Dummy Variables in Cointegration Framework

[anto2209]

Dear All,

I'd like to ask the community over the use of dummy variables in cointegration frameworks and subsequently in TAR and MTAR models.

It is quite usual finding some structural breaks when analysing commodity prices nowadays, e.g. due to the world crisis in 2007 and the rose of oil price in 2010, and more often avoiding their inclusion in the cointegration model may lead to misleading results supporting the absence of cointegration.

1. My first question goes for the 2-step Engle Granger cointegration test: What if one includes dummies (as exogenous variables of course) in the Y=a+bX+Dt+e (where Dt is the dummy)? Can one relies upon the Eview's results when testing for the presence of cointegration or do he/she has to take into consideration different Critical Values?

2. Regarding the Johansen method, Eviews gives the opportunity to include exogenous variables, as dummies, but what if one would include a dummy into the cointegrating space? Do he/she has to programme it? Because when an exogenous (dummy) variable is added, Eviews does not give any further options and the dummy is unrestricted.

3. Generally, if one could spot there are some (or one) structural break in the series but this do not affect the long-run relationship between the series (that is to say, series co-move also during the break and cointegration is found even w/o adding any dummies) could one say that the structural break is cancelled out in cointegration?

4. Finally, when estimating a TAR/M-TAR models in Eviews we can include dummies in the exogenous variables' box. Hence, if one included dummies in testing for cointegration and series resulted to be cointegrated, does she/he have to include dummies in the TAR/M-TAR specifications? I'm asking that because it happened to me that w/o including any dummies in the TAR specification, the output was largely different when including them. I had no cointegration and cointegration and asymmetric adjustment respectively.

Thank you in advance for any comment on this,

Bests

[dakila]

http://davegiles.blogspot.com/2011/05/c ... -hips.html

[anto2209]

Thank you Dakila for the interesting and useful link