I am struggling with the specific understanding of panel regressions in connection with STATA. What I basically want to do is to replicate the following work (http://pages.stern.nyu.edu/~sternfin...c_html/sin.pdf), only for a different data sample.
This work mainly uses “pooled (panel) regressions” (p.24).
In my understanding, a pooled OLS regression in STATA is provided through the command reg or regress (which is completely the same). However, it does not seem that this approach takes the actual panel structure into account. Nevertheless, the researchers of the mentioned paper utilize exactly this term “pooled (panel) regressions” (p.24).
A different approach would be to define the panel first and then to provide a regression:
xtset firmID Time xtreg Y X1 X2 X3
This would take the panel structure into account, but it would not be a pooled regression anymore or? Moreover, the results are very similar, but the ordinary reg command accomplishes a higher level of significance.
I spent a lot of time and search to understand the main difference between the ordinary regresscommand and the xtreg command. Still if I am completely honest I do not understand the difference and would be thankful for a clear explanation. Also the help reg and help xtregcommands in STATA could not provide a clear understanding.
This work mainly uses “pooled (panel) regressions” (p.24).
In my understanding, a pooled OLS regression in STATA is provided through the command reg or regress (which is completely the same). However, it does not seem that this approach takes the actual panel structure into account. Nevertheless, the researchers of the mentioned paper utilize exactly this term “pooled (panel) regressions” (p.24).
A different approach would be to define the panel first and then to provide a regression:
xtset firmID Time xtreg Y X1 X2 X3
This would take the panel structure into account, but it would not be a pooled regression anymore or? Moreover, the results are very similar, but the ordinary reg command accomplishes a higher level of significance.
I spent a lot of time and search to understand the main difference between the ordinary regresscommand and the xtreg command. Still if I am completely honest I do not understand the difference and would be thankful for a clear explanation. Also the help reg and help xtregcommands in STATA could not provide a clear understanding.
Comment