Filter Regression for years

[Julius Schaefer]

Hello all,

for my Master's thesis, I am conducting a regression analysis with Stata.
My dependent variable is "sales_growth" and my independent variable is "patriotism". I am controlling for different variables like Number of Shareholder, Age, R&D etc.

Now the situation is that I have data from 2010-2020, but I want to limit everything to 2010-2016 without having to delete the data from 2017-2020.
What would the command be? xtreg............?

I am a complete amateur as far as Stata is concerned and would be very grateful for an answer!

Best regards
Julius

[Andrew Musau]

help if

xtreg y x1...xn if year<=2016, fe cluster(companyid)

[Julius Schaefer]

Do I need
fe cluster(companyid) ? I want to check the growth of all S&P500 firms

[Maxence Morlet]

So you have panel data right? You're probably going to want to resort to two way fixed effects, so in #2 before the comma you may want to add i.time, and yes as Andrew suggested you do need fe cluster(companyid) to perform fixed effects estimation and get unbiased estimates of standard errors. Unless you want to cluster at a higher level.

[Julius Schaefer]

so what would the command be? Befehl war:
xtreg sales_growth Patriotism sum_rvalue_20_fsup nemp rnd nshareholders sf NationalityMix age if year<=2016 ?

[Maxence Morlet]

xtreg sales_growth Patriotism sum_rvalue_20_fsup nemp rnd nshareholders sf NationalityMix age i.time if year<2017, fe cluster([cluster_var])

If any of your regressors is time-invariant, it will be dropped. Does patriotism vary with time?

Also you may want to model age as a quadratic, otherwise it might drop as well due to collinearity.

xtreg sales_growth Patriotism sum_rvalue_20_fsup nemp rnd nshareholders sf NationalityMix age age_squared i.time if year<2017, fe cluster([cluster_var])

[Julius Schaefer]

Yes, Patriotism varies with time.
Essentially, I just want to observe the period from 2010-2016. Nothing more. And instead of deleting all my data from 17-20. i'm looking for a command that just tests the period 10-16

[Julius Schaefer]

Also, please excuse my stupidity, I am a complete amateur when it comes to statistics.

[Maxence Morlet]

No worries, everyone starts somewhere, no one was blessed with the knowledge of statistics at birth.

Try running the command in #6

[Julius Schaefer]

what does that mean?

[Maxence Morlet]

That means I recommend that you open Stata and paste this into the command box with your data loaded (replace clustervar with whatever variable you want to use for clustering):

xtreg sales_growth Patriotism sum_rvalue_20_fsup nemp rnd nshareholders sf NationalityMix age age_squared i.time if year<2017, fe cluster([cluster_var])