Introduction and Context

Since the 1970s, efforts to improve energy efficiency and save energy have been undertaken by many countries, businesses, and individuals. As the International Energy Agency (IEA—an organization of 29 developed countries) recently noted: “Energy efficiency has been the primary factor in driving down energy consumption in IEA countries over the last decade.” In 2014 alone, IEA estimates that energy efficiency reduced IEA countries’ total primary energy supply by 760 Mega-tonnes of oil equivalent (Mtoe) (32 EJ) as a result of energy efficiency investments since 1990 [1]. To put these savings in perspective, in 2013 (the most recent year with data), the entire world used 13,559 Mtoe [2•], of which IEA countries used about 43 % [3]. Thus, the 760 Mtoe of IEA annual energy savings represents about 13 % of IEA energy demand.

Some countries have saved even more. For example, Nadel et al. examined energy use and energy savings in the USA since 1980 and estimated that energy efficiency efforts since 1980 reduced US primary energy use in 2014 by about 58 quadrillion Btus of energy (61 EJ); without energy efficiency, 2014 energy consumption would have been about 59 % higher (primary energy use is total energy use and includes energy lost in electric generation as well as transmission and distribution of electricity and fuels). On a compound basis, they estimate energy efficiency has reduced US energy consumption by 1.2 % per year. Energy intensity (energy use per dollar of GDP) has declined by about 2.0 % per year over this period, but they estimate that about 60 % of this change is due to energy efficiency and the other 40 % due to structural change in the US economy (e.g., less heavy manufacturing and more services) [4•]. Likewise, Schlomann and Eichhammer estimate that between 1990 and 2011, German absolute primary energy use fell by 12 % (a compound average of 0.6 % per year), while primary energy intensity fell by an average of 2.0 % per year [5] (the same as the USA). Data compiled by the US Energy Information Administration (EIA) shows that five large energy-consuming countries (India, Poland, Russia, Ukraine, and the UK) reduced their energy intensity per dollar of GDP over the 2000–2011 period (the last year with full data) by more than 20 %, and an additional five large energy consumers (Canada, Germany, South Africa, Taiwan, and the USA reduced their energy intensity by 15–20 % [3]. Improvements of 15 and 20 % work out to compound annual improvements of 1.5 and 2.0 %, respectively. More recently, China has had aggressive energy efficiency targets as part of recent five-year plans. For the 2011–2015 plan, they report that energy intensity has improved 18.2 % over the 5-year period [6]. This is a 4.0 % annual compound rate of improvement—an impressive achievement. However, as noted above for the USA, some of the improvement in China, India, Poland, etc., are probably due to structural changes in their economies—not all of the improvement is due to energy efficiency.

Given this past history, an important question is how much energy efficiency potential remains in developed countries? Another important question is what is the energy efficiency potential in developing countries that are seeking to grow their economies and improve standards of living? In this paper, we review some of the major studies on energy efficiency potential, covering studies from the USA, Europe, and Japan and also touching on other countries including several less-developed countries. Our focus is on estimates of potential out to 2040–2050, but we also review some studies that look at shorter-terms.

One factor that affects energy efficiency potential is the rebound effect—when consumers and businesses improve energy efficiency, they sometimes channel some of the efficiency gains into increased comfort (e.g., increasing the thermostat set point in winter) or amenity (e.g., driving their more efficient car more kilometers). Also, if consumers and businesses save energy, they generally reduce their energy bills, and some of these financial savings may be invested in ways that increase energy use such as a factory investing in a new production line. If rebound is substantial, it can reduce the energy savings achieved. We review key studies that estimate the size of rebound effects in different market segments, summarize results across different markets, and then comment on how these results may affect the size of the long-term energy efficiency opportunity. One important point to note is that all of the data on the previous page are based on actual energy use and therefore incorporate rebound effects. The question is how to allow for rebound in projections of potential future energy savings.

Energy Efficiency Potential

There are many studies that estimate energy efficiency potential. In this section, we focus on some of the major ones, emphasizing long-term studies. But first, a brief typology of energy efficiency potential studies is in order. Energy efficiency potential studies look at a variety of potentials, but the most common are technical potential, economic potential, and achievable potential. Technical potential is what is possible without considering cost-effectiveness. Economic potential only includes cost-effective measures but implicitly assumes 100 % adoption. Achievable potential is an estimate of what can be achieved given specific policies and programs. In the discussion below, we concentrate on economic and achievable potential estimates, as many studies estimate one or the other and only some do both. Potential studies can be bottom-up or top-down. Bottom-up studies look at the savings from specific energy efficiency policies and measures, building up the potential as a series of building blocks. Underlying these studies are specific assumptions about which policies and measures and how much they will save. Top-down studies look at overall rates of efficiency improvement, often based on past experience, such as some of the estimates of past experience discussed in the section above. Potential studies can look at primary energy use (defined previously) and/or final energy which does not include upstream losses in power and energy systems (sometimes called site energy or delivered energy). Unless otherwise noted, the figures we discuss below are for primary energy since primary energy is a more complete measure of total energy use.

Dozens of studies have been published in recent years that estimate the potential for energy efficiency improvements in the next few decades including major studies by the IEA [2•], European Union [7, 8•], Deep Decarbonization Pathways Project [9•, 10], Rocky Mountain Institute (RMI) [11], the American Council for an Energy-Efficient Economy (ACEEE) [12, 13••], and the Electric Power Research Institute (EPRI) [14]. There are also several recent journal articles [15–17]. Table 1 summarizes these major studies. These are not the only studies, but they do give a flavor for recent work.

Table 1 Summary of key energy efficiency potential studies
Full size table

A few comments on some of the studies:

  • The deep decarbonization studies were on 15 individual countries. All of the countries found substantial opportunities for energy efficiency and other energy intensity improvements as shown in Fig. 1. In most developed countries, these efficiency improvements were accompanied by absolute reductions in final energy consumption in 2050 relative to 2010 consumption. In developing countries, absolute final energy consumption increased over the 2010–2050 period, with the increases ranging from 36–41 % in Mexico to 225–338 % in India [18].

    Fig. 1
    figure 1

    Estimated available reduction in energy use per dollar of GDP from 2010–2050 by country [9•]

    Full size image

  • In addition to providing numbers, the ACEEE study by Neaubauer [13••] also examines the many assumptions some of these reports make, finding that inputs have a significant impact on results. Key assumptions include participation rates in energy efficiency programs, estimates of the impacts of building codes and equipment efficiency standards, the extent to which emerging technologies are included in the estimates, and the assumptions about avoided costs (the higher the avoided cost, the more efficiency that is cost-effective).

  • Reviewing the detailed assumptions in the EPRI study [14] relative to the other studies, the EPRI study included few new technologies (e.g., LED lighting was not included), little in the way of whole-system improvements (what RMI calls integrative design), tended to use relatively high measure costs and used a relatively restrictive cost-effectiveness test that includes all costs but only the benefits to the utility.

  • The Wang and Brown study [17] looked at savings from 11 specific policies, capturing some but not all of the available efficiency potential. Savings are greater in the first 10 years (0.6 %/year) than the next 15 years (0.35 %/year). In order to sustain savings, policies will need to be periodically updated.

Energy Efficiency Potential Discussion

A substantial majority of the studies discussed above are finding potential future energy efficiency improvements of about 1.0–1.4 % per year. These are mostly achievable potential estimates, but some are economic potentials. These rates of efficiency improvement are in line with efficiency improvements since 1990 in several major countries as discussed in the “Introduction and Context” section of this paper. While achieving these potentials is not easy, as discussed by Alcott and Greenstone [19], past experience as well as a critique of this work [20] leads us to conclude that achieving future savings of this magnitude should be possible.

A few of the studies find higher efficiency potentials of about 2.0–2.6 % per year including the more aggressive scenarios in the RMI and ACEEE studies of the USA; some of the deep decarbonization scenarios if all of the improvement is due to efficiency and none to structural change, and the study on the European Union prepared for the DG ENER. As discussed in the “Introduction and Context” section, this level of efficiency improvement has not been achieved in high-income countries in the recent past and therefore for these potentials to be achievable will require more aggressive interventions that have generally been employed in the past. Such efficiency improvements have been achieved in isolated cases in the high-income countries in the past (e.g., several American utilities have achieved electricity efficiency savings of more than 2 % per year [21], and California achieved a reduction over 6 % in 1 year from a combination of efficiency improvements and conservation behavior during a major energy crisis [22] ) and have also been achieved in some developing countries such as China during their last five-year plan [6]. Overall, it is unclear if these levels of savings can be achieved across entire economies for multiple decades.

The Rebound Effect

Different authors have suggested different types of rebound effects, but these boil down to two general types—direct and indirect. Direct rebound is the impact of a purchase of an efficient product on the purchaser’s use of that product. For example, a car buyer may drive an efficient car more often than an inefficient one or a homeowner who weatherizes his/her house may use a portion of the savings to increase the temperature in the house in the winter in order to increase comfort. Indirect rebound, on the other hand, reflects other upstream impacts including the following: (1) the impact of respending the money that consumers and businesses save from improved energy efficiency (sometimes called an income effect) and (2) the fact that as factories and other parts of the economy get more efficient, production costs may be lower and as a result prices decline and demand for these products can increase (sometimes called a substitution effect). An example of the first is a household that cuts its heating bill and takes back a little of the savings on higher thermostat settings but then spends the money saved on eating out or buying a new television. An example of the second is that efficiency improvements in aluminum smelting can reduce the price of aluminum, thereby fostering increased aluminum sales that require additional energy to produce this aluminum.

Over the years, many studies have examined the rebound effect for specific programs and policies on specific countries, and there have been multiple meta-compilations of these studies [23–25]. There have also been several more recent reviews and compilations including the ones by Economic Consulting Associates (ECA) [26••], Gillingham et al. [27••], and Nadel [28].

Direct Effects

In Table 2, findings on direct rebound effects are summarized, drawing primarily from these more recent studies but also drawing in some key studies referenced in these recent studies. Direct effects have been most extensively studied for passenger vehicles and for space heating in developed countries, but there are also some data for developing countries, other residential energy uses, and for some commercial and industrial applications.

Table 2 Estimates of direct rebound effects from recent studies
Full size table

Indirect Effects

ECA [26••] summarizes a variety of studies that look at indirect rebound effects in particular markets. They report median indirect effects of 6–31 % for high-income countries, varying by the energy use being examined. Based on their review of these studies, their most likely estimate of indirect effects is 10–20 % for high-income countries and 20 % for middle-/low-income countries. Gillingham et al. [27••] discuss a variety of indirect effects. First, they note work by others that found an indirect rebound effect of 5–15 %; for example, Thomas and Azevedo [34] found a 5–15 % indirect rebound effect in US households using an input-output model and a detailed consumer expenditure survey to estimate income elasticities. Second, they explore a macroeconomic price effect, such as the impact of oil price changes on economic growth, concluding that in the case of the oil market, this effect is likely to be on the order of 20–30 %. And third, they discuss use of calibrated simulation models of a country’s economy. For example, studies on the UK and world economies estimated total rebound (direct plus indirect) of 11 % [35] and 21 % [36], respectively.

Rebound Effect Discussion

Most of the results summarized in Table 2 indicate direct rebound effects on the order of 10 % plus indirect effects on the order of 10–20 % for a total rebound of about 20–30 %. The International Risk Governance Council (IRGC) reaches a similar conclusion, finding that: “The evidence to date from econometric studies that generally use price elasticity, income elasticity and elasticity of substitution suggests that direct and indirect rebound effects in developed economies are moderate and that investments in energy efficiency can save between 70 and 85 per cent of the anticipated energy reduction, while allowing households to enjoy the benefits of higher consumption” [37]. However, the data in Table 2 indicate that there are some important exceptions where direct rebound can be higher including with passenger vehicles and space heating outside of the USA (the latter for countries with cold climates where spaces are currently not heated to comfortable temperatures) and appliances and air conditioning in developing countries. In these other cases, direct rebound might generally be on the order of 20 %, except they could well be higher for appliances and air conditioning in developing countries. Furthermore, economies are particularly sensitive to oil prices, with macroeconomic rebound associated with oil savings in the range of 20–30 %. Estimates of indirect rebound effects are often imprecise. As the IRGC notes “indirect rebound effect is likely to depend on the economy under study and most of these drivers have not been thoroughly investigated across a broad number of economies. In addition, prior work has been largely parametric; [more] empirical research on the magnitude of… indirect rebound effects is needed” [37].

There have also been a few claims of “backfire,” such as in reports by the Breakthrough Institute and its contributors [38]. “Backfire” means that rebound is larger than the efficiency savings and therefore energy use increases rather than decreases as a result of efficiency programs, policies, and investments. Most other analysts reject these claims. For example, Gillingham et al. state “the existing literature does not provide support for claims that energy efficiency gains will be reversed by the rebound effect” [27••]. ECA estimates that total economy-wide rebound effects (direct plus indirect) likely range from 10–60 % depending on the sector, end-use, and income of the country—well short of the 100 % needed for backfire [26••]. And Cullenward and Koomey specifically question the data on high rebound effects in the industrial sector [39].

Energy Efficiency Potential Considering the Rebound Effect

In our section on energy efficiency potential, we found that compound energy efficiency savings of 1.0–1.4 % per year appear to be feasible, and savings of 2.0–2.6 % per year might be possible but have been infrequently demonstrated in practice. These estimates of potential future savings by and large do not include rebound effects, although the historic estimates of 0.6–2.0 % per year discussed in the “Introduction and Content” section do include rebound. If we incorporate a range of 20–30 % rebound into these future savings estimates, then the net energy savings potential after rebound becomes 0.7–1.1 % per year that is likely feasible and 1.4–2.1 % per year that is potentially possible. Considering actual accomplishments in recent years, the upper ends of these ranges might be higher—perhaps 0.7–1.5 % per year likely feasible and 1.4–2.5 % per year potentially possible. These are broad ranges but these ranges indicate the wide variety of applications and countries as well as the imprecision of many of the estimates.

Conclusions

Countries around the world have made substantial energy efficiency progress but much more cost-effective energy efficiency potential remains. At a minimum, it appears that recent rates of energy efficiency improvement can be sustained for many years. Some studies estimate that even higher rates of energy efficiency improvement can be achieved but to do so would be entering largely uncharted territory.