Statalist


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

st: why predictnl calculates different values?


From   Sasha Shepotylo <shepotylo@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: why predictnl calculates different values?
Date   Mon, 27 Jul 2009 10:08:42 +0300

Dear Statalist users,

I have a panel 2000-2008 of some bilateral exports
l estimate the model by hausman-taylor:

 xhhtaylor ...

Then I predict

predictnl yhat1=exp(predict(xbu))

Then I change the value of some explanatory variable x but only for
years after 2003:

replace x=c if year>=2004

and predict again

predictnl yhat2=exp(predict(xbu)).

To my surprise, yhat2 is different from yhat1 for 2000-2003 periods.
How should I interpret these results?
My guess is that when I change some value of x in the future, it
changes the whole variance-covariance matrix and this is what drives
the results.
If it is true, does it have any economic interpretation?

Best,

OS

*
*   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   |   What's new   |   Site index