Re: st: Nonlinear ARMAX model

[Robert A Yaffee <bob.yaffee@nyu.edu>]

Sebastien,
   One of the assumptions of a transfer function model is that the input function is uncorrelated with the error of the arma noise model.  You 
are asking to model a situation where you have simultaneity bias.
   You should use instruments or proxy variables that are highly
correlated with your exogenous variable and uncorrelated with the 
error, lest you render your model vulnerable to feedback and the
bias in the estimation and parameters that can follow from it.
   If you had sufficient number of strong instruments to identify 
the model, then this might consistently be estimated with ivregress or
ivreg2. 
    -  Bob

    
   

Robert A. Yaffee, Ph.D.
Research Professor
Silver School of Social Work
New York University

Biosketch: http://homepages.nyu.edu/~ray1/Biosketch2009.pdf

CV:  http://homepages.nyu.edu/~ray1/vita.pdf

----- Original Message -----
From: Sebastian van Baal <s.vanbaal@arcor.de>
Date: Tuesday, March 9, 2010 11:56 am
Subject: Re: st: Nonlinear ARMAX model
To: statalist@hsphsun2.harvard.edu


> Thank you both for your suggestions! My case appears to be different. 
> I
> attempt to estimate a model such as this one: 
> 
> y[t]=b0+x[t]^b1+u[t]
> 
> u[t]=b2*u[t-1]+e[t]
> 
> My problem is more complicated and prevents me from linearizing it. 
> 
> In Eviews, I would enter something like this: 
> 
> y[t]=b0+x[t]^b1+u[t]+[AR(1)=b2]
> 
> Can I enter something like this in Stata?
> 
> Best regards
> Sebastian
> 
> *
> *   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/