Delete firms from panel data having no observation

[Kaleem Khan]

Dear Statlists

I am working on R&D investment of Panel data of more than 2900 firms and study period is 2005-2015. Problem is that most of the firms have not R&D observations. I want to keep only those firms which have at least on year observation of R&D, and all the remaining one I want to delete form sample. For example, company 2 has no observation for R&D, so it must be deleted. how can I do that?

input int(year id) RDinv
2004 2 .
2005 2 .
2006 2 .
2007 2 .
2008 2 .
2009 2 .
2010 2 .
2011 2 .
2012 2 .
2013 2 .
2014 2 .
2015 2 .
2004 3 .
2005 3 .
2006 3 .
2007 3 .
2008 3 .
2009 3 .
2010 3 5767031.08
2011 3 2952345.13
2012 3 2460888.19
2013 3 1936084.84
2014 3 2439346.8
2015 3 4265023.32

[Carlo Lazzaro]

Kaleem:
you can try:

drop if RDinv==.

However, although tecnically feasible, please note that Stata would do it by defaut, by applying listwise deletion to all the observations with at least one missing value in any variable.

[Kaleem Khan]

Dear Carlo

Thank you so much for guidance. I applied it but it deletes all missing R&D observations. I want to delete only those firms which don't have any R&D observations.

Like for this dataex

input int(year id) RDinv
2004 2 .
2005 2 .
2006 2 .
2007 2 .
2008 2 .
2009 2 .
2010 2 .
2011 2 .
2012 2 .
2013 2 .
2014 2 .
2015 2 .
2004 3 .
2005 3 .
2006 3 .
2007 3 .
2008 3 .
2009 3 .
2010 3 5767031.08
2011 3 2952345.13
2012 3 2460888.19
2013 3 1936084.84
2014 3 2439346.8
2015 3 4265023.32


The solution must be like this:

input int(year id) RDinv
2004 3 .
2005 3 .
2006 3 .
2007 3 .
2008 3 .
2009 3 .
2010 3 5767031.08
2011 3 2952345.13
2012 3 2460888.19
2013 3 1936084.84
2014 3 2439346.8
2015 3 4265023.32

is it possible?

[Carlo Lazzaro]

Kaleem:
surely there are more elegant solutions but the one that springs to my mind at present is the following one:

bysort id: egen flag=total( RDinv)
drop if flag==0

[Nick Cox]

bysort id (RDinv) : drop if missing(RDinv[1]) & missing(RDinv[_N])

[Kaleem Khan]

Thank you so much Carlo and Nick. It worked.