[Limdep Nlogit List] Selection with simultaneous equations

[Stephane Mahuteau]

Hi everyone,

My question is related to paragraph E23.2.5 of vol2 Econometric Modeling guide (Limdep 8.0) entitled Simultaneous Equations Models with Selectivity.
In this paragraph are shown the commands for estimating such models implying a Probit selection equation.
My question is the following: can one use the same sort of method when the selection is based on a multinomial logit? Should I write as many SELECT commands as there are choices in the multinomial logit and equations in the simultaneous equations model? I am a bit lost there, any help would be much appreciated.

I have another problem for which I am a bit at a lost regarding the right technique to use. Basically I have these durations of migration in a country (fortunately all complete spells, no censoring) and I am trying to disentangle two theories. One says that the labour force status (not in labour force, unemployed,wage earner and self employed) after return in origin country influences the duration of migration (basically migrants decide their duration depending on what they want to do once back). In this theory people who want to become self employed upon return are supposed to experience the shortest spells of migration. Another theory would say that the labour force status in the host country conditions the duration of stay. Successful migrants should stay longer in the host country. 
So far I had a look at the first theory with a model similar to Lee (1983), that is selection based on a multinomial logit (labour force status after return to origin country) and then estimation of the durations. I looked at the second theory by using the same sort of model except that the selection was on the labour force status in the host country.
I get some results but I don't think it is satisfactory since in the first model I don't account for the potential selection due to the success in the host country and in the second model I don't take into account the same selection due to the intended occupation upon return.
If anybody has any idea on how to implement a model on Limdep that would incorporate both selection, I'd be very grateful.
Thank you very much,

Stephane