Re: st: marginal effects for ivprobit

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

On Sun, 28 Jan 2007, Ramani Gunatilaka wrote:

> Hello,
> I am trying to obtain marginal effects after ivprobit.
> My problem is related to the discussion between  Adrienne and Brian
> Poi in the Stata Archives (Dec4, 2005).
> This is my code:
<snip>
> I tried using
>
> ivprobit hapiv_2  /*income variables*/ (ln_pcy02 =f_edyrs prodassets)
> age agesq /*individual characteristics*/ male married divorced widowed
> ethmin edyrs unempld workhrs netfinassets r_health, first twostep
> mfx compute, predict (p) eq(hapiv_2)
> estimates store hap1
> overid, depvar(hapiv_2)
>
> as Brian suggested to Adrienne. But Stata said:
>
> .
> . mfx compute, predict(p) eq(hapiv_2)
> probabilities not available with two-step estimator
Ramani,

In short, if you use the two-step estimator, then you cannot use the -mfx- 
command because of a scaling issue that arises with the two-step estimator 
(but not with the maximum likelihood estimator).  Wooldridge (2002, 
Econometric Analysis of Cross Section and Panel Data) has a nice 
discussion of this model on pp. 472-477, and he shows how you can obtain 
average partial effects after the two-step estimator.

  -- 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/