Unit root question only, question, multiple

dmar

Does unit root apply to AR model with multiple independent variables or only AR(1)?
If yes for above, is it that if any of the slope coefficient is 1, then we conclude the AR(p) has unit root?
Thanks for your help.

former

I think it is if there is a lag slope coefficient equal to 1 then it has a unit root, ie not covariance stationary.

CFA4Techie

it only applies to AR(1). Remember that a model with the unit root is a random walk. A random walk only has one lag coefficient. Thus the random walk is a special case of AR(1)