The authors examine output growth for the G-7 countries. They use Nathan S. Balke and Thomas Fomby's procedure to identify the date and type of trend breaks, and note that these occur in clusters. The authors estimate country-specific intervention models and test the unit root hypothesis. Their critical values explicitly take account of the prior outliner search procedure. For Italy, Japan, and the United Kingdom, the unit root null hypothesis can be rejected. This suggests that the variance of output growth in these countries is generated by low-frequency, high-magnitude shocks rather than high-frequency, low-magnitude shocks. Copyright 1995 by Ohio State University Press.