Folks, This is more of a stat question than an R question. Apologies in advance! Suppose I fit an AR(1) to a time-series and also fit an AR(1) to the logs of the same time-series and then simulate future paths. In my case I see a big difference in the resulting paths. If I simulate thousands of paths under each scenario (exponentiating the results from the log-fit) and take the "mean" path for each I see a divergence. The average path from log-fit is increasing relative to the average path of the base fit. My guess is that there is a simple computation that would allow me to compare the two analytically but I just haven't been able to find it. Thanks for your time, KW -- [[alternative HTML version deleted]]