st: re: how to run zandrews as a postestimation command

[Kit Baum <baum@bc.edu>]

Purba said

I would like to test for the structural stability of an estimation of
the form y=a+bx+error.
Specifically want to test using the "zandrews" command if "b" which is
the coefficient of x, changes overtime and exactly where the breakpoints
are. The data is 40 years time series.

Can anyone please help. It seems like the "zandrews" command only allows
me to select one variable to test in this case I can pick either y or x
but there seems to be no method to specify the actual equation that I am
estimating.


zandrews is a unit root test that allows for structural breaks. Unit  
root tests are applied to individual time series (or panels of  
individual time series). You could, of course, apply it to the  
residuals of a regression that you have run, but that does not sound  
like what you're trying to do. If you want to test for structural  
stability of a relationship, then you should use a test for  
structural stability. -qll- will do that, but it will not identify  
where breaks occur (only if the relationship appears to violate the  
null of structural stability).

To test for breaks in a y on x regression, it is usual to compute  
split-sample estimates (moving the breakpoint through, for instance,  
the interior 85% of the sample), allowing for a break in every  
feasible period and keeping track of which period registers the  
lowest sum of squared errors. See the code for  -clemio1-, which does  
something like that.

Kit Baum
author of zandrews, qll, clemao1

Kit Baum, Boston College Economics and DIW Berlin
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


*
*   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/