Quandt-Andrews Breakpoint Test
Posted: Mon Dec 03, 2012 5:32 am
Hey guys, at first: thanks a lot for this great forum, you guys already helped me a LOT.
I have a question concerning the Quandt-Andrews Breakpoint test, especially concerning the selection of the "Breakpoint variables: Enter the variables you with to vary across breakpoints". The QA-Test, as far as I'm concerned, consists of multiple Chow-Tests, and takes the one with the highest F-statistic. But the Chow-Test (in a bivariate case) just allows for a change in the slope, not in the level (which would be expressed in an intercept-change). But using the QA-Test I can also choose to vary the "c"-Variable (= intercept) across breakpoints, not just the independent variable (as I said, I have a bivariate example). So how is the QA-Test going to implement the intercept-change in the multiple Chow-Tests? Does it just add another intercept multiplied with an indicatorvariable to allow for an intercept change (like the gregory-hansen cointegration test does)?
Thanks a lot in advance.
Jerry
I have a question concerning the Quandt-Andrews Breakpoint test, especially concerning the selection of the "Breakpoint variables: Enter the variables you with to vary across breakpoints". The QA-Test, as far as I'm concerned, consists of multiple Chow-Tests, and takes the one with the highest F-statistic. But the Chow-Test (in a bivariate case) just allows for a change in the slope, not in the level (which would be expressed in an intercept-change). But using the QA-Test I can also choose to vary the "c"-Variable (= intercept) across breakpoints, not just the independent variable (as I said, I have a bivariate example). So how is the QA-Test going to implement the intercept-change in the multiple Chow-Tests? Does it just add another intercept multiplied with an indicatorvariable to allow for an intercept change (like the gregory-hansen cointegration test does)?
Thanks a lot in advance.
Jerry