# Re: CointegrationMLE

Home 21090308 Forums Re: CointegrationMLE

#2035
osigrtoelt
Member

In fact, since there’s a linear causal relationship here, VAR(1) should probably be the “correct” model to use.
For cointegration, we require p>= 2. Otherwise, there’s no delta(y_{t-i}) term in the VECM and we’re essentially doing VAR(1) anyway.