The Tobit and Heckman models estimate different things, so the resulting
predictions should differ. Tobit treats zero wage as zero while the Heckman
model considers zero wage as unobserved (as your own coding illustrates).
Therefore, the estimate from Heckman should be much larger than from Tobit
since it is estimating what the wage would have been if it were observed
(sort of).
Michael Blasnik
[email protected]
----- Original Message -----
From: "Takako Yuki" <[email protected]>
To: <[email protected]>
Sent: Monday, April 05, 2004 10:07 AM
Subject: st: comparison of "heckman" and "tobit" outputs
> Dear everyone:
>
> I would like to know, with STATA commands, whether Tobit TYPE-I and
> Type-II (i.e., heckman selection model) will provide the same estimation
> results if in the Type-II the regression equation has the same set of
> variables with the selection equation. For example, as described below,
> I compared the means of predicted wages with "tobit" command for TYPE I
> and "heckman" command for Type II. However, they provide the different
> results.
> I would appreciate if anyone can tell me: whether my understanding for
> the relation between Tobit TYPE I and TYPE II is wrong, or whether the
> following STATA commands are wrong for my purpose. If the latter is the
> case, could you also tell me how to do?
>
> .use http://stata-press.com/data/r8/womenwk
> (a) For Type I
> .gen wage1=0
> .replace wage1=wage if wage~=.
> .tobit wage1 educ age, ll(0)
> . predict yhatwage1
> . sum yhatwage1
>
> Variable | Obs Mean Std. Dev. Min Max
> -------------+--------------------------------------------------------
> yhatwage1 | 2000 12.70266 6.876089 -.4998228 34.74677
>
> (b) For Type II
> .heckman wage educ age, select (educ age)
> . predict yhatwage
> . sum wage yhatwage
> Variable | Obs Mean Std. Dev. Min Max
> -------------+--------------------------------------------------------
> wage | 1343 23.69217 6.305374 5.88497 45.80979
> yhatwage | 2000 20.93415 3.953325 14.20508 33.01097
>
>
> Thank you very much for any help.
>
> Takako Yuki
>
> *
> * 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/
*
* 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/
