by Dr. Steven Sorrell
Improved energy efficiency and reduced energy demand are widely expected to provide the dominant contribution to reduced carbon emissions in the short to medium term – and to do so at little or possibly negative cost.
For example, the IEA’s ‘450 scenario‘ has improved energy efficiency, accounting for 71% of emission reductions (relative to the baseline scenario) in the period to 2020, and 48% in the period to 2035 .
However, the link between improved energy efficiency and reduced energy demand (and hence reduced carbon emissions) is not straightforward. The first need not necessarily lead to the second, and both can be interpreted and measured in multiple ways.
Energy efficiency is the ratio of useful outputs to energy inputs for a specified system, be it a motor, an industrial process, a firm, or an entire economy. In all cases, the measure of energy efficiency will depend upon how inputs and outputs are defined and measured. Depending on the system, outputs may be measured in energy terms, such as heat content or physical work; physical terms, such as vehicle kilometres or tonnes of steel; or economic terms such as value-added or GDP.
Different measures may be more or less appropriate in different situations and are unlikely to capture everything of value: for example, vehicle kilometres and passenger kilometres both measure the quantity of mobility, but the former does not capture load factors, the latter does not capture passenger comfort and neither are necessarily correlated with the frequency and ease of access to relevant destinations.
In many cases the most relevant output of a system is an energy service of some form, such as motive power, thermal comfort and accessibility. But energy services are difficult to measure, dependent upon social context and partly subjective, so a different definition, interpretation or understanding of the relevant energy service may lead to a different judgement on the energy efficiency of a particular system.
The measurement of energy inputs also raises issues, especially when different energy carriers are combined. The most common approach is to sum the thermal content of each energy carrier (in joules). But this amounts to summing apples and oranges; energy carriers vary on multiple dimensions (e.g. volumetric and gravimetric energy density, ease of storage and transport, cleanliness) and they are only partially substitutable (try running a truck on battery-stored electricity!).
Higher quality energy carriers receive a higher price since they are more flexible, suitable for a wider range of end uses and produce more economic output per joule. Price-based weighting schemes should therefore be (but rarely are) used to account for the different quality of energy carriers and when this is done aggregate measures of energy efficiency are found to be improving more slowly than is commonly supposed . For example, Kaufmann  shows that much of the reduction in US energy intensity between 1950 and 1990 was linked to the shift towards higher quality and hence more productive energy inputs – such as from coal to oil.
Importantly, improvements in one measure of energy efficiency may not be reflected in improvements in a second measure, or in measures appropriate for a different spatial or temporal boundary. Indeed, it is entirely possible for an improvement in one measure to be associated with deterioration in another. For example, an electric heat pump is more energy efficient than a gas boiler when energy inputs are measured at the building level, but may be less energy-efficient when those inputs are measured at the source level (e.g. the fuel into the power station) or on a life cycle basis.
In a similar manner, improvements in energy efficiency (however measured) may not always reduce energy demand and reductions in energy demand may result from something other than improved energy efficiency. To claim ‘energy-savings’ or ‘demand reduction’ it is necessary to specify the reference against which those savings are measured or estimated. That involves specifying the relevant spatial and temporal boundary and unit of measure, as well as invoking ceteris paribus assumptions.
The reference may be historical energy consumption or a counterfactual scenario of what energy consumption ‘would have been’ in the absence of specified changes. But since data on energy consumption is not always available (or accurate), counterfactuals are unobservable and countervailing variables are difficult to control (for), the causal link between specific changes and the resulting ‘energy savings’ can be hard to establish. Most approaches rely upon decomposition or econometric analysis of secondary data at the aggregate level and the results are frequently lacking in resolution and sensitive to model specification.
Experimental or quasi experimental studies can control for confounding variables at the micro level, but these are costly to conduct and comparatively rare. As a result, the literature is replete with unreliable estimates of historical energy savings and questionable claims about future energy savings – both in relation to specific technologies and policies and in relation to the determinants of aggregate trends.
California is often hailed as an energy efficiency success story since per capita electricity consumption has remained fairly constant since the 1970s and is more than 40% below the US average. A careful analysis of the contributory factors, however, finds that California’s ambitious energy efficiency policies account for less than one third of this difference .
The link between improved energy efficiency and reduced energy demand is further complicated by the presence of multiple rebound effects. Take fuel-efficient cars: they make travel cheaper as consumers may choose to drive further and/or more often, thereby offsetting some of the energy savings achieved. Drivers may also use the savings on fuel bills to buy other goods and services which necessarily require energy to provide – such as laptops made in China and shipped to the UK. Reductions in fuel demand will translate into lower fuel prices which in turn will encourage increased fuel consumption elsewhere.
Similar mechanisms exist in industry, where cost-effective energy efficiency improvements allow firms to expand output, lower product prices and increase market demand which in turn stimulates economic growth and aggregate energy consumption. In some cases, energy efficient innovations may lead to new, unforeseen energy-using applications, products and industries. The Bessemer process, for instance, greatly improved the energy efficiency of steel-making, but also produced cheaper and higher quality steel suitable for a wider range of uses, thereby increasing demand for both steel and coal.
Rebound is therefore an emergent property of complex economic systems, with the multiple mechanisms and effects being difficult to isolate and measure, especially over the longer term. However, a growing body of evidence suggests that these effects are larger than was previously thought and can frequently offset or even eliminate the energy savings from improved energy efficiency [5,6].
From an engineering perspective, energy demand may be reduced by improving the thermodynamic efficiency of energy conversion devices such as boilers and engines; preserving, heat, light, momentum or materials in passive systems, such as houses, cars and steel bars; or reducing demand for final energy services such as thermal comfort and mobility .
Gas use for home heating may be reduced by installing a more efficient boiler, insulating the walls or roof, or accepting lower internal temperatures; petroleum use for car travel may be reduced by improving the efficiency of the engine, reducing the size, weight, rolling resistance and/or air resistance of the vehicle; or simply driving less; and coal use for steel manufacture may be reduced by improving the efficiency of blast furnaces, increasing scrap recovery and product life, or designing buildings and products to use less steel.
Globally, Cullen et al [8,9] estimate that that global average conversion losses could be reduced by a maximum of 89% and passive systems losses by a maximum of 73%, implying that current demand for energy services could be provided with much lower energy consumption. This is a theoretical potential, so the technical and (especially) economic potential is likely to be much less. Also, the rate at which improvements in conversion efficiency or passive systems can be achieved is constrained by the rate of turnover of the relevant capital stock.
Further reductions in energy demand may be achieved by reducing demand for the relevant energy services (‘sufficiency’), but growing incomes create strong pressures in the opposite direction. This is particularly the case for countries at earlier stages of industrial development, but also applies more generally: for example, an analysis of lighting demand over three centuries and six continents finds no evidence of saturation even in the wealthiest countries . Changes in demand for energy services can often occur fairly rapidly, but these too may be constrained by the lifetimes of relevant technologies and infrastructures . For example, the physical characteristics and spatial location of houses, workplaces and other assets can lock-in heating, cooling and mobility needs for decades. More generally, voluntary actions to reduce any form of consumption face multiple obstacles within a growth-based economy .
In sum, equating improved energy efficiency with reduced energy demand is misleading, while reducing energy service demand involves swimming against a strong tide. This does not mean that energy demand cannot be reduced, but does imply that it will be more challenging than many analyses, policy documents and political statements suggest.
Dr. Steven Sorrell is a Senior Lecturer in the Science Policy Research Unit (SPRU) at the University of Sussex. He is an energy and climate policy specialist with more than 20 years of experience in academic and consultancy research.
Read Steve’s full argument in his recent SPRU working paper: Reducing energy demand: issues, challenges and approaches
- IEA World Energy Outlook 2012; International Energy Agency: Paris, 2012.
- Zarnikau, J., Will tomorrow’s energy efficiency indices prove useful in economic studies? The Energy Journal 1999, 20, 139-145.
- Kaufmann, R.K., A biophysical analysis of the energy/real GDP ratio: implications for substitution and technical change. Ecological Economics 1992, 6, 35-56.
- Sudarshan, A.; Sweeney, J., Deconstructing the ‘Rosenfeld Curve’. Precourd Institute for Energy Efficiency, Stanford University 2008, 38.
- Jenkins, J.; Nordhaus, T.; Shellenberger, M. Energy emergence: rebound and backfire as emergent phenomena; Breakthrough Institute: New York, USA, 2011.
- Sorrell, S.; Dimitropoulos, J. UKERC Review of Evidence for the Rebound Effect: Technical Report 3: Econometric studies; UK Energy Research Centre: London, 2007.
- Cullen, J.M.; Allwood, J.M., The efficient use of energy: Tracing the global flow of energy from fuel to service. Energy Policy 38, 75-81.
- Cullen, J.M.; Allwood, J.M., Theoretical efficiency limits for energy conversion devices. Energy 2010, 35, 2059-2069.
- Cullen, J.M.; Allwood, J.M.; Borgstein, E.H., Reducing energy demand: what are the practical limits? Environ. Sci. Technol. 2011, 45, 1711-1718.
- Lenski, S.M.; Keoleian, G.A.; Bolon, K.M., The impact of’Cash for Clunkers’ on greenhouse gas emissions: a life cycle perspective. Environmental Research Letters 2010, 5, 044003.
- Tsao, J.Y.; Waide, P., The World’s Appetite for Light: Empirical Data and Trends Spanning Three Centuries and Six Continents. LEUKOS 2010, 6, 259-281.
- Guivarch, C.; Hallegatte, S., Existing infrastructure and the 2 C target. Climatic Change 2011, 109, 801-805.
- Sorrell, S., Energy, economic growth and environmental sustainability: five propositions. Sustainability 2010, 2, 1784-1809.