just going thru quant from schweser SS:pg 24 - it says: for an AR(1) model to be cov. stationary: b1 must be less than 0 if b1 =0,then there is an unit root. there must be a mean reverting level Please confirm that it should be b1<1 and b1=1 for unit root. i couldnt find this in the errata section. thanx
i saw that typo too and corrected it in my book.
Are you sure that isn’t the dickey fuller test thing? It is a way of testing that the coefficient is one by converting the model first.
Pretty sure that is what they are talking about. For some statistical reason you can’t directly test whether the coefficient is equal to 1 so you have to do a dickey fuller test, which makes you test whether it is equal to 0 or not. I don’t understand the process at all. I just might not be a typo.
good point, mwvt. We should look at the context and see if they are looking at the initial time series or differences.