On forecasting discontinuities

Too much has been written about forecasting trends via "envelope curves." T o plot a straight line on log-paper is a no-brainer. The special phenomenol ogy of phase changes, catastrophes, or "crashes," and their key evolutionar y role is the subject of this article. There is net unique way of forecasti ng such events, but one indicator is an apparent inconsistency between two or more extrapolations with each other. Alternatively, a catastrophe may be signalled when a trend extrapolation encounters a natural limit. A number of possible discontinuity scenarios are sketched, although overall, the art icle laments the lack of theorization in forecasting disconuities. (C) 2000 Elsevier Science Inc.