The market for demand-side products and services appears poised to explode. What began as separate energy efficiency, demand response and distributed energy program offerings are now coming...
Charting the DSM Sales Slump
graphic illustrations. We also provide a sample of responses (all anonymous) that we received from our survey participants.
Method 1 – It’s Already in the Data: The first method is to forecast sales without adjusting for DSM at all. The theory behind this is that DSM impacts are already fully accounted for in the historical data used to estimate sales models. The adjustment is not warranted either because there is no history of DSM and no expected amount of DSM in the future or the level of DSM has stayed constant in the past and is expected to stay constant in the future (as shown in Figure 1).
Method 2 – Going Forward Only: In contrast to the first method described above, each of the three remaining methods attempts to make adjustments to the rate case sales forecast to account for DSM impacts.
The most common method of these last three methods used to adjust sales for DSM impacts entails subtracting incremental DSM savings that are not embedded in historical sales from forecasted sales. In the simplest scenario, a utility with projected DSM programs would have no history of DSM. In this instance, as shown in Figure 2, the utility could simply subtract all future impacts from sales forecasts. The future impacts would be estimated based on engineering assessments, analysis of program results from other utilities or some combination of the two.
Method 3 – Embedded + Incremental: For utilities with a history of DSM programs, estimating incremental DSM is much more difficult. In order to do this, the utility must first estimate how much DSM is already accounted for in its sales data, which presents its own problems and uncertainty. This amount of “embedded” DSM is carried forward in the forecast.
Next, the utility must estimate future DSM impacts. The incremental DSM is therefore the difference between projected DSM and embedded DSM. Once the sales model projects sales for a given year, the incremental DSM impacts above the historical, embedded impacts are subtracted from projected sales. Figure 3 illustrates this adjustment. This is the most commonly used method.
Method 4 – As If No DSM: The last of the four methods, and the second most common approach to making an actual adjustment to DSM, is one where companies add back in the exogenously estimated historical DSM impacts to historical sales to come up with what sales would have been had DSM not occurred.
This method is illustrated in Figure 4. The green dotted line in the historical period is sales gross of DSM. These econometric sales forecasting models are effectively estimated over a reconstructed data series. These models predict a future devoid of DSM. To forecast sales net of DSM, these companies simply subtract the full value of projected DSM impacts from their model forecasts of gross sales.
In all four methods, historical and projected DSM program impacts must be measured. A variety of methods can be used to estimate DSM program impacts; some utilities rely on an engineering-based analysis, some rely on statistical measurement and verification studies to estimate DSM