Recent reviews of forecasting accuracy in the electric utility industry have been pessimistic (Peck and Nelson, 1985; Huss, 1985a Huss, , 1985b)). For example, Huss makes the point that 'direct observation shows little or no evidence that utility forecasting is improving over time'.
Additionally, he states that the use of 'econometric modeling . . . in utility forecasting must be questioned'.
Many readers may take such statements to mean that the 'millions of dollars' spent in developing more sophisticated forecasting techniques have been wasted. Such a perception could arrest the current momentum towards further improvement of forecasting methodologies. This will not bode well for utility planning and operations in the future.
To rectify such a possible misperception, we deal with three major points in this preface:
- What determines the usefulness of forecasts in utility planning and operations? 2. Historically, have sophisticated forecasting techniques been no more accurate than simpler ones? 3. Are econometric techniques of little use in forecasting?
Electric utilities have several needs that are met by forecasts. These include corporate planning, competitive strategy development, system planning, fuel purchasing, revenue planning and budgeting, marketing, and operations and dispatch management. Thus, accuracy of point forecasts is just one of the many criteria for judging the usefulness of forecasts. Others include (a) ability to scan the future environment under a variety of scenarios (b) ability to assess the impact of new end-use technologies on utility systems (c) assistance in developing marketing strategies to shape the future by providing a deeper understanding of customer behaviour, including rate of participation in demand-side management programmes, load response, etc. (d) assessing competitor strengths and weaknesses through quantification of interfuel substitution and productivity assessments.
Given the increased competition that is developing in electricity markets, there is every indication that forecasting will become an even more important utility function in the future (Booz-Allen & Hamilton, Inc., 1985; EPRI, 1986).