Re: st: Calculate Marginal Effect after IVPROBIT

["Brian P. Poi" <bpoi@stata.com>]

On Mon, 9 Jul 2007 07:34:16 -0400 (EDT), Prabal Kumar De 
<prabal.de@nyu.edu> wrote:


> Hi!
>    I need to know what is the best way to calculate the marginal effect 
> after running IVPROBIT. I use -mfx- , but it is not reporting standard 
> errors ( even after specifying the value at which to be calculated). I 
> could not find something like -divprobit-
> Any help will be much appreciated.
>   Thanks,
>   Prabal
> ***************************************************
> Prabal Kr. De
> Department of Economics, New York University

When using Stata 9, you can safely use the "force" option with -mfx- after 
-ivprobit- to obtain the standard errors of the marginal effects:

   . webuse laborsup
   . ivprobit fem_work fem_educ kids (other_inc = male_educ)
   . mfx, predict(p) force


The maximum likelihood estimator fits the probit equation as well as the 
reduced-form equation for other_inc jointly.  By default, -mfx- will 
refuse to compute standard errors if a righthand side variable in one 
equation appears as the dependent variable in another equation of the 
model (as other_inc does in this case).  For most models, that is the 
appropriate thing to do.  However, for -ivprobit-, the probability of a 
positive outcome only depends on the coefficients of the probit equation, 
so it is safe to use the "force" option with -mfx-.

In Stata 10, we made -mfx- report the standard errors after -ivprobit- 
without your needing to specify the "force" option.

   -- Brian Poi
      bpoi@stata.com
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/