Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: Heckman Procedure


From   Sergiy Radyakin <serjradyakin@gmail.com>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   Re: st: Heckman Procedure
Date   Mon, 10 Jun 2013 22:59:12 -0400

Annie,
Stata manual is very clear that "The selection equation should contain
at least one variable that is not in the outcome equation." So perhaps
you need to go back to the drawing board and decide HOW you want to
correct for selection. The problem is not in how to specify the
command to Stata, but how to model the phenomenon you study. Perhaps
people living closer to the cities are more likely to get financial
advice? Or those ones with radios? or ones that have gone through
bankruptcy procedure? Check what other information you can find. Or
wait for the guys in Europe to wake up in a couple of hours. Best,
Sergiy

On Mon, Jun 10, 2013 at 10:33 PM, Annie Zhang
<annieclairezhang@gmail.com> wrote:
> Hi Sergiy,
>
> Thanks so much for your reply. I don't have an instrument variable for
> the fa (financial advice) variable.
> The issue is that:
>
> Age, gender, wealth affects probability of receiving financial advice.
> Financial advice affects asset allocation (variable- cash).
>
> Therefore I need to correct for self-selection bias that receiving
> advice may have on asset allocation, given then certain age, gender,
> wealth affect probability of receiving advice.
>
> My main multivariate regression is:
>
> reg cash fa age sex logfum
>
> so I wanted to run heckman to correct for self-selection bias using:
>
> Heckman cash fa age sex logfum, select(fa=age sex logfum) twostep
>
> Since fa is being dropped, can you suggest anyway to run heckman that
> will work in my situation?
>
> Thank you so very much!!
>
> Kind Regards,
> Annie
>
>
>
> On Tue, Jun 11, 2013 at 2:27 PM, Sergiy Radyakin <serjradyakin@gmail.com> wrote:
>> Annie, Stata is doing the right thing as your variable fa is bringing
>> no additional explanatory power to the model since it is a linear
>> combination of your variable age sex and logfum from the first stage,
>> and thus the separate effect of it can't be identified. Recall that
>> the coefficient on a variable is the effect of change of that var
>> holding all other constant (ceteris paribus). Can your fa change when
>> age,sex, and logfum are fixed? Did you forget perhaps something like
>> an instrument? Best, Sergiy
>>
>> On Mon, Jun 10, 2013 at 10:05 PM, Annie Zhang
>> <annieclairezhang@gmail.com> wrote:
>>> Hi there,
>>>
>>> When using the heckman procedure I want to use the independent
>>> variable in the first stage as well as make it a dependent variable in
>>> the selection model, however stata will drop the variable due to
>>> collinearity each time.
>>>
>>> How can I use the heckman procedure in this case?
>>>
>>> My model is
>>>
>>>
>>> heckman cash fa age sex logfum, select(fa=age sex logfum) twostep
>>>
>>> And the 'fa' variable keeps dropping.
>>>
>>> Thanks for your help!
>>>
>>> Kind Regards,
>>> Annie
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>> *   http://www.ats.ucla.edu/stat/stata/
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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