VECM With GARCH?

[Kahn]

Hello,

I have calculated a VECM model however I am now required to:

1) Re-estimating the VECM but with GARCH(1,1) errors and testing for asymmetries and conditional non-normalities
2) Again Re-Estimate but this time a GARCH with the mean equation consisting of a constant only

I've tried multiple searches and can't find any help with this. I have access to EViews 9.5 and would greatly appreciate any help!

[Kahn]

Hello Again,

After researching a bit more, some people have said that you may not be able to estimate such a model in EViews however my question now is that would it make sense to get one particular variable in the VECM system, say Delta R, and generate a new variable of Delta R but with GARCH errors, then re-estimate the VECM with this newly generated variable?

Sorry if there are mistakes in my understanding as I am new to this field and I am finding this particularly difficult.

[trubador]

That would lead to "generated regressors problem". You may try to migrate your VECM into the system framework and use the ARCH estimation. In order to do that, you should first identify and fix the long-run (cointegration) relationship via VEC and estimate all the remaining coefficients (including adjustment parameters, alphas) within the system. This is a two-step approach, which mitigates (not alleviates) the variance-covariance matrix problem.

[Kahn]

That would lead to "generated regressors problem". You may try to migrate your VECM into the system framework and use the ARCH estimation. In order to do that, you should first identify and fix the long-run (cointegration) relationship via VEC and estimate all the remaining coefficients (including adjustment parameters, alphas) within the system. This is a two-step approach, which mitigates (not alleviates) the variance-covariance matrix problem.

Thanks for your reply. I am actually testing the Expectations Hypothesis but found no cointegration. Despite this, I have been instructed to assume that there is cointegration and estimate a VECM for the long and short rates followed by re-estimation of the VECM equation for Delta R, where R is the long-term interest rate, but with the long-rate having GARCH errors. Could I have misinterpreted the question? Sorry if this sounds naive as I am very new to this.

[trubador]

I do not know the context or the research question here, but I think it would be safe to assume that GARCH errors have an impact on the short term dynamics.