Cointegration test in VCEM-X
Cointegration test in VCEM-X
Dear Tom,
As I am working on the project that have related to co-integration tests. I am wondering RATS does support for Johansen test in the presence of exogenous variables, say VAR-X. To be specific, I have four I(1) variables (monthly data) and want to test their co-integrations while keeping oil price, US T-bill rate as exogenous forces. Could I do that?
Thanks in advance.
As I am working on the project that have related to co-integration tests. I am wondering RATS does support for Johansen test in the presence of exogenous variables, say VAR-X. To be specific, I have four I(1) variables (monthly data) and want to test their co-integrations while keeping oil price, US T-bill rate as exogenous forces. Could I do that?
Thanks in advance.
Re: Cointegration test in VCEM-X
That's one of the things that CATS is designed to do. However, what exactly do you mean by "exogenous"? One assumes that those are I(1) variables---are you including them in the cointegrating vector for other four variables?
Re: Cointegration test in VCEM-X
Dear Tom,
Thank for reply. The exogenous var could be I(1) or I(0) but I do not want to include it as an endogenous variable. Let say oil price or US T-bill rate is exogenous w.r.t to domestic I(1) vars, hence, I want to add those vars as doing so in VAR-X system. Is is feasible with CATS? And could you refer me a reference for that test?!!
Thanks a lot.
Thank for reply. The exogenous var could be I(1) or I(0) but I do not want to include it as an endogenous variable. Let say oil price or US T-bill rate is exogenous w.r.t to domestic I(1) vars, hence, I want to add those vars as doing so in VAR-X system. Is is feasible with CATS? And could you refer me a reference for that test?!!
Thanks a lot.
Re: Cointegration test in VCEM-X
Juselius' The Cointegrated VAR Model has a section on long-run (weak) exogeneity, both testing for it (in the context of a full VECM), and imposing it (in a partial model). Note that if the exogenous series are I(1), you can't just throw them into a VECM as just "extra" variables---they either have to enter the cointegrating vector or be used in differences or you will have I(0) on the left and an I(1) variable on the right.