8, 115) = 117.53
Statistics robust to heteroskedasticity Prob > F = 0.0000
R-squared = 0.9966
Adj R-squared = 0.9964
Within R-sq. = 0.6188
Number of clusters (cid) = 116 Root MSE = 0.0908
(Std. err. adjusted for 116 clusters in cid)
Robust
ln_CO2 Coefficient std. err. t P>t [95% conf. interval]
ln_N .7633596 .1000825 7.63 0.000 .5651154 .9616038
ln_x -5.091496 2.110266 -2.41 0.017 -9.271528 -.9114644
ln_x_2 .5766992 .2410469 2.39 0.018 .0992317 1.054167
ln_x_3 -.0205965 .0089854 -2.29 0.024 -.0383949 -.0027981
ln_epcff .3015928 .04724 6.38 0.000 .2080195 .3951661
urban_pop .0021092 .0036395 0.58 0.563 -.0051001 .0093184
trade_share .0306852 .0095561 3.21 0.002 .0117565 .049614
treat -.0562171 .0213541 -2.63 0.010 -.0985154 -.0139188
_cons 20.15335 6.086037 3.31 0.001 8.098083 32.20862
Absorbed degrees of freedom:
Absorbed FE Categories - Redundant = Num. Coefs
-
cid 116 116 0 *
year 22 0 22
* = FE nested within cluster; treated as redundant for DoF computation
************************************************** ************************************************** **************
I have no trouble running the regressions in the help file using sample dataset. I have properly xtset before running the regression and I have no idea why this happens. Can you tell what went wrong? Please help.
I have encountered an issue with xtdcce2 (xtdcce2 4 - 08.02.2023). I have a balanced panel dataset with N=116 and T=22, 2552 observations in total. Somehow, only 30 obs are used in the regression. I get the following output from running a simple command:
************************************************** ************************************************** **************
. xtdcce2 ln_CO2 ln_N ln_x ln_x_2 ln_x_3 ln_epcff urban_pop trade_share treat, nocross reportc
(Dynamic) Common Correlated Effects Estimator - Mean Group
Panel Variable (i): cid Number of obs = 30
Time Variable (t): year Number of groups = 116
Degrees of freedom per group: Obs per group (T) = 0
without cross-sectional averages = -8.7397176
with cross-sectional averages = -8.7397176
Number of F(1025058.375, 1508)=170757.84
cross-sectional lags none Prob > F = 0.00
variables in mean group regression = 1025058.4 R-squared = -55.20
variables partialled out = 0 R-squared (MG) = 1.00
Root MSE = 0.04
CD Statistic = 1.70
p-value = 0.0890
ln_CO2 Coef. Std. Err. z P>z [95% Conf. Interval]
Mean Group:
ln_N 1.092561 .4976889 2.20 0.028 .1171085 2.068013
ln_x 2.801099 2.275498 1.23 0.218 -1.658795 7.260993
ln_x_2 -.3599327 .4323254 -0.83 0.405 -1.207275 .4874095
ln_x_3 .0174324 .0234778 0.74 0.458 -.0285832 .063448
ln_epcff .4504235 .0372784 12.08 0.000 .3773592 .5234879
urban_pop -.0114534 .0243626 -0.47 0.638 -.0592032 .0362963
trade_share 1.098167 .6406565 1.71 0.087 -.1574965 2.353831
treat -.0006374 .0016545 -0.39 0.700 -.0038801 .0026054
_cons -4.391728 3.390281 -1.30 0.195 -11.03656 2.253101
Mean Group Variables: ln_N ln_x ln_x_2 ln_x_3 ln_epcff urban_pop trade_share treat _cons
Warning:
Collinearities detected. One or more variables are dropped and set to zero.
Use estat ebistructure to display more details.
************************************************** ************************************************** **************
//Here's a reghdfe command and its output for reference.
************************************************** ************************************************** **************
. reghdfe ln_CO2 ln_N ln_x ln_x_2 ln_x_3 ln_epcff urban_pop trade_share treat, absorb(cid year) vce(cluster cid)
(MWFE estimator converged in 2 iterations)
HDFE Linear regression Number of obs = 2,552
Absorbing 2 HDFE groups F( 8, 115) = 117.53
Statistics robust to heteroskedasticity Prob > F = 0.0000
R-squared = 0.9966
Adj R-squared = 0.9964
Within R-sq. = 0.6188
Number of clusters (cid) = 116 Root MSE = 0.0908
(Std. err. adjusted for 116 clusters in cid)
Robust
ln_CO2 Coefficient std. err. t P>t [95% conf. interval]
ln_N .7633596 .1000825 7.63 0.000 .5651154 .9616038
ln_x -5.091496 2.110266 -2.41 0.017 -9.271528 -.9114644
ln_x_2 .5766992 .2410469 2.39 0.018 .0992317 1.054167
ln_x_3 -.0205965 .0089854 -2.29 0.024 -.0383949 -.0027981
ln_epcff .3015928 .04724 6.38 0.000 .2080195 .3951661
urban_pop .0021092 .0036395 0.58 0.563 -.0051001 .0093184
trade_share .0306852 .0095561 3.21 0.002 .0117565 .049614
treat -.0562171 .0213541 -2.63 0.010 -.0985154 -.0139188
_cons 20.15335 6.086037 3.31 0.001 8.098083 32.20862
Absorbed degrees of freedom:
Absorbed FE Categories - Redundant = Num. Coefs
-
cid 116 116 0 *
year 22 0 22
* = FE nested within cluster; treated as redundant for DoF computation
************************************************** ************************************************** **************
I have no trouble running the regressions in the help file using sample dataset. I have properly xtset before running the regression and I have no idea why this happens. Can you tell what went wrong? Please help.
