where id= subject identification and employ=1 if the person is employed
and employ=2 if the person
is unemployed.
I want to keep in the data those subjects whose are unemployed in the
first observation (employ=2)
without losing their second observation if they are observed more than
once, that is: