Dummy variable in panel data

[Thijs Hofman]

Hi all,

I've got monthly panel data on stock split announcements from the years 1999 to 2017 for 600 companies. All companies have a GICS code (industry code) and based on their six-digit GICS code they have got peers (companies who are in the same industry/have the same GICS-code).

I want to make a dummy variable that takes the value of 1 when any of their peers have done a stock split announcement in the last 12 months and takes the value of 0 when none of their peers have done a stock split announcement. Do you have any idea how to obtain this dummy variable in STATA? Can you show me which kind of code to use?

Example of my data

T ; Company ; GICS ; Announcement
1-1999; 1 ; 0
2-1999; 1 ; 1
...

Help will be greatly appreciated.

[Carlo Lazzaro]

Thijs:
welcome to this forum:
Could it be:

bysort GISC: egen flag=total( Announcement)

[Thijs Hofman]

That doesn't give a dummy as a result...

[Igor Paploski]

Hi Thijs,

Please read the FAQ, specially regarding displaying a snippet of your dataset and using code delimiters.

Out of the 2 industries you displayed, both are from the same GICS (and you said in your question that GICS is a 6 digit variable, while in the example you sent it is a 1 character variable). Based on the 2 observations dataset you shared with us, I created extra observation for hypothetical GICS without any annoucements and run Carlo's command, it seemed to work:

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(T Company GICS Announcement)
1 1999 1 0
2 1999 1 1
3 1999 2 0
4 1999 2 0
end

bysort GICS: egen flag=total(Announcement)

browse

Both companies with GICS==1 were assigned flag==1 (because one of the had annoucement==1) and none of the 2 hypothetical GICS==2 were flagged - both had announcement==0.

Please share a portion of your dataset (use dataex to do so) so that people can see the structure of your data and test solutions.