Electricity demand is less elastic than I'd thought.
ISCR's Competition & Regulation Times reports (pdf) on an experiment run by New Zealand power company Mercury Energy. Four hundred households were randomly assigned to four different treatment groups. One group received information about how to reduce electricity usage. The other three groups received the information plus a time-of-day price differential of on- and off-peak power use on weekdays: 4 cents, 10 cents or 20 cents: the experiment added half the differential to the household's existing on-peak price and subtracted half the differential from the off-peak price.
Results? There was no change in power use in summer, autumn and spring. In winter, the group experiencing the largest price differential shifted some power use to off-peak times: the 50% price premium on on-peak use resulted in a 6% decrease in on-peak usage and a 4.5% increase in off-peak use.
I was pretty surprised that demand was this inelastic. The article notes that Mercury didn't use time-of-use pricing with any of its customers prior to the experiment. When we moved to New Zealand, we signed up with Meridian Energy and took its Night-and-Day rate plan; we then shifted our laundry, dishwasher, and hot water heating to the lower night-rate. Current daytime power prices with Meridian on the DayNight plan are just about three times the night rate: $0.3072 per kWh daytime and $0.1180 per kWh at night. After the kids showed up, we stopped turning the hot water cylinder off during the daytime.
We're now with PowerShop which, while not charging an explicit time-of-day price, conditions your average price on your day/night/weekend usage patterns; we've kept our "shift to the evening" strategy. PowerShop's underlying daytime power rate, when I'd enquired back in May, was about double the price of their night-time rate.
The Mercury experiment suggests that larger roll-out of time-of-use dependent pricing can shave some peaks off peak power use, and knocking back those peaks can be pretty important. But I had expected time-of-day power usage to be more price sensitive.