Do all time series models with random walk (with or without drift) claim to have a regression coefficients of 1? The book is explaining how AR models may be invalid for these types of data since they will not be covariance stationary (since it assumes the regression coefficients are 1).

Thanks!

Be careful about your terminology.

In a random walk,

*x*(*i*+1) = _b_0 + _b_1 × *xi*

*Both* _b_0 and _b_1 are *regression* coefficients; however, _b_1 is a **slope** coefficient, while _b_0 is not.

In any case, in a random walk, _b_1 = 1. If _b_0 = 0, it is a random walk *without* drift, and if _b_0 ≠ 0, it is a random walk *with* drift.