Imagine a setback thermostat programmed at the factory that the consumer couldn’t modify. Who would want this device? You could give the customer a big enough discount to get her to accept the...
The changing architecture of demand response in America.
off-peak period price was only 15-percent lower than the standard rate, yielding a ratio of 1.3:1 between peak and off-peak rates. Such minor rate differentials make it difficult for customers to achieve sufficient savings in their utility bills, de-motivating them from making behavioral changes necessary to shave peaks or transfer them to off-peak periods.
Many pilots with real-time pricing also suffer from this weakness because they don’t allocate capacity costs during the critical hours of the peaking season and spread them out uniformly over all 8,760 hours of the year.
Many pilots suffer from another weakness: They offer only a single dynamic price to customers in the treatment group. This makes it difficult to derive price elasticities with any degree of precision. This can become a serious problem if the utility or load- serving entity wants to engage in price-responsive demand bids to the wholesale market. In addition, by limiting themselves to a single price, all they are able to do is estimate the impact of that specific price by using analysis of variance or analysis of covariance. They aren’t able to estimate demand curves, which would allow them to predict future responses to alternative price levels. This capability is critical for making long-term forecasts under conditions of price uncertainty.
A few pilots don’t have any control group at all. It becomes difficult to predict the behavior of treatment group customers on treatment days since such behavior can be inferred only with reference to a comparable group of customers who weren’t on the treatment— i.e., a control group. Such pilots have limited internal validity. The best they can hope to do is to compare the behavior of treatment group customers on treatment days to their behavior on non-treatment days— i.e., the treatment group becomes its own control.
The lesson for those who have done their own pilot is to make sure it’s consistent with the scientific principles of experimental design. If it doesn’t hold up to these principles, they should consider conducting a new pilot that does.
Pathways to the Future
So what does the policy analyst do who has never carried out his or her own pilot? The analyst could decide to rely on the FERC results. However, there is significant variation in a key analytical parameter—the elasticity of substitution—across the pilots. This elasticity measures the extent to which customers are likely to curtail peak usage in response to higher peak prices, and provides a summary measure of the ability of customers to curtail peak loads or shift them to off-peak periods.
The variation in elasticities is illustrated in Figure 7 using information from the pilots that reported this parameter. The results being shown are for customers who are offered dynamic pricing rates but not offered enabling technologies. The lowest elasticity in the figure has a value of -0.073 and the highest one has a value of -0.13. 15 The impact of this variation in elasticities on the magnitude of DR enabled by dynamic pricing is significant. For example, if a dynamic pricing rate were to be offered where