Sanem Sergici

The impact of informational feedback on energy consumption—A survey of the experimental evidence

In theory, In-Home Displays (IHDs) can revolutionize the way utilities communicate information to customers because they can induce changes in customer behavior even when they are not accompanied by a change in electric prices or rebates for purchasing efficient equipment. IHDs provide consumers with direct feedback—real-time information on energy consumption and costs—and turn a once opaque and static electric bill into a transparent, dynamic, and controllable process. However, to what extent do consumers actually respond to the direct feedback provided by IHDs?

Household Response to Dynamic Pricing of Electricity - A Survey of the Empirical Evidence

Since the energy crisis of 2000-2001 in the western United States, much attention has been given to boosting demand response in electricity markets. One of the best ways to let that happen is to pass through wholesale energy costs to retail customers. This can be accomplished by letting retail prices vary dynamically, either entirely or partly. For the overwhelming majority of customers, that requires a changeout of the metering infrastructure, which may cost as much as

Arcturus: International Evidence on Dynamic Pricing

40 billion for the US as a whole. While a good portion of this investment can be covered by savings in distribution system costs, about 40 percent may remain uncovered. This investment gap could be covered by reductions in power generation costs that could be brought about through demand response. Thus, state regulators in many states are investigating whether customers will respond to the higher prices by lowering demand and if so, by how much. To help inform this assessment, we survey the evidence from the 15 most recent pilots, experiments and full-scale implementations of dynamic pricing of electricity. We find conclusive evidence that households (residential customers) respond to higher prices by lowering usage. The magnitude of price response depends on several factors, such as the magnitude of the price increase, the presence of central air conditioning and the availability of enabling technologies such as two-way programmable communicating thermostats and always-on gateway systems that allow multiple end-uses to be controlled remotely. They also vary with the design of the studies, the tools used to analyze the data and the geography of the assessment. Across the range of experiments studied, time-of-use rates induce a drop in peak demand that ranges between three to six percent and critical-peak pricing tariffs induce a drop in peak demand that ranges between 13 to 20 percent. When accompanied with enabling technologies, the latter set of tariffs lead to a drop in peak demand in the 27 to 44 percent range.

Dynamic Pricing of Electricity for Residential Customers: The Evidence from Michigan

This paper introduces Arcturus, an international database of dynamic pricing and time-of-use pricing studies. It contains the demand response impacts of 163 pricing treatments that were offered on an experimental or full-scale basis in 34 projects in seven countries located in four continents. The treatments included various types of dynamic pricing rates and simple time-of-use rates, some of which were offered with enabling technologies such as smart thermostats. The demand response impacts of these treatments vary widely, from 0% to more than 50%, and this discrepancy has led some observers to conclude that we still don’t know whether customers respond to dynamic pricing. We find that much of the discrepancy in the results goes away when demand response is expressed as a function of the peak to off-peak price ratio. We then observe that customers respond to rising prices by lowering their peak demand in a fairly consistency fashion across the studies. The response curve is nonlinear and is shaped in the form of an arc: as the price incentive to reduce peak use is raised, customers respond by lowering peak use, but at a decreasing rate. We also find that the use of enabling technologies boosts the amount of demand response. Overall, we find a significant amount of consistency in the experimental results, especially when the results are disaggregated into two categories of rates: time-of-use rates and dynamic pricing rates. This consistency evokes the consistency that was found in earlier analysis of time-of-use pricing studies that was carried out by EPRI in the early 1980s. Our analysis supports the case for the rollout of dynamic pricing wherever advanced metering infrastructure is in place.

Dynamic Pricing in a Moderate Climate: The Evidence from Connecticut

The rollout of smart meters has enabled the provision of dynamic pricing to residential customers. However, doubts remain whether households can respond to time-varying price signals and that is preventing the full-scale rollout of dynamic pricing and the attainment of economic efficiency. Experiments are being conducted to test price responsiveness. We analyze data from a pilot in Michigan which featured two dynamic pricing rates and an enabling technology. Unlike most other pilots, it also included a group of “information only” customers who were provided information on time-varying prices but billed on standard rates. Similarly, unlike most other pilots, it also included two control groups, one of whom knew they were in the pilot and one of whom did not. This was designed to test for the presence of a Hawthorne effect. Consistent with the large body of experimental literature, we find that customers, including low-income participants, do respond to dynamic pricing. We also find that the response to critical peak pricing rates is similar to the response to peak time rebates, consistent with the finding of one prior experiment but inconsistent with the finding of two prior experiments. We also find that the “information only” customers respond to the provision of pricing information but at a substantially lower rate than the customers on dynamic pricing. We find that the response to enabling technology is muted. We do not find any evidence to suggest that a Hawthorne effect existed in this experiment.

Arcturus : An International Repository of Evidence on Dynamic Pricing

While many dynamic pricing experiments have been carried out in warm climates, few have been carried out in moderate climates. We analyze data from a pilot in New England which featured a time-of-use rate, two dynamic pricing rates and four enabling technologies. Unlike most other pilots, it included small commercial and industrial (C&I) customers in addition to including residential customers, for a total of around 2,200 customers. Using a constant elasticity of substitution model of consumer behavior, we find that customers do respond to dynamic pricing even in a moderate climate, that response to critical-peak pricing rates is higher than response to peak-time rebates, that there is virtually no response to TOU rates with an eight hour peak period and that small C&I customers are less price responsive than residential customers. We also find that some enabling technologies boost price responsiveness.

The Econometrics of Dynamic Pricing in a Mid-Atlantic Experiment

This chapter introduces Arcturus, an international database of dynamic pricing and time-of-use pricing studies. It contains the demand response impacts of 163 pricing treatments that were offered on an experimental or full-scale basis in 34 projects in seven countries located in four continents. The treatments included various types of dynamic pricing rates and simple time-of-use rates, some of which were offered with enabling technologies such as smart thermostats. The demand response impacts of these treatments vary widely, from 0 % to more than 50 %, and this discrepancy has led some observers to conclude that we still do not know whether customers respond to dynamic pricing. We find that much of the discrepancy in the results goes away when demand response is expressed as a function of the peak-to-off-peak price ratio. We then observe that customers respond to rising prices by lowering their peak demand in a fairly consistent fashion across the studies. The response curve is nonlinear and is shaped in the form of an arc: as the price incentive to reduce peak use is raised, customers respond by lowering peak use, but at a decreasing rate. We also find that the use of enabling technologies boosts the amount of demand response. Overall, we find a significant amount of consistency in the experimental results, especially when the results are disaggregated into two categories of rates: time-of-use rates and dynamic pricing rates. This consistency evokes the consistency that was found in earlier analysis of time-of-use pricing studies that was carried out by EPRI in the early 1980s. Our analysis supports the case for the rollout of dynamic pricing wherever advanced metering infrastructure is in place.

Household response to dynamic pricing of electricity: a survey of 15 experiments

The Baltimore Gas & Electric Company (BGE) undertook a dynamic pricing experiment to test customer price responsiveness to different dynamic pricing options. The pilot ran during the summers of 2008 and 2009 and was called the Smart Energy Pricing (SEP) Pilot. In 2008, it tested two types of dynamic pricing tariffs: critical peak pricing (CPP) and peak time rebate (PTR) tariffs. Some thousand customers who randomly placed on these tariffs and some of them were paired with one of two enabling technologies, a device known as the Energy Orb and a switch for cycling central air conditioners. The usage of a randomly chosen control group of customers was also monitored during the same time period. In 2009, the pilot only tested the PTR tariff. In this paper, we estimate a well-known demand model on the hourly consumption, pricing and weather data that was compiled in the SEP pilot. We derive substitution and daily price elasticities and predictive equations for estimating the magnitude of demand response under a variety of dynamic prices. We also test for the persistence of impacts across the two summers. We also report average peak demand reduction for each of the treatment cells in the SEP pilot and compare the findings with those reported from earlier pilots.