[R] VAR and VECM in multivariate time series

[cloris]

Hello to everyone!

I am working on my final year project about multivariate time series. There
are three variables in the multivariate time series model. 

I have a few questions:

1. I used acf and pacf plot and find my variables are nonstationary. But in
adf.test() and pp.test(), the data  are stationary. why?

2.I use VAR to get a model. y is the matrix of data set and I have made a
once difference of it to make it stationary.
 library(tsDyn)
VARselect(y,lag.max=20,type="const",season = NULL, exogen = NULL)
y1=VAR(y, p = 16, type = c("const"),
season = NULL, exogen = NULL, lag.max = NULL,ic = c("AIC"))
summary(y1)
plot(y1)

How can I get estimation of AIC in this model?


3. I also get a VECM model
v1=VECM(y, lag=16,beta=NULL, estim="ML")

what does ETC mean in the output?
and what is a number of cointegrating relationships? 

I want to make forecast by VECM.
j=ca.jo(y,K=16,type='trace',season = NULL)
j.var=vec2var(j)
predict(j.var,n.ahead=80)

Is this a correct way to predict VECM in R?


Could anyone help me? 
Thank you very much 



--
View this message in context: http://r.789695.n4.nabble.com/VAR-and-VECM-in-multivariate-time-series-tp3995951p3995951.html
Sent from the R help mailing list archive at Nabble.com.