NOTE: This blog article revises and updates a previous entry.
“Is There an Energy Efficiency Gap?” is the title of a recent article by Hunt Allcott and Michael Greenstone published in the Journal of Economic Perspectives. ACEEE wrote a critique of this paper. In response, Allcott and Greenstone have posted an updated version of their paper. Allcott and Greenstone examine energy efficiency and specifically “whether there are investment inefficiencies that a policy would correct.” They conclude that “the available evidence from empirical analyses of weatherization, demand-side management programs, automobile and appliance markets, the ‘landlord–tenant’ agency problem, and information elicitation suggests that while investment inefficiencies do appear in various settings, the actual magnitude of the Energy Efficiency Gap is small relative to the assessments from engineering analyses.”
The new version of Allcott and Greenstone’s work corrects a number of problems with their earlier paper. In the new version they generally note all of an author’s findings and not just selected findings. They also include some additional citations to relevant works. Still, while acknowledging uncertainties, they have not changed their conclusion that the available evidence suggests that the energy efficiency gap is small. ACEEE has just released a revised white paper examining their arguments. We find that the authors have some useful insights, but that rather than providing a dispassionate review of the evidence, they interpret available data in ways that best support their points, downplaying other research findings.
In much of the research they cite, they emphasize results using high discount rates, relegating lower discount rate scenarios from the same articles to footnotes. Likewise, they tell only part of the story when they say that landlords paying for investments and tenants paying for energy do not cause a big barrier to efficiency investments, citing a study that insulation levels in multifamily rental homes in California are only slightly lower than single-family homes. They downplay the fact that California has a building code that requires all new homes to have high levels of insulation.
In the case of vehicle fuel efficiency, Allcott and Greenstone assert that existing and proposed CAFÉ standards appear to be overly stringent without reference to recent, dramatic advances in gasoline vehicle technologies. Nor do they mention that the standards now vary with vehicle size, which is relevant to concerns about consumer choice and potential loss of utility. They also ignore investment inefficiencies on the manufacturer end, though these help to explain why fuel economy standards are so useful.
More broadly, it is also worth noting that externalities and information gaps are not the only rationales for energy efficiency policies. Efficiency policies can spur manufacturers to develop higher efficiency products, giving consumers more choices and creating economies of scale help to reduce the price of these efficiency improvements. Past standards have helped spur the development of new motor vehicles, lamps, and clothes washers. Furthermore, utilities offer efficiency programs because they are less expensive than new power plants.
Allcott and Greenstone do have some useful points to make. They argue that consumers are heterogeneous and that policies should be tailored to address this. For example, they argue that appliance standards for heating and cooling equipment should be different in Chicago than in Los Angeles. We tend to agree—for example, we successfully advocated that the latest standards for furnaces and central air conditioners vary between the North and South. However, customizing policies also has costs, and a balance needs to be found between ease and cost of implementation and addressing heterogeneity.
Finally, Allcott and Greenstone argue that additional rigorous empirical work is needed—utilizing randomized controlled trials and quasi-experimental techniques. In fact, Allcott and Greenstone (together and separately) now have two such projects underway. We agree that further study would be useful. Randomized controlled trials can be very useful, but these can be hard to do in the “real world” as they can be expensive and hard to fit in around other programs that are serving the same customers. Thus, while we support randomized control trials where possible, we believe that they can and should be complemented with careful evaluations of consumer responses to existing programs relative to a control group of non-participants. Analyses can also be conducted on important sub-groups of participants in large programs. There are hundreds of current programs that could be evaluated, providing useful information to design the most effective energy efficiency programs.
Therese Langer contributed to this blog post.