Decomposition and measurement of the rebound effect: The case of energy efficiency improvements in Spain

1 DECOMPOSITION AND MEASUREMENT OF THE REBOUND EFFECT: THE CASE OF ENERGY EFFICIENCY IMPROVEMENTS IN SPAIN José M. Cansino1,2,* (ORCID: 0000-0003-1087-5399), Manuel Ordóñez1 (ORCID: 0000- 0002-1644-9131), Manuela Prieto1 (ORCID: 0000-0001-6037-4168) 1 Universidad de Sevilla Avda. Ramón y Cajal, 1 Postal Code 41018 Sevilla Spain 2 Universidad Autónoma de Chile, Av. Pedro de Valdivia 425, Providencia Santiago 758-0150, Chile * Universidad de Sevilla Departamento de Análisis Económico y Economía Política Avda. Ramón y Cajal, 1 Postal Code 41018 Sevilla Spain E-mail: [email protected] Phone: + (34) 954 55 75 28. Fax: + (34) 954 55 76 29 2 ABSTRACT The rebound effect caused by energy efficiency improvements demands greater attention from energy policymakers as it represents an important obstacle to energy consumption reduction measures. Despite the emerging literature about the rebound effect, no studies to date have managed to break down it. To do so, this study introduces the disruptive innovation of replacing the Logarithmic Mean Divisia Index method used in previous literature with Structural Decomposition Analysis. Based on Input-Output tables, a respending model allows us to assess the Indirect Rebound Effect for 14 productive sectors. Finally, Structural Decomposition Analysis enables us to determine the Economy-Wide Rebound Effect caused by energy efficiency improvements. As far as we are aware, no other study to date has done so. Spain is used as a case study for the 2000-2014 period. Major findings indicate that total rebound effect varies from nearly to 10% to around 50%. The part of rebound explained by Direct rebound effect is around 10%. Indirect rebound effect shows values between 1.2% and 1.8%. Finally, Economic wide rebound effect varies from -4.51% to around 40 % but without showing a clear trend for the period considered. The novelty of this study is that it attempts to break the overall rebound effect down to the direct, indirect, and economic wide rebound effect by combining the Cobb- Douglas production function and decomposition techniques. Keywords: Rebound effect, Input-Output, energy consumption, energy efficiency, Spain. 3 DECOMPOSITION AND MEASUREMENT OF THE REBOUND EFFECT: THE CASE OF ENERGY EFFICIENCY IMPROVEMENTS IN SPAIN 1. Introduction Energy efficiency has become a primary energy policy goal in many countries; this is certainly true for Spain, where major efforts in this regard have shaped the policies towards energy-intensive sectors (Llorca and Jamasb, 2017; Cansino, 2020). When considering a ‘stated policies’ scenario 1 , global carbon neutrality targets depend on 37% from energy efficiency improvements by 2050 (IRENA, 2019). In almost two decades, both the European Union (EU) and Spain achieved energy efficiency improvements such as a decrease in units of oil equivalent required per euro of GDP. Energy efficiency improvements could lead to changes in the demand for energy services if the rebound effect offsets some of the energy savings achieved. Should this change take place, the potential abatement of greenhouse gas (GHG) emissions and energy savings due to energy efficiency improvements may not fully materialize (Amjadi, Lundgren and Persson, 2018; Wei, Zhou and Zhang, 2019). Forecasts of energy savings could be overstated due to rebound effect. The rebound effect is defined as the additional energy consumption from overall changes in demand due to behavioral and other systemic responses to the energy efficiency improvements (Binswanger, 2001; Brookes, 1990; Khazzoom, 1980; Saunders, 2000). The rebound effect measures the fraction of energy efficiency improvements offset by an increase in energy consumption (Zhang and Lin Lawell, 2017). It corresponds to an unexpected increase in energy consumption due to the reduction of the effective cost providing a specific energy service. Energy efficiency seeks to reduce this cost (Freire- González, 2017 a). Greening, Greene and Difiglio (2000), Sorrell (2007) and Sorrell and Dimitropoulos (2008) identified three types of rebound effect which lead to greater energy consumption, through three channels: The direct rebound effect has an impact on the energy consumption of the energy service affected by the energy efficiency improvements. The cost of energy services decreases as energy efficiency improves. The downward price triggers a greater energy demand, which 1 Following International Energy Agency criteria, the Stated Policies Scenario reflects the impact of existing policy frameworks and today’s announced policy intentions. 4 means more energy consumption. The impact of the effective energy cost reduction on consumption includes both the substitution effect and the income effect. The direct rebound effect stems from the responses of first-order consumers and producers to energy efficiency improvements. Consequently, the direct rebound effect ignores any changes in the demand for other goods and services due to either the change in relative prices or higher disposable income. For this reason, the direct rebound effect is defined by energy price elasticity (cost). The indirect rebound effect affects the demand for goods stemming from changes in disposable income. Declining effective energy costs generate greater energy demand from downstream industries along the industrial chain. This second-order change is referred to as the indirect rebound effect and results in an increase in the overall energy demand (Gillingham, Rapson and Wagner, 2016). Delving further into the matter, these authors note how most of the available literature estimates the indirect rebound effect by answering the question: if consumers (or productive sectors) are given an extra dollar, how will they spend it? This implies that the evaluation of the size of the direct rebound effect focuses on the price effect while the size of the indirect rebound effect is assessed through the income effect, using a spending model as if it were an input-output model. Economy-wide effects go further than changes in energy service prices and disposable income. They stem from global economic readjustments following energy efficiency improvements, thus becoming the third-order response. The logic behind this concept is that an effective decrease in energy costs results in a reduction in the price of intermediate and final products, which leads to price adjustments throughout the whole economic system. The production-cost gap between energy-intensive industry and other industries thus narrows. A decrease in the relative price of energy-intensive products further increases consumption, thus leading to more energy consumption. Therefore, energy efficiency improvements can increase the energy demand of the entire economic system (Lin, Chen and Zhang, 2017; Yan, Ouyang and Du, 2019; Safarzadeh, Rasti-Barzoki and Hejazi, 2020). Previous studies have reviewed approaches to measuring the rebound effect but do not decompose it into its components. Currently, the literature has yet to find a way to break down the rebound effect into the direct rebound effect, indirect rebound effect and economic wide rebound effect. Various rebound effect policymakers and environmental agencies have echoed concerns raised by academics regarding the need to address the 5 rebound effect to achieve full energy consumption, economic and environmental decoupling. However, due to a lack of consensus on how to measure it, these concerns have generally not been translated into a tangible policy action (Vivanco, Kemp and van der Voet, 2016), thus acting as a barrier to evaluating the welfare implications of energy efficiency policies. As a consequence, despite the efforts made this key point remains unsolved. The overarching aim of this study is therefore to go further and bridge the gap in terms of how to break down and measure the rebound effect in order to contribute to the design of better energy efficiency actions. As far as we are aware, no study to date has attempted to do so. The novelty of the work is twofold. First, we propose a new methodological approach to break down the rebound effect into the direct rebound effect, indirect rebound effect and economic wide rebound effect by combining the Cobb-Douglas production function and decomposition techniques. Our work also is the first to do it. Second, this paper applies the proposed method to Spain for the period 2000-2014. Spain is a representative case study of strong energy efficiency improvements-oriented policy measures. The period under study is determined by the availability of data. Recently, Lin, Chen and Zhang (2017), Wang, Jiang and Li (2017) and Liu, Du and Li (2019) used the Logarithmic Mean Divisia Index (LMDI I) method and the Cobb-Douglas production function to assess direct rebound effect in Chinese industry. Under the indirect rebound effect definition provided by Greening, Greene and Difiglio (2000), the effective decrease in utility energy costs could lower the price of energyconsuming products, leading to a rise in the demand for these products in the economic system, thus increasing the energy demand. This increase has an impact on macroeconomic activity through the inter-sectoral relationship described in Input-Output Tables (IOTs). For that reason, this research takes the methodological advance from Lin, Chen and Zhang (2017), Wang, Jiang and Li (2017) and Liu, Du and Li (2019) as a starting point but contributes the disruptive innovation of replacing LMDI I with Structural Decomposition Analysis. The use of IOTs and a re-spending model allows this study to assess the indirect rebound effect as described below. Finally, Structural Decomposition Analysis enables us to capture the economy-wide rebound effect caused by the energy efficiency improvements. To calculate the economy-wide rebound effect, changes in inter-sectoral relationships must be accounted for, which lends support to the use of Structural Decomposition Analysis due to changes in IOTs from year to year. This 6 is the main methodological contribution to the literature, but not the only one. The ultimate goal and the novelty of this study is that it attempts to break the overall rebound effect down to the direct, indirect, and economy-wide rebound effect by combining the Cobb-Douglas production function and decomposition techniques. This paper is structured as follows: after the Introduction, a review of the literature is shown presented in Section 2. Section 3 details the methodology and materials used. The main results are detailed and discussed in Section 4, and the conclusions drawn are set out in Section 5. 2. Review of the Literature Considerable concerns have emerged in the field of energy economics due to the rebound effect primarily caused by the energy-saving effects of energy efficiency improvements. Most of the related articles focus on high emission areas or countries (i.e., EU countries, US and China) and the most energy-intense sectors (industry, transport or residential). A comprehensive summary of the findings in the literature on the rebound effect can be found in Sorrell (2007) and Jenkins, Nordhaus and Shellenberger (2011). An updated review is provided by Lin, Chen and Zhang (2017), while the most recent review of the literature comes from Jin and Kim (2019) and Safarzadeh, Rasti-Barzoki and Hejazi (2020), with the latter reference referring to the industry sector only. A wide range of methodological approaches have been used to assess the rebound effect. To capture the economy-wide rebound effect, the computable general equilibrium (CGE) model has been widely used. Turner (2009) used a CGE framework to research the conditions under which the rebound effect may occur in response to increases in energy efficiency in the UK national economy. Hanley, McGregor, Swales and Turner, (2009) applied a similar approach for the Scottish economy. Turner and Hanley (2011) used a CGE model for the same economy to analyze the factors influencing impacts of one form of technological change—improvements in energy efficiency—on absolute levels of CO2 emissions, on the carbon intensity of the economy, and the per capita Environmental Kuznets Curve relationship. An article by Yu, Moreno-Cruz and Crittenden (2015) focusing on the state of Georgia, USA, is also worth citing. More recently, Bye, Fæhn and Rosnes (2018) applied a multi-sector CGE model to the Norwegian economy to explore the cost, emission, and energy rebound effects of alternative policy interpretations underlying the proposed 2030 energy efficiency goal. 7 As econometrics approach has also been widely used. Ouyang, Gao, Du and Du (2018) first estimated an energy demand model to assess the magnitude of the rebound effect in Yangtze River Delta Urban Agglomeration industrial sectors by using dynamic Ordinary Least Squares regression and seemingly unrelated regression methods. Llorca and Jamasb (2017) analyzed energy efficiency and the key features influencing the rebound effect for road freight transport in 15 European countries during the 1992–2012 period using a stochastic frontier analysis approach. Freire-González (2017b) developed a hybrid methodology combining econometric estimates, extended environmental input-output analysis, and re-spending models. The econometric total factor productivity (TFP) approach is usually the best way to represent the contribution of technological progress, with the Cobb-Douglas and CES production function being employed to estimate TFP. Yan, Ouyang and Du (2019) used dynamic panel data models to estimate the economywide rebound effect for China’s provinces. Orea, Llorca and Fili (2015) incorporated the measurement of the rebound effect into a stochastic energy demand frontier model and estimated the effect as indicated by Saunders (2000). These authors modified the conventional stochastic energy demand frontier model by adding an interaction term with the inefficiency term, which allowed them to estimate energy efficiency and the rebound effect simultaneously in a one-step procedure. Adetutu, Glass and Weyman-Jones (2016) adopted a two-step strategy for measuring the energy rebound effect. In a first step, they used stochastic frontier analysis to measure the energy efficiency values. In a second step, they used a dynamic energy demand regression model in which a set of variables interacted with the term measuring energy efficiency. Hediger, Farsi and Weber (2018) linked econometrics with behavioral economics by researching how households respond to heating system efficiency improvements. In a similar vein, Santarius and Soland (2018) addressed the current rebound discourse in psychological theories. Along with the aforementioned approaches, decomposition techniques for measuring the rebound effect are also available in the specialized literature. For the most part, the decomposition analysis involves two methods: Index Decomposition Analysis (IDA) and Structural Decomposition Analysis. LMDI I is an IDA technique (Cansino, Sanchez- Braza and Rodriguez-Arevalo, 2015); in fact, it is the most extensively applied model in the field of energy economics. Together with top-down models, decomposition models 8 are among the most commonly used for the estimation of the rebound effect (Safarzadeh, Rasti-Barzoki and Hejazi, 2020). However, Structural Decomposition Analysis uses data from IOTs and offers a broader range of information for both technical aspects and the effects of the final demand than IDA does (De Han, 2001; Cansino, Román-Collado and Ordóñez, 2016). Typical Structural Decomposition Analysis studies can provide more detailed structural factors, such as the Leontief effect (or technical effect) (Xie, 2014) and also shape socio-economic drivers from both production and final demand perspectives. LMDI approaches differ from Structural Decomposition Analysis as they do not allow researchers to carry out an in-depth analysis of the internal production linkages within an economy, and their influence on the changes in GHG emissions or energy consumption levels (Brizga, Feng and Hubacek, 2014). IOTs help capture the indirect rebound effect and economy-wide rebound effect while LMDI does not. This is because IOTs involve macroeconomic inter-industry reactions to changes, meaning that the proposed methodology is capable of accounting for the indirect channels (indirect rebound effect and economy-wide rebound effect) through which the rebound effect operates. This represents one advantage when comparing with the literature on LMDI. 3. Method and Materials 3.1. Method. 3.1.1. The energy rebound effect The total rebound effect generated by technological progress in an economy can be defined as the ratio of the incremental energy consumption to the amount of energy savings caused by technological progress, as shown in Eq (1). 𝑅𝐸𝑡+1 =Incremental energy consumption Amount of energy savings caused by technological progress =𝐴𝑡+1(𝑌𝑡+1 − 𝑌𝑡)𝐼𝑡+1 𝑌𝑡+1(𝐼𝑡− 𝐼𝑡+1) (1) Where t denotes time; It is energy intensity in period t; Yt denotes an output measure in period t (i.e., total gross added value); At denotes total productivity factor in period t; and, finally, REt represents the total energy rebound effect in period t expressed as the percentage of the forecast reduction in energy use lost due to the sum of the consumers’, productive sectors’ and market’s responses to changes after energy efficiency improvements. It includes the direct rebound effect, indirect rebound effect and economywide rebound effect. 9 3.1.2. Direct rebound effect To determine the direct rebound effect, the contribution of energy consumption to economic growth is assessed. It is assumed that energy efficiency improvements derive from technological progress, thus it is first necessary to calculate its contribution to economic growth—σt in Eq (15) below. A neoclassical production function is considered in Eq (2) as a starting point (Griffin, 1981; Howarth, 1997 and Du, Qiang, Li and Bai, 2017), 𝑌𝑡= 𝐴𝑡 .𝐹(𝐾𝑡,𝐿𝑡,𝐸𝑡) (2) where At represents technological progress in the sense of Solow (1957) varying through time, Kt measures capital stock, Lt measures the labor force and Et, energy consumption. The Cobb Douglas production function is used to estimate parameters as follows (Douglas, 1976): 𝑌𝑡= 𝐴0 .𝑒𝑟𝑡. 𝐾𝑡 𝛼. 𝐿𝑡 𝛽 .𝐸𝑡 𝛾 (3) In Eq (3) technological progress has a fixed value (A0) and an unfixed one (ert) (Pack, 1994), α, β and γ parameters are the elasticities of economic growth to capital (K), labor (L) and energy consumption (E). Following Freire-González (2017a), under certain assumptions 2 , elasticities of economic growth for energy consumption (E) may be considered as energy price elasticity, thus defining the direct rebound effect (see Llorca and Jamasb, 2017; Gillingham, Kotchen, Rapson and Wagner, 2013; Gillingham, Rapson and Wagner, 2016; Frondel, Ritter. and Vance, 2012). To soften out layers, natural logarithms are used: ln𝑌𝑡= ln𝐴0+𝑟𝑡 + 𝛼 ln𝐾𝑡+ 𝛽 ln𝐿𝑡+ 𝛾 ln𝐸𝑡 (4) Parameters in Eq (4) could be estimated with Ordinary Least Squares regression after checking the Engel-Granger test. However, when Variance Inflation Factor (VIF) values indicate multicollinearity problems, other types of regression techniques such as Ridge regression must be used (Hoerl and Kennard, 1970). 2 They are: (1) consumers’ reaction to an efficiency improvement is similar to their reaction to a reduction in energy prices, (2) energy efficiency is not affected by energy prices. 16 Labor data (Lt) are from the Spanish Statistical Office (INE) measured as the number of employees. Energy consumption data measured in thousand tons of oil equivalents (ktoe) were taken from IDAE (2020) and from Eurostat (2020 a). For the IO analysis, IOTs at 2014 basic prices (INE, 2019) were used, which included 64 activity sectors. The last available year was 2014; thus, the full period under analysis corresponds to 2000-2014. To standardize the sectors included in the IOTs with those considered in the energy consumption data, they were grouped into 14 sectors as shown in Table A.1 in the supplementary materials. To assess δ values from Eq (11), IOTs were taken from the WIOD database (Timmer, Dietzenbacher, Los, Stehrerand and de Vries, 2015; World Input-Output Database; 2016). Table A.2 details the criteria for grouped activity branches. Finally, the savings rate considered in the re-spending model was taken from Eurostat (2020b). Table A.3 in supplementary materials details the savings rates for the whole period. 4. Results 4.1. Direct rebound effect We initially planned to estimate Eq (5) by Ordinary Least Squares regression, since the assumptions of normality and homoscedasticity of regression residuals are fulfilled and the Engle-Granger test confirms that the variables are co-integrated (Engle and Granger, 1987). The results of this latter test are essential when working with time series in order to be able to apply the Ordinary Least Squares regression approach. If the Engle-Granger test indicates that the variables are co-integrated, as it does in this case, the estimation by means of Ordinary Least Squares regression captures stable, long-term relationships; in other words, it avoids spurious relationships. However, Eq (5) has been estimated using a Ridge regression due to the multicollinearity problems presented by predictor variables, as confirmed by the VIF values. 3 The Ridge regression minimizes the following expression: ∑(𝑌𝑖− 𝑌 𝑖)2=∑(𝑌𝑖−∑𝛽𝑗𝑋𝑖𝑗 𝑝 𝑗=0 )2+ 𝜆 ∑𝛽𝑗 2 𝑝 𝑗=0 𝑀 𝑖=1 𝑀 𝑖=1 (21) 3 The test results conducted to estimate the Cobb Douglas production function are available as supplemental material. 17 where i corresponds to the number of years analyzed and j is the number of regressors (five in our case); λ is calculated using cross validation methods. The λ value indicates the weight given in the estimation to minimize the quadratic error and the penalty term, formed by the sum of the square of the coefficients. In this case, the Ridge regression yields better estimators than Ordinary Least Squares regression (Hoerl and Kennard, 1970). Eq (22) shows the parameter values. 𝑌𝑡∗=(−7.3201 + 0.0043 × 𝑡)+ 0.1618𝐾𝑡 ∗+ 0.5309𝐿𝑡 ∗+ 0.1104𝐸𝑡 ∗ (22) An R2 of 0.994 has been estimated; coefficients are all significant at 1% with an optimal λ of 0.0585. 4 From Eq (22), the size of direct rebound effect is calculated at 11.04 % for the period considered. This means that 11.04% of available resources derived from energy consumption savings triggers additional energy consumption due to the reduction in its implicit price. The result for the direct rebound effect is in line with the findings reported by Allcott (2011) for Illinois (10 %), Barla, Lamonde, Miranda-Moreno and Boucher (2009) for Canada (8%), Ito (2014) for California (8.8%), Jessoe and Rapson (2014) for Connecticut (12%) and Kulmer and Seebauer (2019) for Austria (8% - 10%). These papers focused on developed countries as we do in this paper. When assessing the rebound effect, Eq (5) also provides estimates for each of the 14 productive sectors studied. To do so, sectoral values of σ were first estimated. However, the lack of data disaggregated by sector gave rise to results that were contrary to those indicated by the theoretical models, due to the stationarity of the energy consumption series in certain sectors and recession years within the period studied. For example, for the Agricultural, forestry and fishing sector, the estimation indicated that the increase in any of the production factors had a negative effect on economic growth. In other sectors, (transport and storage), none of the coefficients obtained were statistically significant. 4.2. Indirect rebound effect The direct rebound effect determines the amount of disposable income or budget to be respent. In an IO model, this determines changes in the demand vector shock for economic activity. This shock causes the indirect rebound effect, which is captured through the 4 The statistical significance of each estimator has been calculated following Salmerón and Rodríguez (2017). The values obtained are shown in the supplemental material. 18 extended IO energy model as described in Eq (8). A new demand vector is calculated as follows: The IOT of Spain’s domestic production up to 2014 was taken as the starting point. The final demand of institutional sectors (households and nonprofit entities, companies, public sector and foreign sector) was extracted from the IOT. From these results, and in relation to the size of the direct rebound effect, we determine the available resources for re-spending on goods and services (both domestic and imported) of each productive sector included in the IOT. Since direct rebound effect was 11.04%, effective savings in energy consumption amounts to 88.96%. In general, this implies that a savings of 1% from energy efficiency improvements provokes an increase of 0.8896% in available resources to be re-spent on other goods and services. Further goods and services require new inputs for energy consumption. New energy requirements due to re-spending will determine the size of the indirect rebound effect. To assess the size of the indirect rebound effect, two starting points are needed. The first is the value of final energy consumption for all productive sectors included in Spain’s 2014 IOT. This was 72,023.8 ktoe. The second starting point is the initial spending of institutional sectors for their final demand, thus implying energy consumption. Table 1 shows figures for this second starting point. As an example, take the household institutional sector (column 2 in Table 1). In this case, 1% additional savings would amount to 293.57 M€ when taking its initial spending as the starting point. Due to the direct rebound effect, 11.04% of these potential savings goes to additional spending, so new available resources for the household institutional sector due to energy efficiency improvements amounts to 261.2 M€ (88.96% of potential savings). For households, one needs to bear in mind that spending is not the only choice; savings is another. A savings rate of 6.34% (Eurostat, 2020b) was assumed, so effective and additional spending is equal to 244.6 M€. Households + nonprofit entities Capital investment by companies Public consumption Exports Initial spending on energy consumption 29,356.8 1,594.6 5,686.1 3,509.2 19 Energy consumption savings 261.2 14.2 50.6 31.2 Additional spending in productive sectors other than energy sector 244.6 14.2 50.6 31.2 Table 1: Re-spend from energy consumption savings due to energy efficiency improvements (M €). In order to assess the demand vector shock on the extended IO energy model, we need to know 1) how additional spending goes to every productive sector and 2) how to break it down between domestic spending and imports. After answering 1) and 2) it is also necessary to distribute additional spending in Table 1 into the various productive sectors. Table 2 shows how to do it by taking households and nonprofit entities as an example regarding the case of the Agriculture, forestry and fishing sector. Households and nonprofit entities spend 1.7% of their additional budget, which means 4.1 M€ of their total new available resources (244.6 M€). Only 62.1% of additional spending goes to domestic goods and services; the rest of the new budget goes to imports. An amount of 2.6 M€ corresponds to the re-spending of households and nonprofit entities on goods and services from the Agriculture, forestry and fishing sector. New goods and services will require additional energy consumption, thus determining the indirect rebound effect resulting from energy efficiency improvements, but only for households and nonprofit and the Agriculture, forestry and fishing sector, which are used here as an example. For all other productive sectors included in the IOT, we proceeded in the same way. To simplify, it was assumed that company investment spending goes only to capital instead of capital and stock. Savings decisions were only contemplated for households, with this being a limitation for our research. Further research might consider savings scenarios for the public sector. (1) (2) (3) (4) Agriculture, forestry and fishing 1.7 4.1 62.1 2.6 Mining and quarrying 0.0 0.1 43.7 0.0 20 Food, beverages and tobacco 8.1 19.8 72.4 14.3 Textile and leather 2.3 5.7 22.6 1.3 Paper, pulp and printing 0.4 0.9 78.1 0.7 Chemical and petrochemical; non-metallic minerals 4.1 9.9 64.1 6.4 Basic metals and fabricated metal products 0.3 0.8 67.3 0.5 Machinery 0.9 2.2 31.2 0.7 Transport equipment 2.2 5.3 24.8 1.3 Other industries 0.9 2.2 35.6 0.8 Construction 1.0 2.4 99.1 2.4 Energy sector 3.8 9.3 99.6 9.3 Transport and storage 2.7 6.7 93.4 6.2 Commercial, services and public services 71.6 175.2 97.6 171.0 TOTAL 100.0 244.6 88.9 217.4 (1) Sectoral consumption as a share of total consumption (%). (2) Increase in sectoral consumption due to energy efficiency improvements (M €). (3) Sectoral domestic consumption as a share of total consumption (%). (4) Increase in domestic consumption due to energy efficiency improvements (M €). Table 2: Households and nonprofit entities re-spending broken down by productive sectors (M€). Table 3 shows the increase in demand caused by the re-spending of all institutional sectors. The last column exhibits the value of each element for the final demand. This vector was taken as the shock to the extended IO energy model to assess the size of the indirect rebound effect. (1) (2) (3) (4) TOTAL Agriculture, forestry and fishing 2.6 0.1 0.0 1.2 3.9 Mining and quarrying 0.0 0.1 0.0 0.5 0.6 Food, beverages and tobacco 14.3 0.1 0.0 2.5 16.9 Textile and leather 1.3 0.0 0.0 1.4 2.7 Paper, pulp and printing 0.7 0.0 0.0 0.6 1.4 Chemical and petrochemical; non-metallic minerals 6.4 0.2 0.3 6.6 13.4 Basic metals and fabricated metal products 0.5 0.3 0.0 2.3 3.2 Machinery 0.7 0.5 0.0 2.4 3.6 21 Transport equipment 1.3 0.6 0.1 4.8 6.8 Other industries 0.8 0.3 0.0 0.5 1.6 Construction 2.4 5.9 0.2 0.2 8.8 Energy sector 9.3 0.1 1.1 0.3 10.7 Transport and storage 6.2 0.1 1.4 1.8 9.4 Commercial, services and public services 171.0 3.8 45.8 6.2 226.7 TOTAL 217.4 12.0 49.0 31.2 309.7 (1) Households and nonprofit institutions consumption. Household consumption and Non-profit institutions serving households (2) Investment spending in fixed capital. (3) Public spending. (4) Exports. Table 3: Institutional sectors re-spending broken down by productive sectors (M€). Indirect rebound effect values up to 2014 are shown in Table 4. As a whole, the indirect rebound effect implies an additional energy consumption of 13.2 ktoe. As expected, the energy sector shows a high value for increasing energy consumption, but the same is true for the Transport sector (6 ktoes)–almost half of the total increase—and the Commercial, services and public services (2 ktoes). Others sector with high values but in relative terms are Food, beverages and tobacco and Agriculture, forestry and fishing. Lower impacts correspond to Construction, Textile and leather, Mining and quarrying and Transport equipment. Indirect rebound effect values in Table 4 expressed in percentages are derived from the energy consumption figure. As an initial spending value of 1% (for households and also for the rest of institutional sectors) was assumed, it corresponds to 720.2 ktep of energy consumption (a 1% share of total energy consumption in 2014; 72,023.8 ktep). Taking 720.2 ktep as the reference value, an increase in energy consumption of 13.2 ktep is due to energy efficiency improvements shares of 1.83%. This determines a total indirect rebound effect size of 1.83%. Activity Branches ktep (% on total) Indirect 22 rebound effect Agriculture, forestry and fishing 0.5 4.0 1.9 Mining and quarrying 0.0 0.3 0.8 Food, beverages and tobacco 0.5 3.7 2.1 Textile and leather 0.0 0.3 1.3 Paper, pulp and printing 0.4 3.0 1.7 Chemical and petrochemical; non-metallic minerals 1.0 7.6 1.4 Basic metals and fabricated metal products 0.4 3.1 1.1 Machinery 0.1 0.7 1.1 Transport equipment 0.0 0.3 1.0 Other industries 0.2 1.2 1.4 Construction 0.1 0.9 0.9 Energy sector 1.8 13.7 2.1 Transport and storage 6.0 45.6 1.9 Commercial, services and public services 2.0 15.5 2.3 TOTAL 13.2 100 1.83 Table 4: Sectoral energy consumption after re-spending decisions (ktoe and %) and 2014 indirect rebound effect values. Figure 1 shows the indirect rebound effect trend for the entire period, in total values, and Figure 2 shows it by sector. Table A.4 in the supplementary materials details all results in terms of ktoe. For the entire Spanish economy, the indirect rebound effect varies from 1.2% to 1.8%. An increase in this value appears from 2008 before stabilizing in 2012. This coincides with an extended period of energy-intensive behavior during recession years. Energy intensity may rise during periods of recession due to the decoupling of energy consumption and economic activity. Decreased production levels could provoke an increase in energy consumption per unit produced and a reduction in the use of productive capacities. The limited use of installed capacities and maintaining fixed 23 consumption levels explains the increased consumption per production unit (Cansino, Sanchez-Braza and Rodriguez-Arevalo, 2015). As expected, the energy sector shows high values for the indirect rebound effect. The same happens with the Commercial, services and public services and the Food, beverages and tobacco sectors. These three productive sectors show values from 1.3% up to 2.1%. As with the total indirect rebound effect size for Spain, an increase is detected from 2008 onwards. It should be borne in mind that the Kyoto Protocol came into force in 2005. Thus, it could be the case that the first results in terms of energy efficiency improvements began to appear after 2005, meaning that part of the available resources that sectors started to re-spend have higher values for the indirect rebound effect. The major recession of 2008 forced a strong indirect rebound effect to decrease for the most exposed productive sectors; namely, Construction, Transport equipment, Machinery and Other industries. As discussed in the introduction section, the size of the rebound effect, regardless of its components, remains a matter of debate. Thus, it is informative to compare our results with other studies. Freire-González (2017a) assessed two measures for the indirect rebound effect caused by the behavior of Spanish households after energy efficiency improvements: both differ notably from our findings. When considering a direct rebound effect of 30%, the indirect rebound effect reported by this author was 34.60%, and 24.71% when taking a direct rebound effect size of 50%. He used 30% and 50% values for the direct rebound effect, as these were considered among the range of estimates found in the literature on the direct rebound effect in households for industrialized countries (Greening, Greene and Difiglio, 2000; Chakravarty, Dasgupta and Roy, 2013; Labandeira, Labeaga and López-Otero, 2017). Values were determined by using price elasticities but not through a production function, as standard practice when analyzing productive sectors. Sorrell (2007) and Freire-González (2017a) acknowledged that the price elasticities approach tends to overestimate the rebound effect. Closer to our results, Thomas and Azevedo (2013) found an indirect rebound effect of 10% for US households. These findings are of particular of interest because these authors conducted their research following an IO approach, as we do, under the assumption of a 10% direct rebound effect (near 11.04% as we found) from an energy efficiency intervention which reduces household expenditures in either electricity, natural gas, or gasoline, and the US economic structure, energy prices, and electric grid mix of 2002. Recently, Fullerton and Ta (2020) found an indirect rebound effect of 4.26% from a 24 costless technology shock when no policy measures are implemented to limit it. This result is quite similar to ours, although the cited authors did not deal with the economywide rebound effect assessment. The literature pays attention to most of the productive sectors for which higher indirect rebound effect values were found; for example, the Transport sector, where the increase in ecommerce and logistic activities acts as a catalyst (Ordóñez, Arcos, Cansino and Román-Collado, 2019). Llorca and Jamasb (2017) found that the rebound effect for refrigerator transport in the EU-15 was 4% for 1992-2012, but without breaking down the rebound effect. The results from Craglia and Cullen (2020) for the UK were similar (4.6 %). Other industrial sectors have also received a good deal of scholarly attention (Smeets, Tabeau, van Berkum, Moorad, van Meijl, and Woltjer, 2014; Barker, Ekins and Foxon, 2007; Amjadi, Lundgren and Persson, 2018). An innovative analysis of the rebound effect in Agriculture is provided by Paul, Techen, Robinson and Helming (2019) for the case of Germany. 25 Figure 1: Indirect rebound effect (IRE) total value for the Spanish economy % (2000- 2014). 32 Seville. The first author acknowledges funding provided by the Universidad Autónoma de Chile (Chile). The third author acknowledges funding provided the Andalusian Regional Government (project SEJ-566). 33 References Adetutu, M. O., Glass, A. J. and Weyman-Jones, T. G. (2016). Economy-wide estimates of rebound effects: Evidence from panel data. Energy Journal, 37(3), 251–269 (available at https://mpra.ub.uni-muenchen.de/65409/1/MPRA_paper_65409.pdf). Allcott, H. (2011). Rethinking real-time electricity pricing. Resource and Energy Economics, 33 (4), 820–42. DOI: 10.1016/j.reseneeco.2011.06.003 Amjadi, G., Lundgren, T. and Persson, L. (2018). The rebound effect in Swedish heavy industry. Energy Economics, 71, 140-148. DOI: 10.1016/j.eneco.2018.02.001 Antal, M., and van den Bergh, J. C. (2014). Re-spending rebound: A macro-level assessment for OECD countries and emerging economies. Energy Policy, 68, 585-590. DOI: 10.1016/j.enpol.2013.11.016 Arocena, P., Gómez, A. and Peña, S. (2016). La eficiencia energética, el efecto rebote y el crecimiento económico. Papeles de Energía, 2, 39-63 (available at https://www.funcas.es/articulos/la-eficiencia-energetica-el-efecto-rebote-y-el- crecimiento-economico/). Barla, P., Lamonde, B., Miranda-Moreno, L. F. and Boucher, N. (2009). Traveled distance, stock and fuel efficiency of private vehicles in Canada: price elasticities and rebound effect. Transportation, 36(4), 389-402. DOI:10.1007/s11116-009-9211-2 Barker, T., Ekins, P. and Foxon, T. (2007). Macroeconomic effects of efficiency policies for energy-intensive industries: the case of the UK Climate Change Agreements, 2000- 2010. Energy Economics, 29(4), 760-778. DOI: 10.1016/j.eneco.2006.12.008 Binswanger, M. (2001). Technological progress and sustainable development: what about the rebound effect? Ecological Economics, 36, 119–132. DOI: 10.1016/S0921- 8009(00)00214-7 Brizga, J., Feng, K. and Hubacek, K. (2014). Drivers of greenhouse gas emissions in the Baltic States: A structural decomposition analysis. Ecological Economics, 98, 22-28. DOI: 10.1016/j.ecolecon.2013.12.001 Brookes, L. (1990). The greenhouse effect: the fallacies in the energy efficiency solution. Energy Policy, 18, 199–201. DOI 10.1016/0301-4215(90)90145-T Bruns, S. B., Moneta, A. and Stern, D. I. (2021). Estimating the Economy-Wide Rebound Effect Using Empirically Identified Structural Vector Autoregressions. Energy Economics, 97. DOI: 10.1016/j.eneco.2021.105158 Bye, B., Fæhn, T. and Rosnes, O. (2018). Residential energy efficiency policies: Costs, emissions and rebound effects. Energy, 143, 191-201. DOI: 10.1016/j.energy.2017.10.103 Cansino J.M., Sanchez-Braza A. and Rodriguez-Arevalo M. L. (2015). Driving forces of Spain's CO2 emissions: a LMDI decomposition approach. Renewable and Sustainable Energy Reviews, 48, 749-759. DOI: 10.1016/j.rser.2015.04.011 34 Cansino, J.M., Román-Collado, R. and Ordóñez, M., (2016). Main drivers of changes in CO2 emissions in the Spanish economy: A structural decomposition analysis. Energy Policy 89, 150−159. DOI: 10.1016/j.enpol.2015.11.020 Cansino, J. M., Román-Collado, R. and Merchán, J. (2019). Do Spanish energy efficiency actions trigger JEVON’S paradox? Energy, 181, 760-770. DOI: 10.1016/j.energy.2019.05.210 Cansino, J. M. (2020). El reto de la eficiencia energética en Europa y el riesgo del efecto rebote. Araucaria. Revista Iberoamericana de Filosofía, Política y Humanidades, 22 (45), 269-289 (available in https://hdl.handle.net/11441/105356). Chakravarty, D., Dasgupta, S. and Roy, J. (2013). Rebound effect: how much to worry? Current Opinion in Environmental Sustainability, 5(2), 216-228. DOI: 10.1016/j.cosust.2013.03.001 Craglia, M. and Cullen, J. (2020). Do vehicle efficiency improvements lead to energy savings? The rebound effect in Great Britain. Energy Economics, 88. DOI: 10.1016/j.eneco.2020.104775 De Han, M. (2001). A Structural Decomposition analysis of pollution in the Netherlands, Economic Systems Research, 13(2): 181-196. DOI: 10.1080/09537320120052452 Dietzenbacher, E. and Los, B. (1998), Structural Decomposition Techniques: Sense and Sensitivity, Economic Systems Research, 10(4):307-323. DOI: 10.1080/09535319800000023 Douglas, P.H. (1976). The Cobb-Douglas Production Function Once Again: Its History, Its Testing, and Some New Empirical Values. Journal of Political Economy, 84(5), 903- 915 (available at http://ecocritique.free.fr/douglas1976.pdf). Du, Qiang, Li, Y. and Bai, L. (2017). The Energy Rebound Effect for the Construction Industry: Empirical Evidence from China. Sustainability. 9(5). DOI: 10.3390/su9050803 Duarte, R., Sánchez-Chóliz, J. and Sarasa, C. (2018). Consumer-side actions in a lowcarbon economy: A dynamic CGE analysis for Spain. Energy Policy, 118, 199-210. DOI: 10.1016/j.enpol.2018.03.065 Engle, R. F., and Granger, C. W. (1987). Co-integration and error correction: representation, estimation, and testing. Econometrica: journal of the Econometric Society, 251-276. DOI: 10.2307/1913236 Eurostat (2020 a), Energy balances. Complete energy balances, 2015 (available at www.ec.europa.eu/eurostat). Eurostat (2020 b), Annual sector accounts. Household saving rate (available at www.ec.europa.eu/eurostat). Feng, C., Wang, M., Zhang, Y. and Liu, G. C. (2018). Decomposition of energy efficiency and energy-saving potential in China: A three-hierarchy meta-frontier approach. Journal of Cleaner Production, 176, 1054-1064. DOI: 10.1016/j.jclepro.2017.11.231 Freire-González, J. (2017 a). Evidence of direct and indirect rebound effect in households in EU-27 countries. Energy Policy, 102, 270-276. DOI: 10.1016/j.enpol.2016.12.002 35 Freire-González, J. (2017 b). A new way to estimate the direct and indirect rebound effect and other rebound indicators. Energy, 128, 394-402. DOI: 10.1016/j.energy.2017.04.057 Frondel, M., Ritter, N. and Vance, C. (2012). Heterogeneity in the rebound effect: Further evidence for Germany. Energy Economics, 34(2), 461-467. DOI: 10.1016/j.eneco.2011.10.016 Fullerton, D. and Ta, C. L. (2020). Costs of energy efficiency mandates can reverse the sign of rebound. Journal of Public Economics, 188. DOI: 10.1016/j.jpubeco.2020.104225 Fundación BBVA (2019). El stock y los servicios de capital y su distribución territorial y sectorial. Resultados detallados, magnitudes en términos reales (available at www.fbbva.es). Gillingham, K., Kotchen, M. J., Rapson, D. S., and Wagner, G. (2013). The rebound effect is overplayed. Nature, 493(7433), 475-476. DOI: 10.1038/493475a Gillingham, K., Rapson, D. and Wagner, G. (2016). The rebound effect and energy efficiency policy. Review of Environmental Economics and Policy, 10(1), 68-88 (available in https://www.jstor.org/stable/resrep01115). Greening, L. A., Greene, D. L. and Difiglio, C. (2000). Energy efficiency and consumption—the rebound effect—a survey. Energy policy, 28(6-7), 389-401. DOI: 10.1016/S0301-4215(00)00021-5 Griffin, J. M. (1981). Engineering and Econometric Interpretations of Energy-Capital Complementarity: Comment. The American Economic Review, 71(5): 1100-1104 (available at https://www.jstor.org/stable/1803499). Guerra, A. I. and Sancho, F. (2011). An analysis of primary energy requirements and emission levels using the structural decomposition approach: The Spanish case. In María Llop (ed) Air Pollution Economic Modelling and Control Policies (pp. 112-131). DOI: 10.2174/978160805217211101010112 Hanley, N., McGregor, P. G., Swales, J. K. and Turner, K. (2009). Do increases in energy efficiency improve environmental quality and sustainability? Ecological Economics, 68(3), 692-709. DOI: 10.1016/j.ecolecon.2008.06.004 Hediger, C., Farsi, M., and Weber, S. (2018). Turn it up and open the window: On the rebound effects in residential heating. Ecological Economics, 149, 21-39. DOI: 10.1016/j.ecolecon.2008.06.004 Hoerl, A. E. and Kennard, R. W. (1970). Ridge Regression: Biased Estimation for Nonorthogonal Problems. Technometrics, 12:1, 55-67. (available at https://www.math.arizona.edu/~hzhang/math574m/Read/RidgeRegressionBiasedEstima tionForNonorthogonalProblems.pdf). Howarth, R.B. (1997). Energy Efficiency and Economic Growth. Contemp. Economic Policy, 15(4), 1-9. DOI: 10.1016/j.energy.2006.06.005 IDAE (2020), Balances energéticos. Balance del consumo de energía final, año 2015 (available at www.idae.es). 36 INE. Instituto Nacional de Estadística (2019). Annual Spanish National Accounts: Input-Output Tables. (available at https://www.ine.es/dyngs/INEbase/en/operacion.htm?c=Estadistica_C&cid=125473617 7058&menu=resultados&idp=1254735576581). IRENA (International Renewable Energy Agency) (2019). Global energy transformation: A roadmap to 2050. 2019 edition (available at https://www.irena.org/publications/2019/Apr/Global-energy-transformation-A- roadmap-to-2050-2019Edition). Ito, K. (2014). Do consumers respond to marginal or average price? Evidence from nonlinear electricity pricing. American Economic Review, 104(2), 537-63. DOI: 10.1257/aer.104.2.537 Jenkins, J., Nordhaus, T. and Shellenberger, M., (2011). Energy Emergence: Rebound and Backfire as Emergent Phenomena. Breakthrough Institute, USA (available at https://www.ourenergypolicy.org/wp-content/uploads/2013/11/Energy_Emergence.pdf). Jessoe, K. and Rapson, D. (2014). Knowledge is (less) power: Experimental evidence from residential energy use. American Economic Review 104 (4): 1417–38. DOI: 10.1257/aer.104.4.1417 Jin, T. and Kim, J. (2019). A new approach for assessing the macroeconomic growth energy rebound effect. Applied Energy, 239, 192-200. DOI: 10.1016/j.apenergy.2019.01.220 Khazzoom, J.D. (1980). Economic implications of mandated efficiency in standards for household appliances. Energy 1(4), 21–40 (available at https://www.jstor.org/stable/41321476). Kulmer, V., and Seebauer, S. (2019). How robust are estimates of the rebound effect of energy efficiency improvements? A sensitivity analysis of consumer heterogeneity and elasticities. Energy Policy, 132, 1-14. DOI: 10.1016/j.enpol.2019.05.001 Labandeira, X., Labeaga, J. M. and López-Otero, X. (2017). A meta-analysis on the price elasticity of energy demand. Energy Policy, 102, 549-568. DOI: 10.1016/j.enpol.2017.01.002 Leontief, W. (1966). Input-Output Economics, 2ª ed., New York, Oxford University Press Li, G., Sun, J. and Wang, Z. (2019). Exploring the energy consumption rebound effect of industrial enterprises in the Beijing–Tianjin–Hebei region. Energy Efficiency, 12(4), 1007-1026. DOI: 10.1007/s12053-018-9743-4 Lin, B., Chen, Y. and Zhang, G. (2017). Technological progress and rebound effect in China's nonferrous metals industry: An empirical study. Energy Policy, 109, 520-529. DOI: 10.1016/j.enpol.2017.07.031 Liu, H., Du, K. and Li, J. (2019). An improved approach to estimate direct rebound effect by incorporating energy efficiency: A revisit of China's industrial energy demand. Energy Economics, 80, 720-730. DOI: 10.1016/j.eneco.2019.02.012 37 Llorca, M. and Jamasb, T. (2017). Energy efficiency and rebound effect in European road freight transport. Transportation Research Part A: Policy and Practice, 101, 98-110. DOI: 10.1016/j.tra.2017.05.002 Murray, A. G. and Mills, B. F. (2011). Read the label! Energy Star appliance label awareness and uptake among US consumers. Energy Economics, 33(6), 1103-1110. DOI: 10.1016/j.eneco.2011.04.013 Ordóñez, M., Arcos, Á., Cansino, J. M. and Román-Collado, R. (2019). Efectos económicos y medioambientales de la electro-movilidad en España. Un análisis input output medioambientalmente. Economía industrial, 411, 85-97 (available at https://www.mincotur.gob.es/Publicaciones/Publicacionesperiodicas/EconomiaIndustria l/RevistaEconomiaIndustrial/411/ORDO%C3%91EZ,%20ARCOS,%20CANSINO%20 Y%20ROM%C3%81N.pdf). Orea, L., Llorca, M. and Filippini, M. (2015) A new approach to measuring the rebound effect associated to energy efficiency improvements: an application to the US residential energy demand, Energy Economics, 49, 599 – 609. DOI: 10.1016/j.eneco.2015.03.016 Ouyang, X., Gao, B., Du, K. and Du, G. (2018). Industrial sectors' energy rebound effect: An empirical study of Yangtze River Delta urban agglomeration. Energy, 145, 408-416. DOI: 10.1016/j.energy.2018.01.009 Pack, H. (1994). Endogenous Growth Theory: Intellectual Appeal and Empirical Shortcomings. Journal of Economic Perspectives, 8 (1), 55-72 (available in https://www.researchgate.net/publication/4902348_Endogenous_Growth_Theory_Intell ectual_Appeal_and_Empirical_Shortcomings). Paul, C., Techen, A. K., Robinson, J. S. and Helming, K. (2019). Rebound effects in agricultural land and soil management: Review and analytical framework. Journal of Cleaner Production, 227, 1054-1067. DOI: 10.1016/j.jclepro.2019.04.115 Pérez, F., Mas, M., Serrano, L. and Uriel, E. (2019), El stock de capital en España y sus comunidades autónomas. Evolución de la edad media de las inversiones y envejecimiento del capital, Documentos de trabajo, 1/2019, Fundación BBVA (available at https://www.fbbva.es/publicaciones/el-stock-de-capital-en-espana-y-sus-comunidades- autonomas-evolucion-de-la-edad-media-de-las-inversiones-y-envejecimiento-del- capital/). Román-Collado, R., Cansino J. and Botia C., (2018). How far is Colombia from decoupling? Two-level decomposition analysis of energy consumption changes. Energy 148, 687-700. DOI: 10.1016/j.energy.2018.01.141 Safarzadeh, S., Rasti-Barzoki, M. and Hejazi, S. R. (2020). A review of optimal energy policy instruments on industrial energy efficiency programs, rebound effects, and government policies. Energy Policy, 139. DOI: 10.1016/j.enpol.2020.111342 Salmeron, R. and Rodríguez, E. (2017). Métodos cuantitativos para un modelo de regresión lineal con multicolinealidad. Aplicación a rendimientos de letras del tesoro. Revista de Métodos Cuantitativos para la Economía y la Empresa, 24, 169-189. (available at https://www.econstor.eu/bitstream/10419/195386/1/102241657X.pdf). 38 Santarius, T. and Soland, M. (2018). How technological efficiency improvements change consumer preferences: Towards a psychological theory of rebound effects. Ecological Economics, 146, 414-424. DOI: 10.1016/j.ecolecon.2017.12.009 Saunders, H. D. (2000). A view from the macro side: rebound, backfire, and Khazzoom– Brookes. Energy Policy, vol.28 pp. 439 – 449. DOI: 10.1016/S0301-4215(00)00024-0 Seibel, S. (2003). Decomposition analysis of carbon dioxide emissions changes in Germany. Conceptual framework and empirical results. European Commission. Working Papers and Studies. (available at https://ec.europa.eu/eurostat/web/products-statistical- working-papers/-/KS-DB-03-002). Smeets, E., Tabeau, A., van Berkum, S., Moorad, J., van Meijl, H. and Woltjer, G. (2014). The impact of the rebound effect of the use of first generation biofuels in the EU on greenhouse gas emissions: A critical review. Renewable and Sustainable Energy Reviews, 38, 393-403. DOI: 10.1016/j.rser.2014.05.035 Sorrell, S. (2007). The Rebound Effect: an assessment of the evidence for economy-wide energy savings from improved energy efficiency. Project Report (available at https://ukerc.ac.uk/publications/the-rebound-effect-an-assessment-of-the-evidence-for- economy-wide-energy-savings-from-improved-energy-efficiency). Sorrell, S., and Dimitropoulos, J. (2008). The rebound effect: Microeconomic definitions, limitations and extensions. Ecological Economics, 65(3), 636-649. DOI: 10.1016/j.ecolecon.2007.08.013 Solow, R. M. (1957). Technical change and the aggregate production function. The review of Economics and Statistics, 39(3), 312-320. DOI: 10.2307/1926047 Stern, D. I. (2020). How large is the economy-wide rebound effect? Energy Policy, 147. DOI: 10.1016/j.enpol.2020.111870 Timmer, M. P., Dietzenbacher, E., Los, B., Stehrer, R. and de Vries, G. J. (2015). An Illustrated User Guide to the World Input–Output Database: the Case of Global Automotive Production. Review of International Economics, 23, 575–605 (available at http://acdc2007.free.fr/timmer2015.pdf). Thomas, B. A. and Azevedo, I. L. (2013). Estimating direct and indirect rebound effects for US households with input–output analysis. Part 2: Simulation. Ecological Economics, 86, 188-198. DOI: 10.1016/j.ecolecon.2012.12.003 Turner, K. (2009). Negative rebound and disinvestment effects in response to an improvement in energy efficiency in the UK economy. Energy Economics, 31(5), 648- 666. DOI: 10.1016/j.eneco.2009.01.008 Turner, K. and Hanley, N. (2011). Energy efficiency rebound effects and the environmental Kuznets Curve. Energy Economics, 33(5), 709-720. DOI: 10.1016/j.eneco.2010.12.002 Vassileva, I. and Campillo, J. (2014). Increasing energy efficiency in low-income households through targeting awareness and behavioral change. Renewable Energy, 67, 59-63. DOI: 10.1016/j.renene.2013.11.046 39 Vivanco, D. F., Kemp, R. and van der Voet, E. (2016). How to deal with the rebound effect? A policy-oriented approach. Energy Policy, 94, 114-125. DOI: 10.1016/j.enpol.2016.03.054 Wang, Q., Jiang, X. and Li, R. (2017). Comparative decoupling analysis of energy-related carbon emission from electric output of electricity sector in Shandong Province, China. Energy, 127, 78-88. DOI: 10.1016/j.energy.2017.03.111 Wei, T., Zhou, J. and Zhang, H. (2019). Rebound effect of energy intensity reduction on energy consumption. Resources, Conservation and Recycling, 144, 233-239. DOI: 10.1016/j.resconrec.2019.01.012 World Input-Output Database (2016), World Input-Output Tables and World Input- Output Tables, previous years’ prices. (available at www.wiod.org). Xie, S. C. (2014). The driving forces of China ׳ s energy use from 1992 to 2010: An empirical study of input–output and structural decomposition analysis. Energy Policy, 73, 401-415. DOI: 10.1016/j.enpol.2014.05.035 Yan, Z., Ouyang, X. and Du, K. (2019). Economy-wide estimates of energy rebound effect: Evidence from China's provinces. Energy Economics, 83, 389-401. DOI: 10.1016/j.eneco.2019.07.027 Yu, A., Moreno-Cruz, J. and Crittenden, J.C. (2015). Regional energy rebound effect: The impact of economy-wide and sector level energy efficiency improvement in Georgia, USA. Energy Policy, 87, 250-259. DOI: 10.1016/j.enpol.2015.09.020 Zhang, J. and Lawell, C. Y. C. L. (2017). The macroeconomic rebound effect in China. Energy Economics, 67, 202-212. DOI: 10.1016/j.eneco.2017.08.020 40 Activity branches Activity branches IDEA Activity branches BBVA Foundation 1. Agriculture, forestry and fishing Agriculture and fishing. Agriculture, livestock, forestry and fishing 2. Mining and quarrying Non-energy extractive Extractive industries 3. Food, beverages and tobacco Food products; beverages; tobacco products. Food, beverages and tobacco industry. 4. Textile and leather Textile and leather. Textile industry, clothing manufacturing and leather and footwear industry 5. Paper, pulp and printing Wood, paper and printing. Wood and cork industry, paper and graphic arts industry 6. Chemical and petrochemical; non-metallic minerals Chemical (including petrochemical). Non-metallic mineral products. Coke and oil refined petroleum products Chemical industry Pharmaceutical manufacturing Manufacture of rubber and plastic products and other non-metallic mineral products 7. Basic metals & fabricated metal products Iron and steel. Non-ferrous metallurgy. Metallurgy and manufacture of metal products 8. Machinery Machinery Manufacture of computer, electrical, electronic and optical products. 9. Transport equipment Transport equipment Manufacture of transport equipment 10. Other industries Other industries Other industries 11. Construction Construction Construction 12. Energy sector Energy sector Electricity, gas and water; sewerage services and management services. 13. Transport and storage Transport Transport and storage 14. Commercial, services and public services Commercial, services and public services Trade and repair. Accommodation and food services. Information and communications Financial and insurance services. Real estate services. Professional services. Public administration, health and education services. Other services. 41 Table A.1. Productive sectors in used databases.