Ahmad Faruqui & Stephen George,
Charles River Associates
While many problems conspired to create California’s energy crisis, most people agree that the crisis was exacerbated by the lack of dynamic price signals in retail markets. Dynamic pricing would have given customers an incentive to lower loads during peak times which, in turn, would have reduced market clearing prices and partially mitigated market power. However, dynamic pricing requires the installation of digital interval meters and makes economic sense only if its benefits exceed the costs of the new metering infrastructure. The benefits depend on the impact of dynamic pricing on customer loads and on avoided supply-side costs, both of which are uncertain.
To address the uncertainty in customer loads, the state of California conducted a large-scale test of several innovative pricing designs involving roughly 2,500 residential and small commercial and industrial (C&I) customers. The pilot was conducted jointly by the state’s three investor-owned utilities and two regulatory commissions through a working group process involving several interested parties including consumer groups, researchers, academics and metering vendors.
The scientifically designed experiment was conducted between July 2003 and December 2004. Several time-varying rates were examined, including a two-period, time-of-use (TOU) rate and two versions of a critical-peak pricing (CPP) rate. The CPP rates had prices similar to the TOU rate on all but 12 summer days and three winter days when peak-period prices were much higher than on other days. These 15 high-priced days were called on a day-ahead or day-of basis to simulate dynamic or real-time conditions that might be encountered during a crisis.
The study found that residential customers in California will reduce their peak period energy use by about 5 percent in response to a TOU rate, which features prices that were about twice as high as the standard price of 13 cents/kWh, and about 15 percent in response to a CPP rate, which features prices that were five times as high. The percent reductions in peak load varied by climate zone, being lower in cooler climate zones such as San Francisco and higher in warmer climate zones such as the Central Valley. They also varied with customer characteristics, being higher for customers with central air conditioning, college education, higher incomes and higher overall usage.
Impacts on peak-period energy use on critical days did not differ significantly across the summers of 2003 and 2004. However, impacts were higher during the hot months of July through September than during the milder summer months of May, June and October. Winter impacts were lower than summer impacts. Impacts were lower during the milder months of November, March and April than during the colder months of December, January and February.
A key issue prior to the experiment was whether time-varying rates might induce overall energy conservation. The results showed that the new rates did not have any effect on overall annual energy consumption.
Another issue was whether there would be any degradation of response if multiple critical days were called in a row. The experimental results showed that peak-period impacts did not differ significantly when two or three critical days were called in a row, as might happen during a heat wave.
There was also a question about whether technologies such as smart thermostats enhanced impacts. The experiment showed that reductions in peak loads triggered by CPP rates could exceed 25 percent if customers were given enabling technology.
The new estimates of price responsiveness from California are at the low end of the range reported in the literature. For example, the percent reduction in peak demand is about 60 percent of the reduction observed in a comprehensive multi-experimental study conducted by EPRI in the early 1980s. That study was based on data from the five best TOU experiments conducted around the country, including two in California and single experiments in Connecticut, North Carolina and Wisconsin.
The lower impact may be due to California’s ambitious program of energy conservation and load management undertaken during the past quarter century. In addition, the energy crises of 2000 and 2001 may have fatigued customers and their ability or willingness to further reduce energy use. Nevertheless, it is also very clear that residential customers still display a surprisingly robust amount of demand response to TOU and dynamic pricing.
The experiment has also yielded insights about the response of C&I customers to time-varying tariffs. The C&I population was segmented into two groups: those with peak demands less than 20 kW (LT20) and those with peak demands between 20 and 200 kW (GT20).
The results showed that LT20 customers reduced energy used during the peak period by 6.0 percent while GT20 customers reduced peak-period energy use by 9.1 percent. The absolute size of the reduction in peak-period energy use for GT20 customers is roughly 10 times larger than for LT20 customers, due primarily to the fact that average energy use for GT20 customers was much larger than for LT20 customers.
A subset of C&I customers were provided a smart thermostat as an enabling technology. Within this group, LT20 customers reduced peak-period energy use on critical weekdays by 14.3 percent, and all of this reduction was attributable to the enabling technology. GT20 customers reduced peak-period energy use on critical weekdays by 13.8 percent of which about 80 percent was attributable to the enabling technology.
These new experimental results have been used by the three utilities to develop business cases for advanced metering infrastructure that is estimated to cost several billion dollars for their 10 million residential customers. The California Public Utilities Commission is in the process of reviewing the business cases and is expected to announce a final decision later in the year.
Economists Faruqui and George are vice presidents with Charles River Associates.