ARDL Cointegration Bounds test with exogenous variables
Posted: Tue Jan 31, 2017 5:05 am
by Lasse
Hi! I noticed that the ARDL Bounds test to detect cointegration introduces exogenous variables (such as dummies) in a way that may lead to spurrious cointegration results. The bounds test does not support dummies in the long-run equation, or s o I'm told.
Re: ARDL Cointegration Bounds test with exogenous variables
Posted: Fri Feb 03, 2017 11:57 am
by EViews Gareth
Could you provide more details?
Re: ARDL Cointegration Bounds test with exogenous variables
Posted: Sat Feb 04, 2017 2:15 am
by Lasse
I only referr to others with greater knowledge than me, for instance Kripganz and Schneider, who made the ARDL cointegration module for Stata. In the attached file you can see their ppt on this. Look at page 16-17 in their ppt, and notice that their exogenous variables (dummies) only goes into the short-run part of the equation. Ofc it could also be them who misunderstood, but they are not the only ones making that mistake in that case. If you think of it it makes sense though, since the bounds test has different critical values for different cointegration cases (restricted constant, unrestricted constant etc). For instance, if a step dummy is included as a long-run coefficient, it should influence the critical values, right? I also provide the original papers by Pesaran et al (1999 and 2001). I'm looking forward to your response.