The Latest in Behavioral Energy Efficiency Programs
Behavioral programs have been shown to be some of the most cost-effective energy efficiency measures for energy consumers of many shapes and sizes, and consumers want to participate. But these programs are still relatively new, and effective ways of evaluating, measuring, and verifying (EM&V) them are still being worked out. We've been pioneering our own programs, and conducting our own EM&V, which I had the opportunity to discuss during the recent ACEEE Intelligent Efficiency Conference in Boston.
The Premise of Behavioral Energy Efficiency Programs
Behavioral energy efficiency programs seek to encourage improvements in energy efficiency simply by providing users with feedback and actionable insight on their own energy usage. Energy reports are one tool we use at EnerNOC to facilitate this engagement.
By giving energy users, like small and medium enterprises (SMEs), the means to understand their own energy usage beyond the simple kWh tallies that they receive on their bill each month, we equip them with the knowledge they need to make better energy decisions.
The real challenge with behavioral programs is quantifying the effects these programs actually have on individual decisions. Unlike a building upgrade program, where we know ahead of time what kinds of energy usage changes to expect following the upgrade, with behavioral programs there is some uncertainty about the form that the usage changes will take.
Some customers may simply make sure to turn off their lights and air conditioners at the end of the day, while others might use the information from their reports to justify investing in energy efficient capital upgrades.
However customers choose to use the information in their energy reports, we face the same problem: we cannot observe what that customer would have done without the reports. This is the same problem faced by researchers in all kinds of fields from medicine to sociology to economics: we can never observe the counterfactual—the hypothetical world where the treatment (a report on a business's energy consumption, in this case) is never applied.
Filling the Knowledge Gap
The "gold-standard" solution to this problem is the randomized controlled trial (RCT). To perform an RCT, we start with a group of eligible customers. A portion of these customers—usually half—are then randomly assigned to be in one of two groups. One group, called the treatment group, is sent business energy reports (BERs) for the duration of the experiment. The other group, the control group, is simply observed. At the end of the experiment, the energy usage for the two groups is compared.
RCTs are deceptively powerful in their simplicity. The control group gets us as close as we can ever possibly get to observing the true counterfactual, and the statistical methods for differentiating between real treatment effects and random noise are powerful and well established.
Aside from simply measuring how much less energy is used by the treatment group compared to the control group, these experiments also equip us with even better insight into how different segments of business energy customers respond to behavioral efficiency programs, and what kinds of messages are most effective to promote energy efficiency actions.
What Comes Next
At the moment, we're rolling out some of the largest RCTs in behavioral energy efficiency that have ever been carried out for business customers, with 30,000 and 45,000 customers each in the treatment groups. This volume of data will afford us extremely high granularity in understanding how different businesses make use of feedback on their energy consumption.