Re: st: RE: fractional probit

[Austin Nichols <austinnichols@gmail.com>]

Arina Viseth <arina@udel.edu> :
As I said, I don't think you can use -fmlogit- with endog variables;
use Wooldridge's approach instead:
 http://www.stata.com/meeting/snasug08/abstracts.html#wooldridge

On Thu, Aug 26, 2010 at 11:25 AM, Arina Viseth <arina@udel.edu> wrote:
> Dear Martin, Austin,
>
> Thank you very much for your answers and comments. How should I then apply fmlogit when one explanatory variable is endogeneous - Could you please help?
>
> Thank you very much again,
>
> Arina
>
> ---- Original message ----
>>Date: Thu, 26 Aug 2010 10:59:33 -0400
>>From: owner-statalist@hsphsun2.harvard.edu (on behalf of Austin Nichols <austinnichols@gmail.com>)
>>Subject: Re: st: RE: fractional probit
>>To: statalist@hsphsun2.harvard.edu
>>
>>Martin Weiss <martin.weiss1@gmx.de>:
>>
>>Or Maarten, really:  I think I understand what -fmlogit- is doing now, and
>>the description as given is a lot better than my reading of it.
>>
>>I guess it would be hard to extend this approach to a panel setting
>>esp. with IV ;
>>in such cases one seems to be stuck estimating L-1 equations using
>>Wooldridge's approach separately for all but one of the proportions.
>>
>>On Thu, Aug 26, 2010 at 10:39 AM, Austin Nichols
>><austinnichols@gmail.com> wrote:
>>> Martin Weiss <martin.weiss1@gmx.de>:
>>>
>>> That says
>>> "fractional multinomial logit model"
>>> which is not the same thing--
>>> also I do not understand
>>> "It is a
>>>      multivariate generalization of the fractional logit model"
>>> --is that supposed to read:
>>> "It is a generalization of the fractional logit model
>>> for polychotomous outcome variables" or somesuch?
>>>
>>>
>>> On Thu, Aug 26, 2010 at 3:09 AM, Martin Weiss <martin.weiss1@gmx.de> wrote:
>>>>
>>>> <>
>>>>
>>>>
>>>> Maarten has published a command for fractional logit at -ssc d fmlogit-...
>>>>
>>>>
>>>> HTH
>>>> Martin
>>>>
>>>> -----Original Message-----
>>>> From: owner-statalist@hsphsun2.harvard.edu
>>>> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Arina Viseth
>>>> Sent: Donnerstag, 26. August 2010 03:44
>>>> To: statalist@hsphsun2.harvard.edu
>>>> Subject: st: fractional probit
>>>>
>>>> Dear Statalist,
>>>>
>>>> I am currently trying to apply the fractional probit model,  as in Papke and
>>>> Wooldridge (2008).
>>>> http://pages.stern.nyu.edu/~wgreene/Econometrics/Papke-Wooldridge-Fractional
>>>> Response.pdf
>>>>
>>>> My understanding is that, in order to control for unobserved heterogeneity,
>>>> time average of the independent variables are added to the regression, such
>>>> as for example:
>>>>
>>>> Yit = X1it + X2it + X3it + Cit + timeaverageof X1 + timeaverageof X2
>>>>
>>>> with
>>>> Y dependent variable
>>>> X1 and X2 endogenous independent variables
>>>> X3 exogenous independent variable
>>>> C vector of dummy variables
>>>>
>>>> I am looking for the STATA command for the fractional probit, is it the
>>>> following: glm y x1 x2 c ex1 ex2, family(binomial) link(probit) robust?
>>>> (with ex1 time average of X1, and ex2, time average of X2)
>>>>
>>>> Your comments will be very much appreciated. In advance, thank you very much
>>>> for your kind help.
>>>>
>>>> Arina
>>>

*
*   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/