VAR stability Test

1. VAR stability test
A professor from Algeria was asking whether the VAR model is stable or not?

J.R. Paladines commented> Just a stability test

Muhammad Khurram Shabbir commented> No unit root problem..

Younas Iqbal Qazi  coemmented> It shows that there is "satisfactory" stability as per
data inputs!

Keith Araneo-Yowell commented> It means you have a stationary stochastic process.
This output shows whether your model is stable. Since your roots are all inside the
complex unit circle your model will converge (the lagged variables' values won't have a
compounding effect on one another and the contemporary values will stabilize over time.

2. Can we say that model is stable as one dot is on the line?
Saeed Aas Khan Meo posted the figure.

Sayed Hossain commented> All dots should be inside the circle to have stable model.

Nicat Gasim For the stable model, all roots must be inside in the unit circle. If the roots
outside or on the unit circle, it means that model is not stable.

Atif Salar commented> Only one dot is on line rest are inside with this model can be
treated as stable there cannot be 100% accuracy with some conditions we can say
model is stable

Nicat Gasim > All roots must be inside unit circle for stationary or stable model. Please
see Chatfield, Oliver Anderson, Box and Jenkins, Hamilton etc. Time Series analysis

Sami Ozturk commented> I think he posted this picture since one of the roots is on the
line. The value of that roots will be .99956 something like this. Since it is less than 1, we
would say that this is a stable model.Having this type of root value might be have some
indication such as some of your variables might suffer from stationarity problem or you
might be using non optimal lag number to run your model. You should recheck them to
get rid of this type of root problems. I think the goal of posting this picture is not inverse
root or normal root. It is all about having one root on the line. We learn or teach
checking the stability conditions by finding the inverse root. However, computer
programs' outputs are normal roots, which require all roots must be in the unit circle.
Prof. Nicat Gisam explained these differences.
Meo School of Research
Shishir Shakya
Noman Arshed
VAR Stability Test
Hossain Academy Note
Univariate Models
Multivariate Models
Panel Data Model