Hello all,

For a large sample of firms, I need to count the number of (1) business segments and (2) geographic segments the business is in for the years 2010-2019. Using Ben Earnhart's code (https://www.statalist.org/forums/for...variable-group); bysort fyear CUSIP: gen howmany=_N gives me the number of segments for every firm in a particular year.

Now this is good, but I have to separate the business segments from the geographic segments.

For a particular firm, this is what it looks like:

SegmentType CompanyName fyear howmany
GEOSEG NEW SOURCE ENERGY PRTRS LP 2013 7
GEOSEG NEW SOURCE ENERGY PRTRS LP 2013 7
BUSSEG NEW SOURCE ENERGY PRTRS LP 2013 7
BUSSEG NEW SOURCE ENERGY PRTRS LP 2013 7
GEOSEG NEW SOURCE ENERGY PRTRS LP 2013 7
BUSSEG NEW SOURCE ENERGY PRTRS LP 2013 7
BUSSEG NEW SOURCE ENERGY PRTRS LP 2013 7
GEOSEG NEW SOURCE ENERGY PRTRS LP 2014 8
BUSSEG NEW SOURCE ENERGY PRTRS LP 2014 8
GEOSEG NEW SOURCE ENERGY PRTRS LP 2014 8
GEOSEG NEW SOURCE ENERGY PRTRS LP 2014 8
BUSSEG NEW SOURCE ENERGY PRTRS LP 2014 8
BUSSEG NEW SOURCE ENERGY PRTRS LP 2014 8
BUSSEG NEW SOURCE ENERGY PRTRS LP 2014 8
BUSSEG NEW SOURCE ENERGY PRTRS LP 2014 8

Anyone knows how to distinguish in the code between the two types of segments?