1
$\begingroup$

I am working in a VAR model for an economic research. I have been reading about this but there is something i still don't get. Based in the books I have read:

1) if I have 2 time series, both are stationary in level, I can use VAR

2) if I have 2 time series with trend but they are co integrated I can use VECM with this series.

3) if I have 2 time series with trend but they are not co integrated, then I must use VAR with differences

This is based in:

Loría, E. (2007). Econometría con aplicaciones. Editorial Pearson Prentice Hall. México.

Gujarati, D. N. D. N. (1992). Econometría. McGraw-Hill,.

You can also see that in this book: here

This was so far what i knew, before i start to use some software to fit my model. Every book i have been reading about VAR never tell nothing about the inclusion of a "trend" or when to include it, and then i found this

https://onlinecourses.science.psu.edu/stat510/node/79

This site tell me that de-trend a series is not longer necessary, you can fit a VAR with no stationary time series inside the model, using the "trend" argument you can find in some software.

my questions are: How to use the trend inside a VAR model? Using a VAR with trend, when i have co integration, is better than a VECM?

$\endgroup$
  • $\begingroup$ a VAR with deterministic trend is different from a VAR with stochastic trend, which is again different than a VECM. But yes, all of these are nonstationary. $\endgroup$ – Taylor Nov 29 '17 at 16:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.