In this paper, the stationarity of aggregate output is examined using the scaled variogram for a sample of twelve countries. It is argued that if the true data generating process of output is stationary around a deterministic trend with a change in mean and/or growth rate, the results of previous studies are biased. Once the authors allow for a one time exogenous 'shock,' the conclusions are indeed less clear cut and for some countries even reversed. Copyright 1994 by Ohio State University Press.