Does the mean reverting level formula [b0/(1-b1)] apply for linear trend model also?
Yes I believe so. Can somebody confirm?
No , I dont think so as this formula is used for AR models Linear trend models and Log models are of the form yt = bo + b1* t Pls note that the independenet variable is based on a time -dependent variable, whereas AR models use the lagged dependent variable ( Thats why you have the formula as yt = bo + b1* y(t-1)
A linear trend model doesn’t have a long term mean to revert to…it grows at its linear trend. Now, if you detrend the series you may find that it becomes trend-stationary, and then it will have a mean-reverting level, but this is rarely the case in financial series.
no mean reverting level for linear trend models 
that’s why its called linear trend model
I believe all the mean reversion, unit root, random walk, covariance stationary and cointegration only applies to AR(k) models and when you regress two time series together. Linear trend model is just an ordinary linear regression with time as the independent variable.
Thank you all. I asked this question because a question&answer in schweser notes: (Warranty expense)t = 74.1 - 2.7* t + et The mean reverting level is X1 = b0/(1-b1) X1 = 74.1/[1-(-2.7)] = 20.03 So it seems a wrong answer.
deriv108 Wrote: ------------------------------------------------------- > Thank you all. I asked this question because a > question&answer in schweser notes: > > (Warranty expense)t = 74.1 - 2.7* t + et > > The mean reverting level is X1 = b0/(1-b1) > X1 = 74.1/[1-(-2.7)] = 20.03 > > So it seems a wrong answer. in my opinion schweser is often crap