An indicator variable

[RUBY JULY]

Hi everyone,

I have a dataset including over 500 firms that have CEO changes based on CEO_ID and lagged CEO_ID. Please see the following table to understand my dataset. I want to create an indicator variable that is equal to one for one year after the CEO changes and zero for one year before the CEO changes. Could you give me some suggestions on how I can create that indicator variable in this case? Thanks in advance for your help.

gvkey	fyear	execid	execid_n	lag1_execid	CEO_change
1004	2003	9251	9251	9252	-1
1004	2004	9252	9252	9251	1
1004	2005	23781	23781	9252	14529
1004	2006	9249	9249	23781	-14532
1004	2008	23781	23781	9249	14532
1004	2009	36199	36199	23781	12418
1004	2010	9249	9249	36199	-26950
1004	2011	41787	41787	9249	32538
1004	2012	33979	33979	41787	-7808
1004	2013	46404	46404	33979	12425
1004	2015	9249	9249	46404	-37155
1004	2016	51706	51706	9249	42457
1045	2002	3661	3661	3662	-1
1045	2004	16654	16654	3661	12993
1045	2007	28280	28280	16654	11626
1045	2008	14591	14591	28280	-13689
1045	2009	28280	28280	14591	13689
1045	2011	14591	14591	28280	-13689
1045	2015	46190	46190	46189	1
1045	2016	46193	46193	46190	3
1075	2004	19092	19092	10739	8353
1075	2005	10739	10739	19092	-8353
1075	2009	39817	39817	10739	29078
1075	2010	33218	33218	39817	-6599
1075	2011	12755	12755	33218	-20463
1075	2012	5835	5835	12755	-6920
1075	2013	39817	39817	5835	33982
1075	2014	12755	12755	39817	-27062
1078	2003	18569	18569	20862	-2293
1078	2004	20862	20862	18569	2293
1078	2005	24400	24400	20862	3538
1078	2006	14300	14300	24400	-10100

[Ken Chui]

Welcome to Statalist. Please read and follow the FAQ (https://www.statalist.org/forums/help) regarding posting data as Stata code (Section 12 has the details).

And back to the question, I am not sure how this new variable can logically work. If a company constantly changes CEO (which seems to be the case based on the sample data), then a cell in that new indicator will have to be 0 and 1 at the same time (it's a year after a change, and it's a year before the next change). Perhaps try manually inserting the new variable information in the sample data to better show the recoding logic.