Hi,
I want to analize the transition from school to labour market into
account three exits: Work in a fully matched job with his
studies, work in a not matched job and unemployment. I would want to
make this estimation taking into account the unobserved heterogeneity
introducing a random intercept. I think the gllamm command from stata
could serve to me for this.
Follow de GLLAMM's Manual I wrote the model taking into account my
three alternatives:A fully matched job, not matched job and
unemployment.
In think, I have a three level model here: the first level are options
for each individual at each point in time, the combination of {ijt};
the second level are either individual and their options over time
{ij}; and the third level are individuals per se {i}.
The model that I wrote is:
qui by tid: gen alt=_n
gen chosen =alt==salida_AT
sort tid alt
tab alt, gen(a)
eq a1:a1
eq a2:a2
gllamm alt variables_x , i(tid id) lin(mlogit) expand(tid chosen m)
basecategory(3) family(binom) eqs(a1 a2) weight(wt)ip(f)nip(2)trace
tid is indentifying each individual in each alternative in each time.
id is identifiying clustered observations (persons) that belong to the
same alternative over time
Is this ok? How long it can spend running?
I have been running this for 3 weeks and the results are:
error in remcor
numerical derivatives are approximate
nearby values are missing
log likelihood = -13033.349
(not concave)
and results as always stata shows.
I don't know what means "error in remcor" and if the results are ok
when the program said (not concave).
Could someone help me to know if that estimation is correct?
Any help and suggestions would be very much appreciated.
Thanks,
Clara
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
