Panel data

[Kerri Agnew]

I have a data set which includes firm id's and a year variable, e.g.

id Year

1 2004
1 2005
1 2006
2 2006
2 2007
3 2005

etc


Is there a way I can calculate the numbers of new id's (firms) each year, and the number of id's that leave each year?

Thank you for your help

[Harald Leber]

Quick but rough solution: set panel: tsset id Year, and simply create two dummy variables as follows: bysort year: gen missinglag = (l1.id == .); bysort year: gen missingforward = (f1.id == .) ; count them by year: bysort year: egen count... = total(missing...)

[Nick Cox]

Harald: Missing L1. identifiers occur at the beginning of each panel, but also if there are any gaps within panels.

I would do something like this:

bysort id (year) : gen flag = cond(_n == 1, 1, cond(_n == _N, -1, 0))

or equivalently

bysort id (year) : gen flag = (_n == 1) - (_n == _N)

Then the numbers concerned are

egen entry = total(flag == 1), by(year)
egen exit = total(flag == -1), by(year)

You may want to select just one value for display

 
egen tag = tag(year) 
sort year 
l year entry exit if tag