Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: biprobit with partial observability


From   Robert Duval <rduval@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: biprobit with partial observability
Date   Mon, 23 Nov 2009 08:20:42 -0600

It is also very unlikely that your problem will converge if you don't
impose exclusion restrictions, between the determinants of each
equation. Furthermore, without those you won't be able to identify the
equation of leaving from the one of returning to the workforce.

Hope this helps
robert

> Date: Sun, 22 Nov 2009 10:04:51 -0800 (PST)
> From: Celia Vera <veracelia@ymail.com>
> Subject: st: biprobit with partial observability
>
> Hi statalisters,
>
> I am estimating a bivariate probit for a two-stage process in which the second event is observed conditional on the outcome of the first one (the first event is leaving the workforce (left) and the second event is returning to the workforce (ret). So, return only happens if the person left the workforce).
> I understand i have to use a biprobit model with partial observability.
> Getting some help from stata faqs i have done so far:
> replace ret=0 if left=0
> gen leftret=  left*ret so leftret=1 only when left=1 and ret=1, 0 otherwise.
> gen y2=leftret.
> biprobit leftret y2 x1 x2...., partial
> but the likelihood function never converges (it is not concave) in all the iterations.
> Your help is greatly appreciated.
>
> Celia P.


> Date: Sun, 22 Nov 2009 11:46:00 -0800 (PST)
> From: Maarten buis <maartenbuis@yahoo.co.uk>
> Subject: Re: st: biprobit with partial observability
>
> - --- On Sun, 22/11/09, Celia Vera wrote:
>> I am estimating a bivariate probit for a two-stage process
>> in which the second event is observed conditional on the
>> outcome of the first one (the first event is leaving the
>> workforce (left) and the second event is returning to the
>> workforce (ret). So, return only happens if the person left
>> the workforce).
>> I understand i have to use a biprobit model with partial
>> observability.
>> but the likelihood function never converges (it is not
>> concave) in all the iterations.
>
> The help file warns that this model often won't converge
> and advises to use the -dificult- option. Alternatively
> you could look at -heckprob-.
>
> Hope this helps,
> Maarten
>
> - --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
> http://www.maartenbuis.nl
> - --------------------------
>
>

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index