For most things people purchase – like toothpaste, books, and televisions – prices don’t change over the course of the day: Consumers pay the same price regardless of whether they buy it at 8am, 2pm or 10pm. Residential households in the United States pay for the electricity that they consume like this too: Regardless of the time of day – 8am, 2pm or 10pm – households pay the same price per kilowatt-hour to power their iPad, run their air-conditioning, or charge their electric vehicle.
There are good policy reasons, however, for favoring dynamic electricity pricing – that is, charging a different price per kilowatt-hour of electricity over the course of the day – rather than time-invariant electricity pricing. One good reason is that there is often a large difference in the demand for electricity during peak periods – say, the middle of the afternoon on a hot summer day – compared to the demand for electricity during off-peak time periods – say, the middle of the night during that same hot summer day. To satisfy peak demand on a hot summer day, very expensive peaking plants need to be dispatched. As a result, shifting demand from peak periods to off-peak periods would create large economic gains.
One open question related to dynamic electricity pricing is how consumers would change their electricity consumption behavior as there is a dearth of rigorous, scientific program evaluations of time-varying pricing. As well, an important hurdle to implementing dynamic pricing is that many residential consumers lack the in-home technology that would allow them to respond to dynamic pricing, such as smart meters that would allow households to see their electricity consumption over the course of the day.
Koichiro Ito, a Postdoctoral Fellow at the Stanford Institute for Economic Policy Research and alumni of the Energy Institute at Haas from his days as a PhD student at the University of California, Berkeley, has a really neat new paper joint with Takanori Ida and Makoto Tanaka that addresses this open question. You can find their working paper here.
Ito and his coauthors are interested in understanding how residential households respond to dynamic electricity pricing. They find that households reduce their electricity consumption in peak hours when the peak hourly price is increased relative to the non-peak hourly price.
While that is not terribly surprising, they do find several more surprising results. First, they find no evidence that households shift electricity consumption to non-peak hours when households know they will face higher prices in peak periods compared to non-peak periods. In other words, households do not appear to “pre-cool” or “post-cool” their homes when prices are higher in peak periods compared to non-peak periods. In addition, although they find households continue to reduce their electricity consumption as the marginal peak price increases, they also find the incremental reduction becomes smaller as the marginal peak price increases.
These researchers have a very nice setting available to them for looking at these questions. Their findings are based on a randomized field experiment, a rigorous scientific approach to program evaluation, implemented in four communities in Japan. Each residential household in the field trial was provided a smart meter with an in-home display that shows the household’s electricity consumption over the course of the day as well as the hourly schedule of electricity prices for the current day as well as the next day. Households in the control group are charged (nearly) the same price at all hours of the day, while households in the treatment group are charged substantially higher prices during peak hours of the day (1pm-5pm) compared to off-peak hours of the day. For the treatment group, the peak hourly price is randomly assigned from four different peak price levels.
Ito and his coauthors are clear: their results are highly preliminary. First, these results are based on the first few months of a field experiment that is on-going for the next several years, and second, these results are based on just one of the four communities.
Nevertheless, the results are intriguing and provide some well-needed insight into dynamic electricity pricing based on rigorous, scientific methods.