The relationship between the concepts of economic and energy efficiency is a point of continuing debate. Most analysts agree that policies should balance the benefits of energy efficiency against the associated costs. However, some view competitive markets as sufficient to achieve an optimal level of energy efficiency (Sutherland, 1991). Others point to empirical evidence suggesting that the level of energy efficiency achieved in today's markets falls short of the level that would prevail given the full implementation of cost-minimizing technologies (Carlsmith et al., 1990). Though the origins of this "efficiency gap" remain in dispute, a nascent literature attributes it in part to market failures relating to information asymmetries, transaction costs, and bounded rationality (DeCanio, 1997). If the costs of programs and policies aimed at surmounting such market barriers are less than the pursuant benefits, then government interventions facilitating the adoption of cost-effective, energy-efficient technologies may be justified on economic grounds (Sanstad and Howarth, 1994).
Technology analysts typically view energy efficiency as a way to reduce the environmental costs of energy use without corresponding reductions in the provision of energy services. Improved technologies, in this view, avoid the core controversies that surround efficient energy pricing, especially the difficulties of measuring the external costs of energy use and political objections to energy taxation. However, the notion that improved energy efficiency necessarily entails reduced energy use is questioned by Khazzoom (1980), who argues that energy efficiency improvements reduce the effective cost of energy services. Since reduced costs imply increased demand, energy efficiency improvements generally will induce less-than-proportional reductions in energy use. If the demand for energy services is sufficiently cost-elastic, energy efficiency improvements might, in principle, actually augment energy demand.
Following Khazzoom's contribution, the empirical magnitude of this so-called "takeback effect" has been intensively analyzed in the literature. Dumagan and Mount (1993), for example, use a generalized logit model to examine the impacts of efficiency improvements on residential electricity demand in New York State and find that takeback effects are numerically unimportant. A literature review by Greening and Sanstad (1995) finds that engineering studies that take energy services as fixed typically overestimate the energy savings of energy-efficient technologies by 1% to 20%. According to these authors, the small magnitude of takeback effects may be explained by two sets of factors. First, the demand for energy services is generally cost-inelastic, especially in the household sector where energy services are typically near or at saturation. Second, energy costs often are a small component in the cost of energy services, so that large changes in energy intensities are accompanied by modest changes in decision-makers' incentives.
This generalization does not imply that improved energy efficiency is always accompanied by reduced energy use. A counterexample is provided by Scott (1980), who examines the impacts of improved energy efficiency on residential space heating. In cases where consumers hold down energy costs by limiting the portions of their homes that are maintained at comfortable indoor temperatures, building shell retrofits and related efficiency measures can provide powerful incentives to increase the level of energy services. This phenomenon is most relevant in the case of low-income dwellings characterized by low energy efficiency and hence high energy costs - a small fraction of the housing market.
Khazzoom's work is based on a partial equilibrium framework that takes incomes and aggregate economic activity as fixed. A contribution by Brookes (1990) extends this argument to consider the impacts of energy efficiency on long-run economic growth. …