VECM representation for different deterministic models

[TomDoan]

It's not the job of software to give you the answers that you would like. It's the job of the software to give you the results to the analysis that you asked for. Robust standard errors don't fix the fact that the "error correction" mechanism breaks down in the periods where the circuit breakers are activated---robust standard errors take care of heteroscedasticity around a correct model, not a model which is (by law, in effect) incorrect.

In a BEKK model, there's no reason for B (in particular) to be non-zero and even one of the A's can be zero so a negative but insignificant value isn't impossible.

[John]

Tom,
How can I use @BDStest in RATS 9? When I run it got the following error: cp24. Procedure/function BDStest compiled with errors. Please correct.
should I install it first?

[TomDoan]

It's just

@bdstest series

[John]

Dear Tom, thank you very much for your reply.
Is there any way to define a VECM model in such a way that residuals of each equation (dependent series) to follow a univariate GARCH(1,1) model (or any ARCH process) with cross variable as an explanatory variable?

[TomDoan]

Didn't you already do a VECM with a GARCH error process? If you want to do the same thing with univariate GARCH processes, just do MV=DIAG on the GARCH instruction.

[John]

Dear Tom,

I would be grateful if you could possibly guide me how I can get standardized residuals fro a VECM model. I can get residuals from estimate procedure but I not sure if they are standardized or not.

[TomDoan]

For a GARCH model, it's the same as any other GARCH model. If it's not a GARCH model, then there really isn't the issue with standardization.

[John]

Thank you for your reply.
I want to perform a test on the residuals of VECM model and for doing this test standard deviation of these residuals should be equal to one. Do you mean that VECM residuals are standardized?

[TomDoan]

What test requires that? No, they aren't. But in the absence of GARCH, they are assumed to have a fixed covariance matrix.