1
$\begingroup$

I have used Stata to run a time series multiple regression. I know that there is in fact a structural break in the data and the point at which it occurs; therefore, I have estimated the regression over two different periods (one before the break and one after).

My question is;how do I prove (statistically) that there is a structural break in the data that requires this to be split up? Would this require an F test? If so, what are the steps to accomplish this in Stata?

$\endgroup$
  • $\begingroup$ Try searching chow test, unobserved components model(ucm) is a great alternative as a ell,SAS has an option to do automation level shift detection within ucm I'm not sure if stata has this option, $\endgroup$ – forecaster Jul 29 '15 at 13:42
2
$\begingroup$

As you guessed, you just need a Wald test. You can approach this problem in different ways: you can run multiple regressions on the split and combined samples and compute the test statistic by hand; or you can just use interaction and conduct a joint test for significance.

For details on how to implement this in Stata please refer to http://www.stata.com/support/faqs/statistics/computing-chow-statistic/ and http://www.stata.com/support/faqs/statistics/chow-tests/

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.