I was amused and concerned by the allegations of marketing warfare that Mr. Krebs felt compelled to address in his December 1996 article.
Charting the DSM Sales Slump
Demand-side management (DSM) has become increasingly important for utilities across the country. In an effort to address environmental concerns and rising energy costs, utilities are deploying DSM to reduce the amount of energy consumed over the course of the year and/or during peak periods. 1 These programs are not new to the utility industry, having been around since the 1970s. DSM programs and technologies encourage customers through the use of incentives to buy more energy efficient technologies and to shift demand from peak hours (where the power grid is stressed due to high demand) to off-peak hours.
But DSM, by lowering sales growth, has made it difficult for utilities to recover their fixed costs, since a large portion of these costs are recovered through volumetric charges expressed in cents per kWh sold. In order to set new rates for an upcoming period of time, utilities file rate cases with state utility commissions. Rates are set so that the utility will make an allowed profit on sales above their projected costs over the period. Sales must be projected in order to set rates to meet the utility’s revenue requirement. 2
In recent years, many utilities across the industry have consistently over-forecasted sales due to factors like DSM and codes and standards that are difficult to account for in sales forecasting models. Forces such as these are putting downward pressures on sales. There are also certain intangible factors, such as a relative decline in consumer sentiment, that influence utility sales but simply cannot be captured in a quantitative model. 3
Currently, one of the greatest problems in utility sales forecasting is determining how much, if any, DSM is accounted for in the historical data used to estimate sales forecasting models. In instances where utilities employ DSM programs whose impacts are not fully captured in their sales models, certain techniques can be used to account for their impacts. By not accounting fully for the amount of DSM captured in sales models, utilities would otherwise overestimate electric sales for a given period of time. For this reason, the majority of utilities with DSM programs make adjustments to sales forecasts to account for DSM program impacts not captured by their sales models.
In order to determine the methods utilities are currently employing to deal with DSM, The Brattle Group reached out to 15 utilities to ask if they made any exogenous adjustments to their load forecasts for DSM. These North American utilities were medium-to-large sized utilities and had significant regional variation, spanning the Mid Atlantic, Midwest, Northeast, Southeast, and Western parts of the United States and parts of Canada. The survey told us that there are five main methods used to deal with DSM in sales forecasting, which, for the sake of convenience, we can characterize as follows: 1) Already Reflected – No Adjustment Needed; 2) No Prior History – Forward Only; 3) Prior DSM History – Embedded + Incremental; 4) Reconstructed Data – As If No DSM; 5) Econometric estimation of DSM impacts.
In the discussion that follows, we provide explanations of these four methods, plus