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 at the end of May, and its replacement, statalist.org is already up and running.


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

st: RE: Endogenous interaction terms in 2SLS


From   "Wooldridge, Jeffrey" <wooldri1@msu.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Endogenous interaction terms in 2SLS
Date   Thu, 3 Mar 2011 07:21:04 -0500

Let me modestly (:-)) suggest the solution in Chapter 20 of 2e of my MIT Press book. It takes advantage of the binary nature of Y1 but is fully robust to the probit model being misspecified.

1. Estimate a reduced form probit for Y1 using ALL exogenous variables. Get the fitted probabilities, say phat.

2. Construct instruments (NOT regressors!) phat*X1, phat*X1*X2

3. Estimate the equation

Y2 = b0+ b1*Y1 +b2* Y1*X1 + b3*Y1*X1*X2

Using IVs phat, phat*X1, phat*X1*X2.

No need to adjust the standard errors asymptotically, but you can bootstrap if you have a smaller sample.

One comment: Many get nervous -- me included -- when variables that show up as interactions do not show up in level form. How come X1 and X2 don't appear by themselves (and act as their own IVs)? Leaving the level terms out can result in the interactions being spuriously significant.

JW

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Rene Algesheimer
Sent: Thursday, March 03, 2011 4:52 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: Endogenous interaction terms in 2SLS

Dear Statalisters,

I have a very specific Stata question that I wasn't able to solve with the Statalist. I would therefore highly appreciate to hear your advice on this.

I would like to estimate the following model: 

Y2 = b0+ b1*Y1 +b2* Y1*X1 + b3*Y1*X1*X2

Y1 is a dummy variable and endogenous,

X1 is another dummy variable and the instrument for Y1.

The situation therefore is similar to http://www.stata.com/statalist/archive/2008-10/msg01239.html with endogenous interaction terms. However, in my case one interaction is between the instrumented and the instrumental variable.

Any advice how to deal with this situation is highly appreciated.


Many, many thanks in advance.

best wishes,

René
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


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