C ONTEXT AND RATIONALE
To reduce carbon emissions, governments are focusing on enhancing energy efficiency across the economy, which is expected to lower overall energy consumption However, rebound effects can diminish the anticipated energy savings, as evidence suggests that some energy-efficient technologies have historically led to increased energy demand This phenomenon is especially evident with widespread innovations, like 19th-century steam engines, which not only improved energy efficiency but also boosted overall economic productivity.
Rebound effects from energy efficiency improvements can significantly impact energy and climate policy in the UK and worldwide Although enhancing energy efficiency is generally beneficial for welfare and the economy, it may sometimes hinder climate change efforts Nonetheless, not all energy efficiency advancements lead to increased overall energy consumption, especially those prompted by policy initiatives.
The ongoing debate in energy economics centers on the nature, operation, and significance of rebound effects, particularly regarding whether enhancements in energy efficiency lead to a proportional decrease in energy consumption For instance, a 20% increase in passenger car fuel efficiency does not necessarily result in a 20% reduction in fuel usage for personal travel Economic theory posits that as energy efficiency improves, the marginal cost of energy services decreases, potentially increasing demand for those services Consequently, lower driving costs may encourage consumers to drive more frequently or over longer distances, thereby diminishing the anticipated reduction in overall energy consumption.
The direct rebound effect, first highlighted by energy economist Daniel Khazzoom in 1980, has been extensively researched (Greening et al., 2000b) Even if the direct rebound effect is absent for a specific energy service—such as consumers not increasing their driving distance with a fuel-efficient vehicle—there are still various factors that can lead to a smaller-than-expected reduction in overall energy consumption For instance, savings from reduced fuel costs may be redirected towards activities like overseas travel, thus impacting kerosene usage These indirect rebound effects can manifest in different ways, as summarized in Box 1.1 Both direct and indirect rebound effects are relevant to energy efficiency improvements made by consumers, such as purchasing more fuel-efficient cars, as well as those made by producers, like adopting energy-efficient motors in machinery.
The manufacturing and installation of energy efficiency equipment, such as thermal insulation, consume energy, which can offset the energy savings gained from its use.
Energy efficiency improvements can lead to significant cost savings for consumers, which they may then redirect towards purchasing other goods and services that also require energy For instance, savings from upgrading to a more energy-efficient central heating system could be invested in an overseas holiday, illustrating the broader impact of energy efficiency on consumer spending.
Producers can leverage cost savings from energy efficiency enhancements to boost production, potentially counterbalancing some of the energy savings achieved When these improvements are implemented across an entire sector, they can result in reduced product prices, heightened demand for those products, and ultimately, an increase in overall energy consumption.
Cost-effective energy efficiency enhancements boost overall economic productivity, fostering economic growth As the demand for goods and services rises, this can lead to an increase in energy consumption.
Lower energy demand can lead to decreased energy prices, which in turn may promote higher energy consumption Additionally, the drop in energy prices boosts real income, fostering investment and providing an extra impetus for both aggregate output and energy usage.
Improvements in energy efficiency and the resulting decrease in energy prices will significantly lower the costs of energy-intensive goods and services compared to non-energy-intensive options, leading to an increased consumer demand for the former.
The economy-wide rebound effect from energy efficiency improvements encompasses both direct and indirect impacts, typically expressed as a percentage of anticipated energy savings A 100% rebound effect indicates that the expected energy savings are fully negated, resulting in no net savings While often analyzed within the context of a national economy, these effects can also influence other countries through shifts in trade patterns and international energy prices As markets, technology, and behaviors evolve, the significance of rebound effects is likely to increase over time Ultimately, from a climate change perspective, the long-term implications for global energy consumption are of paramount importance.
The Rebound Effect Balance (REB) can be calculated using the formula REB = [(DIR + IND) / ENG] * 100% In this equation, ENG denotes the anticipated energy savings from a specific energy efficiency improvement, excluding rebound effects DIR refers to the rise in energy consumption due to the direct rebound effect, while IND accounts for the increase in energy consumption stemming from indirect rebound effects.
The economy-wide rebound effect encompasses various complex and interdependent mechanisms that differ in significance based on the type of energy efficiency improvement Consequently, accurately estimating the magnitude of this rebound effect in specific cases is challenging However, it is generally understood that the actual energy savings resulting from energy efficiency enhancements are likely to be lower than commonly believed.
The concept that enhancements in energy efficiency may lead to increased energy consumption was initially proposed by British economist William Stanley Jevons in 1865 This idea has since been recognized as the 'Khazzoom-Brookes (K-B) postulate,' named after contemporary economists Len Brookes and Daniel Khazzoom who further explored this notion The K-B postulate suggests that improvements in energy efficiency can paradoxically result in higher overall energy use.
‘with fixed real energy prices, energy efficiency gains will increase energy consumption above what it would be without these gains’ (Saunders, 1992)
The K-B postulate raises significant concerns regarding energy efficiency policies, suggesting they may not effectively reduce energy consumption or carbon emissions, challenging the conventional beliefs held by analysts, policymakers, and the public Even if the postulate is proven incorrect, the mechanisms involved could still undermine the effectiveness of these policies Consequently, the rebound effect poses serious implications for global climate change efforts, yet it remains largely overlooked and surrounded by controversy and confusion.
H OW THE ASSESSMENT WAS CONDUCTED
The UK Energy Research Centre (UKERC) is conducting an assessment on the rebound effect as part of its Technology and Policy Assessment (TPA) function This topic was chosen after extensive consultations with stakeholders and guidance from the TPA Advisory Group, which highlighted ongoing debates and varying opinions on the rebound effect The Group pointed out the significant gap between the topic's potential importance and the limited research available, stressing the need to examine rebound effects comprehensively, including indirect and economy-wide impacts Consequently, UKERC initiated this study to address these concerns.
This assessment aims to review the existing knowledge on rebound effects rather than conducting new research It seeks to present the findings in a clear and accessible way that benefits analysts and policymakers.
The assessment utilized systematic review techniques commonly used in medicine, starting with a Scoping Note that defined the rebound effect and evaluated the evidence base (Sorrell and Dimitropoulos, 2005b) However, the complexity and contested nature of the evidence made traditional systematic review methods challenging to implement, a common issue in energy policy-related inquiries (Sorrell, 2007).
An Advisory Group was formed for the project, leading to the distribution of a Scoping Note to key stakeholders, which resulted in recommendations regarding the assessment's scope and focus The established Assessment Protocol (Sorrell and Dimitropoulos, 2005a) aimed to expand upon, rather than replicate, Greening et al.'s previous literature review (2000a), emphasizing indirect effects over direct ones It also prioritized clarifying conceptual frameworks and understanding the reasons behind differing opinions on the rebound effect Given the technical nature of the economic and econometric concepts involved, significant effort was dedicated to making these ideas accessible to a non-technical audience.
Box 1.1 Overview of the TPA approach
The TPA assessment approach utilizes Evidence Based Policy and Practice (EBPP) techniques, such as meta-analyses and systematic reviews, to enhance the evidence base for policymakers and practitioners This method aims to minimize research duplication, promote higher research standards, and highlight research gaps However, the application of systematic review techniques in energy policy faces challenges and has been criticized for its rigid methodologies (Sorrell, 2007) In response, the UKERC has developed a flexible process inspired by these methods, allowing for a broader range of techniques.
The process carried out for each assessment includes the following components:
Publication of Scoping Note and Assessment Protocol.
Establishment of a project team with a diversity of expertise.
Convening an Expert Group with a diversity of opinion and perspective.
Systematic searches of the the evidence base.
Categorisation and assessment of evidence.
Expert feedback on initial drafts.
Peer review of final draft.
The assessment was organised around six broad categories of evidence, as follows:
Evaluation studies: micro-level evaluations of the impact of specific energy efficiency improvements on the demand for energy or energy services;
Econometric studies: use of secondary data sources to estimate the elasticity of the demand for energy or energy services at different levels of aggregation;
Elasticity of substitution studies: estimates of the elasticity of substitution between energy and other factors of production at different levels of aggregation;
Computable general equilibrium modelling studies: estimates of economy-wide rebound effects from computable general equilibrium (CGE) models of the macroeconomy; and
Research on energy, productivity, and economic growth encompasses various studies that align with the K-B postulate This includes insights from economic history, neoclassical production and growth theories, ecological economics, as well as decomposition and input-output analyses.
The article discusses two main categories of evidence concerning direct rebound effects, while the remaining categories focus on economy-wide rebound effects, with the latter receiving significant attention and diversity Initially, the project employed systematic literature searches using keywords like "rebound effect," but ultimately, it expanded to include a wider array of evidence, reviewing over 500 studies globally The comprehensive findings of this assessment are detailed in six extensive Technical Reports.
Technical Report 1: Evidence from evaluation studies
Technical Report 2: Evidence from econometric studies
Technical Report 3: Evidence from elasticity of substitution studies
Technical Report 4: Evidence from CGE modeling studies
Technical Report 5: Evidence from energy, productivity and economic growth studies
In addition, there is a shorter Supplementary Note that provides a graphical analysis of rebound effects All these reports are available to download from the UKERC website
The present report summarises the main conclusions of the assessment in a non-technical manner, identifying priorities for further research and major policy implications.
R EPORT STRUCTURE
The report is structured as follows.
Section 2 explores various definitions of energy efficiency and examines the options for independent and dependent variables related to the rebound effect, highlighting how these choices impact the estimated magnitude of the effect The article contends that disagreements regarding the size of rebound effects stem partly from misunderstandings of these fundamental definitions.
Section 3 examines the nature and impact of direct rebound effects, highlighting that these effects tend to be low to moderate (under 30%) for various household energy services in developed countries While direct rebound effects may be more significant in developing nations and for energy efficiency enhancements made by producers, the empirical evidence supporting these claims remains limited.
Section 4 explores the characteristics and functioning of indirect and economy-wide rebound effects, summarizing various studies that offer quantitative estimates of these phenomena The existing evidence is limited and presents significant methodological flaws, complicating the ability to reach broad conclusions Nonetheless, the findings indicate that economy-wide rebound effects often surpass 50%.
Section 5 highlights the evidence supporting the K-B postulate, particularly through the contributions of W.S Jevons, Len Brookes, and Harry Saunders Rather than offering quantitative estimates of rebound effects, this section presents indirect evidence from various fields, including neoclassical growth theory, economic history, and decomposition analysis A key argument is that energy's role in economic growth is underestimated in conventional views The discussion contrasts traditional perspectives with those of ecological economics, suggesting that resolving the debate over the K-B postulate may depend on addressing this broader issue regarding energy's significance in economic development.
Finally, Section 6 summarises the key conclusions from the assessment, identifies the main research needs and highlights some important policy implications.
I NTRODUCTION
Energy efficiency improvements are expected to lower energy consumption compared to previous levels However, the rebound effect can diminish the actual energy savings achieved Accurately estimating the extent of these energy savings is a complex challenge.
Real-world economies lack the ability to conduct controlled experiments, making it challenging to directly assess how changes in energy efficiency influence energy consumption This relationship is often influenced by various confounding factors that can obscure the true impact of energy efficiency improvements.
The uncertainty surrounding the estimated savings from energy efficiency improvements arises from the inability to observe the counterfactual scenario—what energy consumption would have been without these enhancements.
Energy efficiency is affected by various technical, economic, and policy factors, as it is not externally regulated by an experimenter Notably, changes in energy consumption, regardless of their origins, can also impact energy efficiency, suggesting a potential reverse causality.
Energy efficiency is defined as the ratio of useful outputs to energy inputs within various systems, including devices like boilers, buildings, industrial processes, firms, sectors, or entire economies The determination of energy efficiency hinges on the definitions of 'useful' and the methods used to measure inputs and outputs When outputs are assessed in thermodynamic or physical terms, the term "energy efficiency" is applied; conversely, when outputs are evaluated economically, the term "energy productivity" is preferred The inverse of these measures is referred to as "energy intensity."
Enhancing energy efficiency, such as by installing a condensing boiler, can significantly lower energy consumption, particularly in heating systems However, it is crucial to consider the broader economic implications of these improvements, as savings on heating expenses may lead to increased spending on other energy-consuming goods and services.
The definitions of inputs and outputs, as well as the suitable 'system boundaries' for assessing energy efficiency and consumption, can differ significantly across various studies Consequently, the conclusions about the size and significance of the rebound effect are likely influenced by the specific choices made in each analysis.
This section explores various definitions of energy efficiency and the methods for measuring energy inputs and useful outputs It highlights that these definitional complexities significantly impact the understanding of rebound effects, which are often overlooked in discussions about energy efficiency.
M ULTIPLE DEFINITIONS
Energy efficiency is fundamentally defined by the first law of thermodynamics, which measures the ratio of useful energy outputs to the calorific value of fuel inputs For instance, a conventional lightbulb has a first-law efficiency of only 6%, meaning that only 6% of the energy input is converted into light, with the rest lost as waste heat However, the definition of 'useful' energy impacts this efficiency measurement When considering waste heat and other losses, the first-law efficiency can be viewed as 100%, as energy is transformed rather than consumed.
First-law efficiency measures can be misleading as they overlook the availability of energy inputs and outputs and their capacity to perform useful work For instance, high-pressure steam can generate more useful work compared to the same energy amount in low-temperature heat The concept of exergy offers a comprehensive assessment of the ability to perform useful work, applicable to both inputs and outputs in conversion processes Notably, the exergy content of an energy carrier differs significantly from its heat content, despite both being measured in kilowatt-hours For example, a unit of electricity has a higher exergy ranking than an equivalent unit of oil or natural gas Additionally, unlike energy, exergy is consumed during conversion processes.
Exergy provides a second definition of energy efficiency based on the second law of thermodynamics, often revealing a greater potential for improvement compared to first-law efficiency For instance, while electric resistance space heating may achieve a first-law efficiency exceeding 99%, its second-law efficiency drops to approximately 5% due to the conversion of high exergy electricity into low exergy space heat This second-law measure is advantageous as it emphasizes the importance of conserving exergy rather than energy alone.
Measuring useful energy outputs through physical indicators is often more straightforward than using heat content or exergy For instance, in personal transportation by private car, vehicle kilometres serve as an appropriate output measure Consequently, energy efficiency can be quantified as vehicle kilometres per litre of motor fuel.
Physical measures of energy efficiency are often applied at higher levels, such as industrial processes or specific sectors, rather than at the level of individual energy conversion devices These measures can be influenced by various factors beyond the thermodynamic efficiency of the devices themselves For instance, changes in vehicle load factors can significantly affect energy efficiency metrics, such as passenger kilometres per litre of fuel Additionally, suitable physical indicators differ across sectors and types of energy services, rendering a comprehensive economy-wide measure of energy efficiency ineffective.
Work can be defined as the increase in kinetic, potential, physical, or chemical energy within a subsystem of a larger system, adhering to the principle of energy conservation as stated in the first law of thermodynamics.
By utilizing economic value indicators in place of traditional numerators, we can effectively compare the energy efficiency across various sectors, such as brewing and dairy, by measuring value added per GWh of energy input Transitioning from physical to economic indicators broadens the range of influencing factors, especially when applied at higher levels of aggregation Consequently, the indicator that deviates most from a thermodynamic assessment of energy efficiency is the ratio of GDP to total primary energy consumption within a national economy.
When discussing energy efficiency and energy productivity, it is crucial to define the measurement of inputs and outputs In this context, the term "useful work" will be used to describe the beneficial outputs of energy conversion systems, except when referring to economic assessments of energy efficiency or productivity.
Maximizing energy efficiency alone is an inadequate objective, as it overlooks the costs related to capital and labor inputs Economists prioritize enhancing total factor productivity (TFP), which assesses how effectively a firm, sector, or national economy utilizes all its resources.
Energy efficiency improvements relevant to the rebound effect are categorized into two types: those linked to total factor productivity (technical change) and those not linked (substitution) Technical change is typically viewed as independent of relative price changes, while substitution responds to such shifts The rebound effects can be related to both categories, but the impact of technical change is particularly significant as it drives economic growth Additionally, many energy efficiency advancements often arise as a by-product of efforts to enhance total factor productivity rather than from direct initiatives aimed at boosting energy efficiency.
Technical change is classified as 'neutral' when it equally reduces the use of all inputs, whereas it is considered 'biased' if it affects some inputs more than others 'Energy-saving' technical change decreases the proportion of energy costs in output value more significantly than other inputs, while 'energy-using' technical change has the opposite effect This bias in technical change is intricately linked to the growth rate of energy efficiency over time, assuming constant relative prices, a crucial factor in various energy-economic models.
The traditional distinction between substitution and technical change can be deceptive, as it presumes that the economy is utilizing all inputs efficiently Moreover, technical change is not independent; rather, it is affected by shifts in relative prices (Grubb et al., 2002).
Box 2.1 Substitution and technical change
Rising energy prices can drive enhancements in energy efficiency by encouraging the replacement of energy with capital or labor, such as using thermal insulation instead of gas for heating However, if the costs of other inputs remain stable, the overall production costs for a specific output level will rise Conversely, advancements in energy productivity and total factor productivity can stem from technical innovations, which are typically viewed as beneficial since they occur without affecting economic output, regardless of changes in relative prices.
In neoclassical economics, a production function illustrates the maximum output (Y) achievable from energy (E) and other inputs (X) based on current technology An 'isoquant' reflects various input combinations for a specific output level, with the optimal mix influenced by input prices Changes in relative prices can lead to substitutions along the isoquant, often requiring investments in existing technologies like energy-efficient motors Conversely, technical change involves the introduction of new technologies and organizational methods that shift the isoquant leftward, enabling the same output with fewer inputs.
The concept of substitution suggests a seamless transition between techniques; however, this process necessitates time and investment The production function illustrates the optimal blend of factor inputs, yet companies often utilize less efficient combinations Significant investments may lead to a shift from these inefficient methods to more efficient ones, all while staying within the production function's frontier.
‘overcoming inefficiency’ arrow in the diagram.
U NACKNOWLEDGED IMPLICATIONS
The significance of definitional issues in estimating the rebound effect is frequently overlooked, as many assume that improvements in thermodynamic efficiency are the key independent variable However, such enhancements only lead to comparable gains in energy efficiency if certain rebound effect mechanisms do not activate For instance, a reduction in fuel consumption per vehicle kilometer may not translate to lower fuel use per passenger kilometer if vehicle load factors change Numerous studies on rebound effects utilize broad measures of energy efficiency, potentially neglecting the rebound effects linked to improvements in narrower, thermodynamic efficiency metrics Furthermore, advancements in these aggregate measures are unlikely to stem solely from the adoption of more efficient conversion devices.
Rebound effects are anticipated to grow over time and expand with the system boundary for the dependent variable While the K-B postulate typically considers the national economy as the relevant system boundary, energy efficiency advancements can influence trade patterns and international energy prices, impacting energy consumption in other nations To evaluate the role of energy efficiency in reducing carbon emissions, a global system boundary is essential However, measuring the impact of energy efficiency improvements on trade patterns presents significant challenges (Allan et al., 2006).
Energy efficiency measures are influenced by how various energy inputs are combined, typically aggregated by heat content However, this method overlooks each energy type's capacity to perform useful work, which is crucial for determining economic productivity Other important factors include cleanliness, storage capability, safety, and flexibility of use, along with the specific applications of the energy (Cleveland et al., 2000) In a market free from significant distortions, the relative price per kilowatt hour of different energy carriers can serve as a general indicator of their marginal productivity (Kaufmann, 1994).
Economists have developed methods to aggregate different types of energy based on relative prices and marginal productivity (Berndt, 1978) When considering the quality of energy inputs, it becomes evident that improvements in aggregate energy efficiency are slower than often believed (Hong, 1983; Zarnikau, 1999; Cleveland et al., 2000) For instance, from 1970 to 1991, per-capita energy consumption in the US residential sector reportedly decreased by 20% based on thermal inputs However, when factoring in changes in energy quality, particularly the rising use of electricity, per-capita energy consumption actually increased by 7% (Zarnikau et al., 1996) This highlights that advancements in energy utilization extend beyond mere thermodynamic efficiency improvements.
The marginal product of energy input in production refers to the additional value generated by using one more unit of energy, such as heat By replacing low-quality fuels with higher-quality alternatives, particularly electricity, producers can enhance output while maintaining the same heat content in their energy inputs.
Neglecting the impact of energy inputs on their marginal productivities can lead to erroneous conclusions regarding the relationship between GDP growth and primary energy consumption While it is often believed that OECD countries have successfully decoupled GDP growth from energy consumption through structural changes and energy efficiency improvements, a closer analysis reveals a strong correlation between the two when energy inputs are appropriately weighted This suggests that the perceived decoupling may be misleading, and the role of enhanced thermodynamic efficiency in this process may be exaggerated.
Improvements in energy efficiency and fuel switching can significantly influence overall energy consumption, but the rebound effects vary between these strategies Energy efficiency advancements rarely occur in isolation; they often coincide with enhanced productivity from other inputs, driven by new technologies For instance, research by Pye and McKane (1998) illustrates that energy-efficient motors not only lower energy costs but also reduce wear and tear, extend component lifespans, and decrease capital and labor expenses, leading to greater savings Consequently, when evaluating the total impact of energy-efficient technologies on consumption, the rebound effect may be larger if all benefits are considered Conversely, attributing only part of this impact to energy efficiency could result in a smaller rebound effect Ultimately, understanding how energy-efficient technologies influence overall energy demand is crucial for effective policy-making, making this a key focus of the assessment.
S UMMARY
Energy efficiency can be assessed through various methods and system boundaries, each serving as a potential independent variable for estimating the rebound effect The selection of the most suitable approach is contingent upon the study's goals, but challenges may emerge if the term is misinterpreted by commentators.
‘energy efficiency’ in different ways.
The rebound effect is characterized by changes in energy consumption, which can be assessed across various system boundaries Expanding these boundaries enables a more comprehensive understanding of the diverse rebound effects that may occur.
Numerous theoretical and empirical studies on rebound effects typically utilize broad physical or economic measures of energy efficiency However, these approaches may fail to capture the rebound effects associated with enhancements in physical or thermodynamic energy efficiency that are relevant to more confined system boundaries.
Since the early 1970s, major OECD countries have reduced their primary energy consumption by one third per unit of GDP when measured by heat content (Geller et al., 2006).
Numerous theoretical and empirical studies aggregate various energy carriers based solely on their thermal content, often ignoring their economic productivity differences This oversight can result in a failure to recognize how shifts in fuel mix impact overall energy efficiency metrics When energy inputs are assessed according to their marginal productivities, the historical increase in energy consumption shows a strong correlation with GDP growth.
Economists focus on energy efficiency improvements that optimize the use of economic resources, typically classified into price-induced substitution and technical change However, this perspective overlooks the significant impact of organizational and market failures on inefficiencies, as well as the influence of relative prices in driving technical advancements.
Improvements in energy efficiency by producers can enhance the productivity of capital, labor, and material inputs Likewise, consumers who improve their energy efficiency often experience cost savings beyond just energy expenses It is crucial to consider these factors when estimating rebound effects and changes in fuel mix, as neglecting them may result in an underestimation of rebound effects and an overestimation of the decoupling between energy consumption and economic output.
3 Evidence for direct rebound effects
I NTRODUCTION
This section provides a summary of empirical evidence regarding direct rebound effects, specifically emphasizing energy services within the household sector, which has been the primary focus of research For a comprehensive analysis of this evidence, please refer to Technical Reports 1 and 2.
Section 3.2 explores the dynamics of the direct rebound effect, emphasizing critical aspects related to its measurement and the circumstances that influence its magnitude Additionally, a Supplementary Note provides a visual analysis of rebound effects, enhancing the insights presented in this section.
Section 3.3 outlines a methodology for estimating direct rebound effects through quasi-experimental techniques borrowed from policy evaluation It also provides a summary of findings from several studies that have applied this approach to assess the direct rebound effects resulting from enhancements in energy efficiency for space heating.
Section 3.4 presents a widely used method for estimating direct rebound effects through econometric analysis of secondary data, focusing on the elasticity of demand for useful work or energy consumption Sections 3.5 and 3.6 summarize findings from various studies that apply this method to assess direct rebound effects in personal automotive transport, household heating, and select household energy services However, Section 3.7 identifies potential biases in this approach that could result in an overestimation of the direct rebound effect The article concludes in Section 3.8.
U NDERSTANDING THE DIRECT REBOUND EFFECT
Direct rebound effects occur when enhanced energy efficiency in services like heating, lighting, and refrigeration leads to increased consumption due to lower marginal costs For instance, after acquiring an energy-efficient vehicle, consumers might drive more because the cost per kilometer has decreased This behavior results in a rise in energy demand, which can counterbalance the anticipated reduction in consumption from the efficiency gains.
The Supplementary Report reveals that the direct rebound effect for consumers consists of a 'substitution effect', where cheaper energy services replace other goods and services, and an 'income effect', which allows for increased overall consumption due to higher real income Likewise, producers experience a direct rebound effect that includes both substitution and an 'output' effect Over time, as markets, technologies, and consumer behaviors adapt, the magnitude of these direct rebound effects is likely to grow For instance, while energy efficiency improvements can reduce a firm's production costs, it may take time for the firm to expand its output and capture a larger market share.
Energy services involve a blend of capital equipment, labor, materials, and energy, with a key aspect being the useful work they provide, which can be quantified in various ways such as vehicle kilometers or passenger kilometers While all cars offer passenger kilometers, they differ significantly in features like speed, comfort, and prestige This variability allows consumers and producers to balance useful work against other desirable attributes of an energy service, as well as to weigh energy, capital, and other market goods in the production process and consider different types of energy services.
Improvements in energy efficiency can reduce the marginal cost of energy services, potentially leading to increased consumption of energy conversion devices, larger sizes, higher utilization rates, and greater load factors over time For instance, consumers may opt for more and larger cars, drive longer distances, or purchase more washing machines with larger capacities and increased usage frequency The impact of these changes will vary across different energy services and time periods While advancements in refrigerator efficiency may not significantly boost their average usage, they could result in a rise in both the quantity and size of refrigerators due to reduced costs Over the long term, lower energy service costs may drive transformative changes in technologies, infrastructure, and lifestyles, such as a shift towards car-dependent commuting and greater distances between homes, workplaces, and retail areas However, as the time frame extends, distinguishing these effects from those influenced by income growth and other factors becomes increasingly challenging.
The direct rebound effect's estimated size varies based on the definition of energy efficiency, particularly in personal automotive transport, where it is often measured by vehicle kilometres travelled This measurement typically considers the number of vehicles and the average distance each car travels annually However, significant factors such as the rise in average vehicle size and weight due to energy efficiency improvements, like the popularity of SUVs, and the decline in average load factors, such as reduced car sharing, have been largely ignored, despite their potential importance.
The extent of direct rebound effects is likely to be proportional to the proportion of energy costs within the overall expenses associated with energy services.
When evaluating energy efficiency through fuel consumption per vehicle kilometre, rebound effects manifest primarily as increased driving distances Conversely, if energy efficiency is assessed based on fuel consumption per tonne kilometre, rebound effects are reflected in a rise in tonne kilometres driven, calculated by multiplying the distance travelled by the average vehicle weight.
Improving energy efficiency can significantly impact total service costs, depending on the proportion of energy costs For instance, if energy comprises 50% of total costs, enhancing efficiency can lower overall expenses by 25% Conversely, if energy only represents 10% of total costs, the reduction in total expenses would be a mere 5% It's important to note that while energy efficiency improvements are beneficial, they can also entail substantial costs.
As the consumption of energy services rises, saturation effects, or declining marginal utility, can diminish the direct rebound effect For instance, improvements in the energy efficiency of household heating systems may lead to reduced rebound effects once indoor temperatures reach optimal comfort levels This suggests that lower-income groups may experience higher direct rebound effects, as they typically consume energy services at levels further from saturation.
Rising demand for energy services can stem from both current users and new consumers who previously hesitated to make a purchase For instance, enhanced energy efficiency in home air-conditioners may motivate first-time buyers to invest in portable units The significant presence of these 'marginal consumers' in developing nations suggests a potential for substantial rebounds in energy consumption, which may be somewhat mitigated by saturation effects among existing users.
Improvements in energy efficiency can lower energy service costs, but the extent of the direct rebound effect varies based on other cost factors For instance, if energy-efficient equipment is pricier than its less efficient counterparts, the rebound effect may be minimized, as efficiency gains shouldn't lead to increased usage or capacity of these devices However, once acquired, these devices often see higher utilization rates In reality, many types of equipment have not only become more energy-efficient over time but have also decreased in total cost relative to income.
Energy efficiency improvements can be limited by real or opportunity costs related to increased demand, such as the opportunity cost of space and time For instance, opting for a larger refrigerator may not optimize available space, while driving longer distances may not be the best use of time However, advancements in technology could reduce the size of appliances, and rising incomes might lead to larger living spaces, potentially diminishing space constraints over time Conversely, as incomes rise, the opportunity cost of time is likely to increase Consequently, the direct rebound effect for specific energy services is expected to fluctuate over time, influenced by income levels and other factors.
E STIMATING DIRECT REBOUND EFFECTS FROM ‘ EVALUATION ’ STUDIES
To estimate the direct rebound effect, one effective method is to assess the change in demand for 'useful work' after implementing energy efficiency improvements, such as evaluating the shift in internal temperatures after a fuel-efficient boiler is installed.
Kempton and Montgomery (1982) found that the average household energy bill provides less information than a single monthly supermarket bill for food However, advancements in smart metering and electronic display technologies can enhance the information available to consumers, potentially increasing the price elasticity of demand for household energy services.
Estimating the demand for useful work prior to energy efficiency improvements can serve as a benchmark for what demand might have been without these enhancements However, it is essential to account for other influencing factors that may have also altered the demand for useful work.
Measuring useful work for energy services can be challenging, so an alternative method involves assessing the change in energy consumption after implementing energy efficiency improvements To accurately estimate direct rebound effects, this change must be compared to a counterfactual scenario, which includes two potential errors: the energy consumption that would have occurred without the efficiency improvement and the consumption expected post-improvement without any behavioral changes The first scenario helps estimate energy savings, while the second isolates the rebound effect Although engineering models can provide estimates for the latter, they require detailed data on specific installations and are often subject to inaccuracies.
A limited number of studies have estimated direct rebound effects for household energy services, with Nadel (1993) indicating that US utilities report rebound effects of 10% or less for lighting and negligible effects for water heating, while refrigeration results remain inconclusive However, access to these studies has been challenging due to their small-scale, short-term nature and methodological weaknesses Consequently, we focus on summarizing the findings from 15 published studies concerning household heating.
Recent studies on energy efficiency improvements in households across various income groups in the US and UK reveal significant methodological flaws Most research relies on simplistic before-and-after comparisons without control groups, leading to potential biases Additionally, selection bias is prevalent as households self-select into studies rather than being randomly assigned Despite the availability of techniques to mitigate these biases, they are seldom employed Other notable weaknesses include small sample sizes, high variance in results, inconsistent reporting of statistical errors, and inadequate monitoring periods that fail to capture long-term behavioral changes These limitations diminish confidence in the findings and complicate comparisons across different studies.
For household heating, it is helpful to distinguish between:
shortfall, representing the difference between actual savings in energy consumption and those expected on the basis of engineering estimates;
temperature take-back, representing the change in internal temperature following the energy efficiency improvement; and
12 The power of a statistical test is the probability that the test will reject a false null hypothesis
behavourial change, representing the proportion of change in internal temperature that derives from adjustments of heating controls and other variables by the user (e.g opening windows)
Temperature take-back is influenced by various factors, with only a fraction attributed to behavioral change, as noted by Sanders and Phillipson (2006) Similarly, the shortfall in expected savings can stem from temperature take-back, but is also significantly impacted by poor engineering estimates, equipment performance issues, and installation deficiencies Therefore, while behavioral change contributes to temperature take-back, it is not the sole or most critical factor, and the same applies to the shortfall phenomenon Many studies inaccurately equate shortfall solely with behavioral change, neglecting to address the associated uncertainties.
Direct rebound effects can be seen as behavioral responses to reduced energy service prices, but interpreting this change solely as a rational reaction to lower heating costs may be misleading Energy efficiency improvements often alter other factors, such as airflow, which can also drive behavioral changes Additionally, measuring temperature take-back may pose challenges in translating these into shortfall estimates due to the non-linear and household-specific relationship between energy consumption and internal temperature.
Measuring behavioral responses in thermostat settings is complex and uncommon, leading most studies to focus on monitoring internal temperatures and energy consumption instead These studies often emphasize concepts like temperature take-back and shortfall, which hold greater significance for public policy However, inconsistencies arise as different studies use varying terminology for similar concepts, and the same terms can refer to different ideas For instance, the UK government's use of the term "comfort factor" to describe the shortfall resulting from household insulation improvements may be misleading when equated with the direct rebound effect.
Improving thermal insulation in a home typically leads to an increase in daily average household temperatures, even without changes to heating controls Enhanced insulation promotes a more consistent warmth distribution, slows down the cooling rate when heating is off, and extends the duration before the heating system needs to be activated again.
A study by Hong et al (2006) utilized thermal camera imaging to assess homes after insulation upgrades, revealing that 20% of the cavity wall area and 13% of the loft area lacked proper insulation.
Box 3.1 Measurement issues for the direct rebound effect
Estimating direct rebound effects poses significant challenges due to measurement difficulties, often rendering it problematic or even impossible Many energy services lack relevant data, while for others, the available data requires estimation or is prone to substantial errors.
Measuring useful work presents significant challenges, which is why much of the existing research focuses on passenger transport and household heating and air-conditioning, where relevant proxies exist Aggregate data for useful work is limited to specific energy services, such as vehicle kilometers for passenger transport, while detailed measurements often necessitate costly monitoring of individual households or firms In household heating studies, researchers have tracked thermostat settings and internal temperatures, but to fully evaluate efficiency improvements, they must also account for temperature variations in each room, external temperature influences, occupancy changes, and other factors Moreover, internal temperature alone does not determine thermal comfort, as it is influenced by activity levels, air velocity, relative humidity, and the mean radiant temperature of surrounding surfaces Ignoring these variables may result in inaccurate assessments of the direct rebound effect.
Data on individual household energy consumption is often scarce, typically available only through sub-metering during evaluations While total household energy consumption figures may be fairly accurate, estimating the specific energy use for services like lighting, water heating, and cooking requires techniques such as conditional demand analysis (Parti and Parti, 1980) Additionally, even in well-researched areas like personal automotive transport, uncertainties persist regarding the share of total petrol and diesel consumption linked to passenger cars and the distances traveled by various vehicle types (Schipper et al., 1993).
To estimate system-wide energy efficiencies, data on both energy consumption and useful work is essential Alternatively, one can infer the demand for useful work using energy consumption data alongside estimated energy efficiency However, obtaining accurate system-wide energy efficiency can be challenging, as it is often influenced by various factors beyond just the thermodynamic efficiency of conversion devices and may be subject to inaccuracies depending on the defined system boundaries.