Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

st: Three simulateneous equations - one being logit

From   "Wiemann, Markus" <>
To   <>
Subject   st: Three simulateneous equations - one being logit
Date   Mon, 14 Feb 2011 10:12:21 +0100


I am trying to run a 3SLS (with seemingly unrelated regressions) in the
following form:
First stage: y1_binary x1 x2 x3 
Second stage: y2 x4 x5 x6 dummy_for_y1
Third stage: y3 x1 x4 x7 dummy_for_y1

I thought I might use - reg3 - to run these regressions. But I am not
sure whether I run into trouble because of the first stage being binary.
I did not find a way to define the binary outcome and hence use an
according model like logit/probit in the first stage. 

Is it indeed a problem to have a binary in outcome in -reg3-? Is there a
better model for my approach? Like, for example, using Inverse Mills'
Ratios from Heckman in a simultaneous equation model? 
Do I face the same problems using 2SLS?

I use Stata/SE 10.0

Thanks in advance

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index