marcel spijkerman <[email protected]>:
Sounds like a case for discrete-time survival analysis to me
("failure" or "death" being end of one job), but you cannot use
characteristics of the new job... as you say you do not observe all
potential choices at all points in time, in which case you would have
a very interesting and complex competing risks framework, or even at
one point in time for a choice problem (-asclogit-, -asmprobit-,
etc.);
read for starters:
http://www.iser.essex.ac.uk/iser/teaching/module-ec968
On Sun, May 10, 2009 at 1:18 PM, marcel spijkerman
<[email protected]> wrote:
> Dear Stata users,
>
> I am currently working on job transitions. I have paneldata of labour market positions of individuals. Some of them enter a job from a position of unemployment, others have a job but change to another job, and the rest keeps the same job from t to t+1. My interest is in the influence of both personal characteristics as well as job characteristics on these transitions. I would expect that for those who change their job characteristics of their former job as well as their new job have an effect on the probability of a transition. For example one would expect that a wage improvement has a positve effect on making such a transition. The the point is that for those who do not change their job, characteristics of their potential new jobs are not observed. I am not sure how to model this problem. Does anyone have suggestions?
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/