[R-SIG-Finance] Johansen Procedure for VAR question

[Matthieu Stigler]

Manuel Serna a écrit :
> Good Morning,
>
> I'm using vars package to do a Johansen Procedure for VAR, and I'm
> interested in using with K=1, but the next error comes up
>
>   
>> ca.jo(Canada, type ="eigen", ecdet ="const", K = 1,
>>     
> + spec="longrun", season = NULL, dumvar = NULL)
> Error en ca.jo(Canada, type = "eigen", ecdet = "const", K = 1, spec =
> "longrun",  :
>
> K must be at least K=2.
>
> I will be really thankful if you can answer me the reason of this.
>
> Thanks,
>
>   
I suppose you want to estimate a VECM model with one lag and the ECM?

Note that the parameter K corresponds to the number of lags in the model 
in levels (VAR), and the VECM is in difference. Hence, to have one lag 
in differences you need two lags in levels! A model with one lag in 
level would correspond to no lags in the VECM formulation.

To make sure, look at the help function and compare the lags.

Beside of that, note actually that the specification with 
spec=transitory is also often seen and may correspond to the usual VECM 
model (Hamilton 1994, p 580).
Mat
> Manuel Serna Cortés
> Universidad Icesi
> Colombia
>
> 	[[alternative HTML version deleted]]
>
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> R-SIG-Finance at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only.
> -- If you want to post, subscribe first.