Peer Turnover Stata

[Maria de Wit]

Dear readers,

I am new on the forum, but I'm looking for your help. For my thesis I'am studying relative performance eveluation and peer replacement within a group, including some control variables. I use paneldata on the firms from the S&P 1500. All the firms in my subsample have peers (firms) which are used in relative performance evaluation benchmarking to determine executive compensation. I presented below a smaller sample of my full dataset. Three variables are shown: fiscalyear, cik (code of the company) and peercik (cik code of the peer firm)

Example generated by -dataex-. To install: ssc install dataex
clear
input double fiscalyear str12 cik double peercik
2006 "0000002488" 932787
2006 "0000002488" 703360
2006 "0000002488" 70530
2006 "0000002488" 1129446
2006 "0000002488" 97476
2006 "0000002488" 1272547
2006 "0000002488" 791915
2006 "0000002488" 50863
2007 "0000002488" 703360
2007 "0000002488" 791915
2007 "0000002488" 70530
2007 "0000002488" 50863
2007 "0000002488" 97476
2007 "0000002488" 932787
2007 "0000002488" 1272547
2015 "0000002969" 884905
2015 "0000002969" 915389
2015 "0000002969" .
2015 "0000002969" 313616
2015 "0000002969" 79879
2015 "0000002969" 30554
2015 "0000002969" 76334
2015 "0000002969" 31462
2015 "0000002969" 29905
2015 "0000002969" 1306830
2015 "0000002969" 1024478
2015 "0000002969" 1307954
2015 "0000002969" 49826
2014 "0000004127" 1045810
2014 "0000004127" 1097864
2014 "0000004127" 6281
2014 "0000004127" 827054
2014 "0000004127" 895419
2014 "0000004127" 763532
2014 "0000004127" 316793
2014 "0000004127" .
2014 "0000004127" 913885
2014 "0000004127" 743988
2014 "0000004127" 791915
2014 "0000004127" 743316
2014 "0000004127" 1054374
2014 "0000004127" 310568
2014 "0000004127" 1036960
2014 "0000004127" 791907
2014 "0000004127" 768251
2015 "0000004127" 310568
2015 "0000004127" 1272547
2015 "0000004127" .
2015 "0000004127" 1045810
2015 "0000004127" 743988
2015 "0000004127" 1054374
2015 "0000004127" 1097864
2015 "0000004127" 827054
2015 "0000004127" 743316
2015 "0000004127" 768251
2015 "0000004127" 763532
2015 "0000004127" 791907
2015 "0000004127" 6281
2015 "0000004127" 1036960
2015 "0000004127" 895419
2005 "0000004281" .
2005 "0000004281" 4285
2005 "0000004281" 1171264
2005 "0000004281" 1003467
2005 "0000004281" 889211
2005 "0000004281" 900268
2005 "0000004281" 863064
2006 "0000004281" 1171264
2006 "0000004281" 863064
2006 "0000004281" 889211
2006 "0000004281" .
2006 "0000004281" 900268
2006 "0000004281" 4285
2012 "0000004447" 717423
2012 "0000004447" 101778
2012 "0000004447" 1306965
2012 "0000004447" 93410
2012 "0000004447" 313807
2012 "0000004447" 821189
2012 "0000004447" 1140625
2012 "0000004447" 201283
2012 "0000004447" 1163165
2012 "0000004447" 797468
2012 "0000004447" 6769
2012 "0000004447" 34088
2012 "0000004447" 773910
2012 "0000004447" 1090012
2012 "0000004447" 879764
2013 "0000004447" 313807
2013 "0000004447" 879764
2013 "0000004447" 1163165
2013 "0000004447" 797468
2013 "0000004447" 1306965
2013 "0000004447" 773910
2013 "0000004447" 717423
2013 "0000004447" 93410
2013 "0000004447" 34088
2013 "0000004447" 1140625
2013 "0000004447" 101778
2013 "0000004447" 201283

As an example, for the fiscal year 2012, the company with cik identifier 0000004447 has 15 different peers (different peercik code). In 2013 the same company (cik=0000004447) has 12 different peers. It is possible dat a peer group do not change over the years, but I'am looking at those peergroups that do change.

Now, I would like to create two dependent variables indicating whether a peer firm is 'added' to the peer group or 'dropped'. I want to express both dependent variables as a dummy variable. For the added dummies this means that the dependent variable is equal to 1 if the peer firm is added to the peer group in fiscalyear X and 0 otherwise.
For the dependent variable dropped this means that the dependent variable equals 1 if the peer firm is excluded the next fiscalyear and 0 otherwise.

My question is, how would I compute this in STATA?

Thank you for your time.

[Daniel Bela]

Hi Maria,

I'm not sure if I fully understand what you're up to. Anyways, how about this?

clear
input double fiscalyear str12 cik double peercik
2006 "0000002488" 932787
2006 "0000002488" 703360
2006 "0000002488" 70530
2006 "0000002488" 1129446
2006 "0000002488" 97476
2006 "0000002488" 1272547
2006 "0000002488" 791915
2006 "0000002488" 50863
2007 "0000002488" 703360
2007 "0000002488" 791915
2007 "0000002488" 70530
2007 "0000002488" 50863
2007 "0000002488" 97476
2007 "0000002488" 932787
2007 "0000002488" 1272547
2015 "0000002969" 884905
2015 "0000002969" 915389
2015 "0000002969" .
2015 "0000002969" 313616
2015 "0000002969" 79879
2015 "0000002969" 30554
2015 "0000002969" 76334
2015 "0000002969" 31462
2015 "0000002969" 29905
2015 "0000002969" 1306830
2015 "0000002969" 1024478
2015 "0000002969" 1307954
2015 "0000002969" 49826
2014 "0000004127" 1045810
2014 "0000004127" 1097864
2014 "0000004127" 6281
2014 "0000004127" 827054
2014 "0000004127" 895419
2014 "0000004127" 763532
2014 "0000004127" 316793
2014 "0000004127" .
2014 "0000004127" 913885
2014 "0000004127" 743988
2014 "0000004127" 791915
2014 "0000004127" 743316
2014 "0000004127" 1054374
2014 "0000004127" 310568
2014 "0000004127" 1036960
2014 "0000004127" 791907
2014 "0000004127" 768251
2015 "0000004127" 310568
2015 "0000004127" 1272547
2015 "0000004127" .
2015 "0000004127" 1045810
2015 "0000004127" 743988
2015 "0000004127" 1054374
2015 "0000004127" 1097864
2015 "0000004127" 827054
2015 "0000004127" 743316
2015 "0000004127" 768251
2015 "0000004127" 763532
2015 "0000004127" 791907
2015 "0000004127" 6281
2015 "0000004127" 1036960
2015 "0000004127" 895419
2005 "0000004281" .
2005 "0000004281" 4285
2005 "0000004281" 1171264
2005 "0000004281" 1003467
2005 "0000004281" 889211
2005 "0000004281" 900268
2005 "0000004281" 863064
2006 "0000004281" 1171264
2006 "0000004281" 863064
2006 "0000004281" 889211
2006 "0000004281" .
2006 "0000004281" 900268
2006 "0000004281" 4285
2012 "0000004447" 717423
2012 "0000004447" 101778
2012 "0000004447" 1306965
2012 "0000004447" 93410
2012 "0000004447" 313807
2012 "0000004447" 821189
2012 "0000004447" 1140625
2012 "0000004447" 201283
2012 "0000004447" 1163165
2012 "0000004447" 797468
2012 "0000004447" 6769
2012 "0000004447" 34088
2012 "0000004447" 773910
2012 "0000004447" 1090012
2012 "0000004447" 879764
2013 "0000004447" 313807
2013 "0000004447" 879764
2013 "0000004447" 1163165
2013 "0000004447" 797468
2013 "0000004447" 1306965
2013 "0000004447" 773910
2013 "0000004447" 717423
2013 "0000004447" 93410
2013 "0000004447" 34088
2013 "0000004447" 1140625
2013 "0000004447" 101778
2013 "0000004447" 201283
end

* pre-save sort order
tempvar sortorder
generate `sortorder'=_n

* mark first occurence of peercik per cik
bysort cik peercik (fiscalyear) : generate added=(_n==1)

* mark last occurence of peercik per cik
bysort cik peercik (fiscalyear) : generate dropped=(_n==_N)

* re-sort in original sort order
sort `sortorder'

It generates a dummy added that is 1 for the first occurence of each peercik inside of a cik, 0 otherwise. And a dummy dropped that is 1 for the last occurence of each peercik inside of a cik, 0 otherwise.

Is that what you need?

Kind regards
Bela

PS: Please have a look at the FAQ, also linked at the top of each page, in order to learn how to post well readable data examples. Using -dataex- was a good start (thanks!), but your question will be even better when using [CODE] delimiters.

PPS: If you do not want added / dropped peercik's in the very first / last observed year per cik to be marked, add the following code before restoring the sort order:

* dis-mark first year per cik ("naturally" added==1)
bysort cik (fiscalyear peercik) : replace added=0 if added==1 & fiscalyear==fiscalyear[1]

* dis-mark last year per cik ("naturally" dropped==1)
bysort cik (fiscalyear peercik) : replace dropped=0 if dropped==1 & fiscalyear==fiscalyear[_N]