Economic growth and global warming effects on electricity consumption in Spain: a sectoral study

https://doi.org/10.1007/s11356-022-22312-5 APPLIED ECONOMICS OFENERGY ANDENVIRONMENT INSUSTAINABILITY Economic growth andglobal warming effects onelectricity consumption inSpain: asectoral study Maria del P.Pablo‑Romero1 · AntonioSánchez‑Braza1· DanielGonzález‑Jara1 Received: 6 October 2021 / Accepted: 26 July 2022 © The Author(s) 2022 Abstract This paper analyzes the effect of certain factors on electricity consumption in Spain at a sectoral level. An electricity consumption function has been estimated by using panel data, depending on gross value added (GVA), temperatures, capitalization, and human capital. This function is obtained for total productive electricity consumption and for the agricultural, construction, industrial, service, and public administration sectors, referring to the 17 Autonomous Communities of Spain for the 2000–2013 period. The obtained results show important sectoral differences in the effect that GVA has on electricity consumption, indicating a positive and increasing effect of temperatures above 22°C in the total economy and in the tertiary sector, and a negative effect of temperatures below 18°C in some sectors. These results may indicate that global warming may induce an electricity demand growth in Spain, especially related to cooling needs. The results also highlight the positive effects of capitalization in all sectors, and the negative effects of human capital, except for the public administration sector. In this context, it may be appropriate to carry out policies that mitigate this consumption growth, reinforcing energy efficiency measures, and human capital investments. Keywords Electricity consumption· Economic growth· Economic sectors· Global warming· Spain· Heating and cooling degree days (HDD & CDD) Introduction Signatory countries of the 2015 Paris Agreement were committed to adopting national targets to reduce greenhouse gas (GHG) emissions (UNFCCC 2015). In order to achieve these targets, the control of energy consumption from non-pollut- ing sources has acquired great importance. In this sense, according to the US Energy Information Administration (2016), energy consumption analysis is set at the center of the political and economic debate on climate change, since most GHG emissions are caused by fossil fuel consumption. In this sense, according to the International Energy Agency (IEA 2015), energy is responsible for two-thirds of GHG and 80% of CO2 emissions. All energy sectors are relevant to the reduction of CO2 emissions. However, the electricity sector has special relevance, because it was responsible for 38% of worldwide energy-related CO2 emissions in 2018 (IEA 2019). Therefore, the analysis of electricity production and consumption evolution is relevant in order to define policies aimed at achieving the national targets. In this vein, it is worth noting that between 2010 and 2016, emissions from electricity production grew by 7% worldwide (IEA 2018). Nevertheless, in both the Americas and Europe, increases in renewable share of the energy mix and improvements in fossil generation efficiency led to emission reductions. In fact, the GHG emission intensity of total electricity generation in the EU-27 decreased from 510g CO2e/kWh to 281g CO2e/kWh from 1990 to 2018 (European Environment Agency 2021). In Europe, it is worth highlighting Spain for its lower proportion of emissions generated by the electricity sector. The IEA (2018) report shows that, while at the European level these emissions accounted for 34.5% of total CO2 emissions in 2016, these were 27.7% in Spain. The GHG emission intensity of total electricity generation was 276g CO2e/kWh in Spain in 2018, lower than that by the EU-27 (European Responsible Editor: Roula Inglesi-Lotz * Maria del P. Pablo-Romero [email protected] 1 Department ofEconomic Analysis andPolitical Economy, Faculty ofEconomics andBusiness Sciences, Universidad de Sevilla, Av. Ramon y Cajal 1, 41018Seville, Spain / Published online: 6 August 2022 Environmental Science and Pollution Research (2023) 30:43096–43112 Environment Agency 2021). Therefore, it is particularly interesting to analyze the situation of the electricity sector in this country. Some previous studies have analyzed the Spanish electricity sector from different perspectives. For example, that by Del Río (2008) on the effects of feed-in-tariff premiums on electricity production and that by Burgos-Payán etal. (2013) on the costs and benefits of electricity from renewable energies. Along with them, other studies have focused on the effect of economic growth on Spanish electricity consumption. In this vein, the studies by Ciarreta and Zarraga (2010) and Blázquez etal. (2013) focused on the residential sector. These latter studies are of special interest, since numerous previous studies (see “Literature review”) show a clear link between both variables, which can condition the final growth of electricity consumption. In the context of global warming, the analysis of this study is interesting. According to previous research, electricity consumption may be affected by rising temperatures in Spain (Ministry of the Environment 2006). This country belongs to the Mediterranean area, a territory particularly vulnerable to the global warming process, where temperature increases above the European average are expected (Ministry of the Environment 2006; European Environment Agency 2017). The aim of this paper is to analyze the effect of economic growth on Spanish electricity consumption, at a sectoral level, in a global warming context. To this end, panel data are used to estimate an electricity consumption function that depends on the GVA, its squared and cube values, production factors, and temperatures. The data panel refers to the 17 Spanish Autonomous Communities for the period from 2000 to 2013, years for which there is enough statistical information available. The electricity consumption function is estimated for the total productive electricity consumption (without residential electricity consumption) and for the following sectors: agriculture, construction, industry, service, and public administration. Therefore, this study enlarges the previous literature by analyzing the non-linear relationship between electricity consumption and economic growth from a sectoral perspective. Nevertheless, to our knowledge, this study goes beyond previous studies. Firstly, it refers to Spain, for which there are no previous studies. Secondly, it considers the human and physical capital productive factors as additional factors that influence energy consumption, in non-linear economic growth effects analysis, which has not been considered previously. Finally, this study analyzes the effect of temperatures on electricity consumption by sectors. The results of this study can help to design energy policy measures aimed at achieving the Spanish national targets related to energy and emissions. This study is structured in the following sections: after this introduction, the previous literature is reviewed in “Literature review”; “Data” presents the databases used; “Methodology” describes the methodology used; “Results” presents the results, while “Discussion” discusses these results; and, finally, “Conclusion and policy implications” concludes the study. Literature review Previous studies that analyze the effect of certain factors on energy consumption in general, and on electricity consumption in particular, are very broad. Below, previous literature considered relevant for this study is presented. Energy consumption andeconomic growth: theenergy‑EKC hypothesis The relationship between energy consumption and economic growth was analyzed for the first time by Kraft and Kraft (1978). Since then, numerous studies have focused on this topic. An extensive review of these studies has been carried out previously in Tiba and Omri (2017). These reviews indicate that a large part of the studies show that economic growth causes, and positively affects, energy consumption. Following this line of analysis, some studies have shown that this relationship is not linear, but it depends on income level evolution. The study by Suri and Chapman (1998) may be considered one of the pioneers in analyzing this nonlinear relationship, in what has come to be called the energy Kuznets curve (energy-EKC). The EKC hypothesis states that, initially, low-income countries have low pollution levels, which grow as their income level does. However, once a certain level of income has been reached, economic growth will cause a pollution decrease. In short, the relationship between the variables has an inverted U shape. Following this hypothesis, some researchers have taken the name of energy-EKC (Dong and Hao 2018) to refer to the inverted U shape observed in the relationships between energy consumption and income level growth. Among the studies that have found evidence in favor of the energy-EKC hypothesis, those by Nguyen-Van (2010), Yoo and Lee (2010), and Sbia etal. (2017) may be cited. Among those that do not find said evidence, those by Luzzati and Orsini (2009), Zilio and Recalde (2011), and Pablo- Romero and De Jesús (2016) may be cited. Nevertheless, these last studies observe non-linear relationships between the considered variables. Thus, the previous studies’ results highlight the convenience of considering non-linear relationships between the variables. 43097Environmental Science and Pollution Research (2023) 30:43096–43112 Energy consumption andeconomic growth: sectoral studies The sectoral analysis of the relationships between energy consumption and economic growth is highly relevant. There are numerous previous studies that show differences in the relationships between these variables when considering different economic sectors. Among these studies, those by Costantini and Martini (2010), Zhang and Xu (2012), Rahman etal. (2015), and recently Bal etal. (2022) might be pointed out. Nevertheless, studies that simultaneously consider the non-linear relationships between these variables and a sectoral perspective are scarce. Some studies have referred to specific sectors. Among them, for instance, the studies by Liu etal. (2016), Pablo-Romero and Sánchez-Braza (2017), and Pablo-Romero etal. (2019) refer to the residential sector; those by Katircioglu (2014) and Pablo-Romero etal. (2017a) refer to tourism; and those by Lin and Du (2015), Pablo-Romero etal. (2017b), and Rehermann and Pablo- Romero (2018) to the transport sector. However, very few studies have compared the effect of several sectors simultaneously in a context of non-linearity between the variables. Among these, the early study by Judson etal. (1999), which referred to 69 countries, may be cited. Likewise, the studies by Lescaroux (2011), which referred to 101 countries; the study by Burke and Csereklyei (2016), which referred to 132 countries; and that by Liddle (2017), which referred to the 50 USA states, may also be highlighted. Finally, it is also worth noting the study by Howarth etal. (2017) which referred to Gulf Cooperation Council States. In general, all these studies show clear elasticity of energy consumption with heterogeneity regarding income among the sectors studied. This justifies the sectoral disaggregation in the present study. Energy consumption andeconomic growth: influence ofother factors The inclusion of additional factors in the study of the relationship between economic growth and energy consumption is frequent in the literature. In this sense, Costantini and Martini (2010) consider that the inclusion of these factors is convenient. Thus, different variables have been considered in previous studies, depending on the scope of the study. Among these variables, for example, trade openness has been included in Kasman and Duman (2015), population density in Holden and Norland (2005), and urbanization in Dujardin etal. (2014). Taking into account their relative novelty, the study of the sectoral perspective, and the climate change context, two additional variable groups will be considered in the current study. The first group of variables to take into account refers to the combination of productive factors used in the production process. Numerous previous studies have observed that both physical and human capital have complementarity or substitutability relationships with respect to energy consumption. Among them, the studies by Lin and Wesseh (2013), Pablo-Romero and Sánchez-Braza (2015), and Li and Lin (2016) may be cited. Additionally, in the study by Román- Collado and Colinet (2018), the effect of labor productivity on energy consumption is analyzed. Taking into account this fact, some recent studies have included the physical and/or human capital stock as additional energy consumption explanatory variables. Among them, the studies by Fang and Chen (2017), Salim etal. (2017), Chen and Fang (2018), and Fang and Yu (2020) may be cited, although none of them consider the non-linear relationships between economic growth and energy consumption. Only two studies that, as far as we know, also consider these non-linear effects are those by Balaguer and Cantavella (2018) and by Fang and Wolski (2021) (although only the human capital effects are analyzed in these studies). In addition, recently, the study by Hanif etal. (2020) evaluates the effect of human capital and technology innovation on renewable and non-renewable energy consumption. The second group of variables is introduced in the model to take into account the effect of temperatures on electricity consumption. According to Burke etal. (2015), it is expected that global warming will directly affect productive income. Therefore, it is relevant to deepen the knowledge of the effects it has on the productive sector electricity demand. Numerous econometric studies incorporate temperature in the energy consumption analysis. Auffhammer and Mansur (2014) carry out a review of these works, indicating that they mainly focus on the households’ electricity consumption with a microeconomic perspective. Nevertheless, there are also an increasing number of macroeconomic studies referring to the residential sector, for example, that by Auffhammer (2018) and Nie etal (2018). Despite these macroeconomic studies, there are few macroeconomic studies focused on other economy sectors. In this vein, there are very few studies jointly analyzing temperature and productive effects on electricity consumption (Ge etal. 2017) and considering the temperature effects by economic sectors (Fan etal. 2015). In this last study, the authors conclude that although the results indicate that the temperature changes have more impact on the household and tertiary sectors, the primary and the secondary sectors are also influenced by them. Likewise, some studies have highlighted the effect of temperatures on several productive sectors, especially on the commercial ones (Hirano and Fujita 2012). Therefore, it is interesting to study also the effect of temperature on the electricity demand of these productive sectors. 43098 Environmental Science and Pollution Research (2023) 30:43096–43112 Focusing on Spain, there are few studies analyzing the temperature effects on electricity consumption. To our knowledge, these studies focus on the residential sector (Blázquez etal. 2013; Pablo-Romero etal. 2021), on households (Romero-Jordán etal. 2014), on total electricity consumption (Moral-Carcedo and Vicens-Otero 2005), and on the tourism sector (Pablo-Romero etal. 2017a). Additionally, the study by Moral-Carcedo and Pérez-García (2015) considers the effect on different sectors, although they do not consider the possible non-linear effect of CDD and HDD. Taking into account the previous studies, this study enlarges the existing literature by analyzing the non-linear relationship between electricity consumption and economic growth from a sectoral perspective. In this vein, there is little empirical evidence to date, and none related to Spain. Likewise, this study considers the human and physical capital productive factors and integrates them into a non-linear economic growth effects analysis. To our knowledge, this has not been considered previously. Finally, this study analyzes the non-linear effect of temperatures on electricity consumption by sectors, which is also a novelty in the literature. Data This study analyzes the electricity consumption of the 17 Spanish Autonomous Communities in the 2000–2013 period. The study refers to total productive electricity consumption and to the following sectoral consumption: agriculture, industry, construction, service, and public administration. Total productive electricity consumption refers to the total electricity consumption, excluding residential consumption. The sectoral aggregation detail is shown in Table1. Data onelectricity consumption Data related to electricity consumption are from the annual electrical statistics of the Spanish Ministry of Energy (Ministerio de Industria, Energía y Turismo 2019). Data are expressed in megawatt hours (Mwh) per thousand employed, converted into natural logs. Figure1 shows the total productive electricity consumption per thousand employed in the Spanish Autonomous Communities in 2000 and 2013. Noticeable regional differences with relatively small changes between periods may be observed. Therefore, although greater amounts of electricity are consumed in 2013, the regional composition has not changed significantly. Figure2 shows the electricity consumption trend by productive sectors. Total productive electricity consumption grew until 2008, decreasing later. In 2000, sectors with the highest electricity consumption participation on total productive value were industry (58%), service, and public administration (38%, jointly). In 2013, these sectors were also the ones with the highest participation. Nevertheless, industry sector participation decreased until reaching 49%, while service and public administration increased. Data onGVA andproductive factors Data on total and sectoral production are obtained from the Spanish regional accounting offered by the National Statistics Institute (INE 2019). Gross value added (GVA) at 2010 basic prices, in millions of euros per thousand employees logs, is used. Data on physical and human capital stock are from the Instituto Valenciano de Estudios Económicos database (IVIE 2019). Productive physical capital value is considered. Data are expressed in millions of euros per thousand employees, in logs. Human capital is expressed in terms of persons with higher studies per thousand employees, in logs. Data ontemperatures Data from the Agencia Estatal de Meteorología (2019) have been used to calculate the temperature variables. Specifically, data on average daily temperatures from the main Table 1 Sectoral aggregation detail Sector CNAE 2009 Excluded subsectors Additional included subsectors Agriculture 01 to 03 Industry 05 to 39 23.1, 23.5, 24.4, 24.46, 24.53, 24.54, 33.15 Construction 41 to 43 Service 45 to 82 31.01, 31.02, 31.03, 31.09, 33.12 Public administration 85 to 99 85.51, 85.52, 85.53, 85.56, 85.59 36 to 39 43099Environmental Science and Pollution Research (2023) 30:43096–43112 provinces’ meteorology stations have been used. From these data, CDD (cooling degree days) and HDD (heating degree days) variables have been calculated for the Spanish provinces and autonomous communities. CDD and HDD have been extensively used in recent literature, for example, in Fan and Hyndman (2011), Blázquez etal. (2013), and Mohammadi and Ram (2017). The CDD variable is calculated as follows. Firstly, it is necessary to calculate the CDD variable for each day in a year, as in Eq. (1): where Ti is the average temperature registered in a Spanish province on day i, and Tb is a base or reference temperature. Tb is the threshold temperature. Above this value, it is necessary to turn on the cooling devices. If the temperature is higher than Tb, then γ = 0, and the variable CDD takes the value equal to (Ti-Tb) on day i. If the average temperature is lower than Tb, then γ = 1 and CDD = 0 for that day. (1) CDDi = (1− γ)(Ti − Tb) Fig. 1 Total productive electricity consumption in Spain by Autonomous Communities: 2000 and 2013 (MWh per thousand employed). Source: own elaboration from Ministerio de Industria, Energía y Turismo (2019) Fig. 2 Total productive electricity consumption by sectors: 2000–2013 (MWh). Source: own elaboration from Ministerio de Industria, Energía y Turismo (2019) 43100 Environmental Science and Pollution Research (2023) 30:43096–43112 Once the daily value is calculated, the annual CDD value for each Spanish province is calculated as the sum of the daily values for each province. Finally, the CDD for the Spanish autonomous communities is calculated as the weighted average (according to population) of the provinces’ CDD values. Likewise, the HDD variable is calculated as follows. The HDD variable for each day in a year is calculated for each Spanish province, as in Eq. (2): In this case, if the daily average temperature falls below the reference value (Tb), it is necessary to turn on the heating devices. Then, analytically, γ = 0, and HDD = (Tb—Ti). On the contrary, if it does not fall below that value, γ = 1 and HDD = 0. As before, the annual value is calculated from the sum of the daily values. Likewise, the HDD for each Spanish autonomous community is calculated as the weighted average (according to population) of the provinces’ HDD values. The reference temperatures for calculating the CDD and HDD variables are 22°C and 18°C, respectively. According to Blázquez etal. (2013) and Pablo-Romero etal. (2021), these values can be considered appropriate for Spain. In this study, the CDD and HDD values have been converted into logs. Main descriptive statistics The main descriptive statistics of variables are shown in Table2. The overall statistics refer to the panel sample, the within statistics to the variation from each community’s average, and the between ones to the variation from time’s average. As observed, in general, the typical standard deviation is higher across the Spanish Autonomous Communities than it is across time. Methodology The model The initial theoretical modeling in this study is the following electricity consumption function: where E is the electricity consumption per employee, Y represents the GVA per employee, K is the capital stock per employee used in production, H is the human capital indicator described above, temp refers to some temperature indicator, i indicates the Spanish Autonomous Communities, and t is a time variable. (2) HDDi = (1− γ)(Tb − Ti) (3) Eit = f(Yit,Kit,Hit ,tempit ) Therefore, electricity consumption depends on the productive level, productive factor combination, and temperatures. Two considerations related to this model specification should be noted. Firstly, in some previous recent studies (such as, for example, Balaguer and Cantavella 2018; Chen and Fang 2018; Fang and Chen 2017; and Salim etal. 2017), it has been indicated that electricity consumption depends not only on the level of production level, but also on the way in which the productive factors are combined in the production process. Therefore, the aforementioned studies include physical and/or human capital as additional explanatory variables. In this study, that research line is followed and both variables are included in the electricity consumption function specification. Secondly, previous studies consider also that temperatures may influence energy or electricity consumption (Gam and Rejeb 2012; Lee and Chiu 2011, among others). In this study, this research line is followed, and therefore temperature variables have been included in the model. In addition to these considerations, two other questions may be taken into account. On the one hand, many previous studies have observed that the relationship between production and energy consumption is not linear (Nguyen-Van 2010; Pablo- Romero and Sánchez-Braza 2017; Sbia etal. 2017; Suri and Chapman 1998, among others). Thus, when specifying the energy consumption function, not only the production variable has been considered, but also the squared and cubed value of production. This consideration has given rise to what has come to be called the energy-EKC hypothesis (Dong and Hao 2018). On the other hand, the effect of temperatures on energy consumption has also been considered non-linear, in a double sense. Firstly, previous studies have shown that the temperatures’ effect on energy consumption varies depending on whether the temperatures are low or high. That is to say, it varies according to the heating or refrigeration needs, generating greater energy consumption when both situations appear. As a consequence, two variables have been used, instead of one, to analyze the temperatures’ effect on electricity consumption. Usually, in previous studies (Blázquez etal. 2013; Fan and Hyndman 2011; Mohammadi and Ram 2017, for example), these variables are the CDD and HDD. Secondly, other authors have additionally considered that these two variables may also have a non-linear effect on electricity consumption. Thereby, in these studies (for example, in Boyd 2014), the CDD and HDD squared values have also been included to analyze this non-linear temperature effect. In this study, this last specification has been considered. Taking into account the above considerations, the electricity consumption function in this study is specified as follows: (4) lnE it =Ait +𝛽1lnYit +𝛽2(lnYit) 2 +𝛽3(lnYit)3+𝛽4lnKit +𝛽5lnHit +𝛽6lnHDDit +𝛽 7 (lnHDD it )2+𝛽 8 lnCDD it +𝛽 9 (lnCDD it )2+e it 43101Environmental Science and Pollution Research (2023) 30:43096–43112 Table 2 Descriptive statistics Electricity consumption per employee in log Mean Std. dev Min Max Observations  Total productive sectors Overall 9.088 0.400 7.993 9.975 N = 238 n = 17 T = 14 Between 0.400 8.311 9.875 Within 0.093 8.316 9.338  Agriculture sector Overall 8.435 0.810 5.997 10.832 N = 238 n = 17 T = 14 Between 0.753 6.860 9.432 Within 0.345 7.549 10.410  Construction sector Overall 7.008 0.710 5.069 9.149 N = 238 N = 17 T = 14 Between 0.485 5.649 7.470 Within 0.530 5.414 8.928  Industrial sector Overall 10.174 0.619 8.657 11.654 N = 238 N = 17 T = 14 Between 0.611 9.066 11.462 Within 0.174 9.035 11.447  Public administration sector Overall 8.284 0.421 6.503 8.965 N = 238 N = 17 T = 14 Between 0.377 7.097 8.822 Within 0.207 6.908 8.854  Servicesector Overall 8.732 0.378 7.309 9.438 N = 238 N = 17 T = 14 Between 0.343 7.477 9.061 Within 0.179 7.371 9.309 GVA per employees in log Mean Std. dev Min Max Observations  Total productive sectors Overall 10.776 0.089 10.545 10.994 N = 238 N = 17 T = 14 Between 0.078 10.636 10.918 Within 0.046 10.685 10.899  Agriculture sector Overall 10.242 0.220 9.738 10.780 N = 238 N = 17 T = 14 Between 0.171 9.961 10.524 Within 0.144 9.893 10.609  Construction sector Overall 10.803 0.166 10.365 11.298 N = 238 N = 17 T = 14 Between 0.114 10.590 11.074 Within 0.124 10.544 11.138  Industrial sector Overall 11.000 0.143 10.661 11.367 N = 238 N = 17 T = 14 Between 0.103 10.822 11.162 Within 0.101 10.778 11.225  Public administration sector Overall 10.567 0.052 10.448 10.689 N = 238 N = 17 T = 14 Between 0.033 10.501 10.621 Within 0.040 10.464 10.660  Service sector Overall 10.862 0.102 10.587 11.120 N = 238 N = 17 T = 14 Between 1.393 1.887 6.355 Within 0.495 1.107 5.893 Productive physical capital per employees in log Mean Std. dev Min Max Observations  Total productive sectors Overall 11.374 0.177 10.993 11.806 N = 238 N = 17 T = 14 Between 0.129 11.184 11.602 Within 0.125 11.162 11.670  Agriculture sector Overall 11.212 0.454 10.128 12.572 N = 238 N = 17 T = 14 Between 0.420 10.668 12.056 Within 0.197 10.672 11.728  Construction sector Overall 11.553 1.486 8.800 15.511 N = 238 N = 17 T = 14 Between 1.472 9.224 14.807 Within 0.401 11.045 12.460 43102 Environmental Science and Pollution Research (2023) 30:43096–43112 where A represents the sum of the temporary and individual effect, and e is the error term. The estimated coefficients related to production (variable Y) inform about the relationship between electricity consumption and production. In this sense, as Pablo-Romero etal. (2021) state, it is possible to find different types of relationships. The following non-linear relationships should be noted: β1 > 0, β2 > 0, and β3 = 0, progressively increasing relationship. β1 > 0, β2 < 0, and β3 = 0, U-shaped inverted relationship. β1 < 0, β2 > 0, and β3 = 0, U-shaped relationship. β2 2-3*β3β1 > 0 and β3 > 0, N-shaped relationship. β2 2-3*β3β1 > 0 and β3 < 0, N-shaped inverted relationship. Likewise, the β6 and β7 coefficients and the β8 and β9 ones inform about the relationships between electricity consumption and that of cold and heating temperature, respectively. For the two coefficient pairs, if both are positive, then there is an increasing relationship between both variables. If the second coefficient is equal to zero, the relationship is linear. If the first coefficient is positive and the second is negative, there is a positive but decreasing relationship. Finally, if the Table 2 (Continued)  Industrial sector Overall 12.083 1.215 9.822 14.462 N = 238 N = 17 T = 14 Between 1.235 10.147 14.100 Within 0.184 11.758 12.556  Public administration sector Overall 11.341 0.203 10.765 11.733 N = 238 N = 17 T = 14 Between 0.189 10.967 11.609 Within 0.086 11.114 11.535  Service sector Overall 11.008 0.179 10.672 11.465 N = 238 N = 17 T = 14 Between 0.164 10.755 11.348 Within 0.081 10.851 11.276 Human capital per employees in log Mean Std. dev Min Max Observations  Total productive sectors Overall 5.959 0.235 5.119 6.549 N = 238 N = 17 T = 14 Between 0.197 5.463 6.332 Within 0.136 5.614 6.307  Agriculture sector Overall 5.155 0.591 3.852 7.136 N = 238 N = 17 T = 14 Between 0.455 4.362 6.475 Within 0.392 3.891 6.303  Construction sector Overall 5.469 0.353 4.449 6.246 N = 238 N = 17 T = 14 Between 0.231 4.982 5.824 Within 0.273 4.855 6.189  Industrial sector Overall 5.852 0.333 4.629 6.478 N = 238 N = 17 T = 14 Between 0.292 5.184 6.253 Within 0.173 5.297 6.316  Public administration sector Overall 5.931 0.199 5.363 6.476 N = 238 N = 17 T = 14 Between 0.188 5.648 6.415 Within 0.078 5.647 6.134  Service sector Overall 6.186 0.268 5.368 6.847 N = 238 N = 17 T = 14 Between 0.241 5.638 6.665 Within 0.130 5.850 6.575 Temperature Mean Std. dev Min Max Observations  HDD Overall 6.289 1.263 0.969 7.771 N = 238 N = 17 T = 14 Between 1.274 2.539 7.647 Within 0.249 4.719 7.622  CDD Overall 4.159 1.442 − 0.916 6.579 N = 238 N = 17 T = 14 Between 1.393 1.887 6.355 Within 0.495 1.107 5.893 43103Environmental Science and Pollution Research (2023) 30:43096–43112 first coefficient is negative and the second is positive, there is a negative but growing relationship. Thus, the study of these coefficients will inform about the CDD and HDD non-linear effect on electricity consumption. Estimation procedure Two considerations have been taken into account when estimating Eq.(4). Firstly, the introduction of some variables in quadratic or cubic terms may generate multicollinearity problems. However, according to Jaccard and Turrisi (2003), these problems can be mitigated by centering their values. Therefore, variables in Eq.(4) have been transformed in terms of geometric mean deviations. This transformation implies that β1, β6, and β8 coefficients now represent the electricity consumption elasticity with respect to production, to HDD, and to CDD, at the sample center point. Secondly, once the variables have been transformed, the stochastic nature of the series has been studied. Table3 shows Pesaran’s cross-sectional (CD) dependence test results (Pesaran 2004). For all series, the Pesaran CD test results indicate that the cross-sectional independence null hypothesis is rejected. Thus, a second-generation panel unit root test should be performed. In this study, the cross-sectionally augmented Im, Pesaran, and Shin panel unit root tests (CIPS) by Pesaran (2007) are used. Table4 shows the CIPS test results for series in levels and first differences when considering intercept and trend. The results indicate that all variables cannot be considered stationary at a 5% significance level, but are I(1). Therefore, estimating in first differences may be appropriate. According to Stern etal. (2017), expressing Eq.(4) in first differences is like expressing the energy-EKC in long-run growth rates terms. Therefore, the variables have also been converted into first differences. Taking into account both variable transformations, Eq.(4) is now expressed as follows: where Δ A it = δt, Δ indicates that variables are in first differences terms, and the line on the variables indicates they are in terms of geometric mean deviations. Finally, autocorrelation and homoscedasticity have also been analyzed to determine the estimated model of (5). Therefore, Wooldridge (2002) and the Wald test for homoscedasticity (Greene 2000) have been performed. Taking into account the test results, Eq.(5) is estimated by using the feasible generalized least squares (FGLS) method when controlling for autocorrelation, heteroscedasticity, and contemporaneous correlation. Results Tables5, 6, 7, 8, 9 and 10 show the estimate results of Eq.(5), Table5 shows the results when referring to all productive sectors, and Tables6, 7, 8, 9 and 10 show the results for the considered sectors. All estimates have been performed by using FGLS and include time dummies. Estimate results oftotal productive sector Column 3 in Table5 shows the estimate results of Eq.(5) for total productive sectors. The results indicate non-linear relationships between production and electricity consumption. Specifically, an N-shaped relationship is observed. Thus, although a negative relationship is observed for low (5) Δ lnEit =𝛿t+𝛽1ΔlnYit +𝛽2Δ(lnYit )2+𝛽3Δ(lnYit )3 +𝛽4ΔlnKit +𝛽5ΔlnHit +𝛽6ΔlnHDDit +𝛽7Δ(lnHDDit ) 2 +𝛽8Δ lnCDD it +𝛽9Δ( lnCDD it ) 2 + e it Table 4 CIPS test t-bar statistics *** denotes significance at the 1% level, ** at the 5% level, and * at the 10% level. Logs calculated with an iterative process from 0 to 3 based on F-joint test. The truncated version of the test is applied Variables Level First differences Intercept and trend Intercept and trend E − 2.702* − 3.362*** Y − 2.435 − 4.286*** Y2 − 0.612 − 3.137 *** Y3 − 2.035 − 2. 771** K − 2.290 − 3.179*** H − 1.889 − 3.379*** HDD − 3.008** − 4.249*** HDD2 − 2.968** − 4.060*** CDD − 2.776* − 3.323*** CDD2 − 3.002** − 3.782 *** Table 3 Pesaran’s CD tests *** denotes significance at the 1% level, ** at the 5% level, and * at the 10% level Variables CD test E6.21*** Y34.27*** Y22.45** Y327.99 *** K42.83*** H41.13*** HDD 28.85*** HDD21.84* CDD 27.79*** CDD21.67* 43104 Environmental Science and Pollution Research (2023) 30:43096–43112 Urban Stud 42(12):2145–2166. https:// doi. org/ 10. 1080/ 00420 98050 03320 64 Howarth N, Galeotti M, Lanza A, Dubey K (2017) (2017) Economic development and energy consumption in the GCC: an international sectoral analysis. Energy Transit 1:6. https:// doi. org/ 10. 1007/ s41825- 017- 0006-3 IEA (2015) Energy and climate effects. International Energy Agency, París IEA (2018) CO2 emissions from fuel combustion 2018: overview International Energy Agency, París IEA (2019) Global Energy & CO2 Status Report 2019 International Energy Agency, París INE (2019) INEbase, Economía/Cuentas Económicas. Instituto Nacional de Estadística, Madrid. https:// www. ine. es/ Accessed may 2021 IVIE (2019) Bases de Datos/El Stock y los Servicios de Capital, Capital Humano. Instituto Valenciano de Investigaciones Económica, Valencia. https:// www. ivie. es/ Accessed may 2021 Judson RA, Schmalensee R, Stoker TM (1999) Economic development and the structure of demand for commercial energy. Energy J 20:29–57. https:// doi. org/ 10. 5547/ ISSN0 195- 6574- EJ- Vol20- No2-2 Kasman A, Duman YS (2015) CO2 emissions, economic growth, energy consumption, trade and urbanization in new EU member and candidate countries: a panel data analysis. Econ Model 44:97–103. https:// doi. org/ 10. 1016/j. econm od. 2014. 10. 022 Katircioglu ST (2014) International tourism, energy consumption, and environmental pollution: the case of Turkey. Renew Sust Energ Rev 36:180–187. https:// doi. org/ 10. 1016/j. rser. 2014. 04. 058 Kraft J, Kraft A (1978) On the relationship between energy and GNP. The Journal of Energy and Development 3(2):401–403 Lee C, Chiu Y (2011) Electricity demand elasticities and temperature: evidence from panel smooth transition regression with instrumental variable approach. Energy Econ 33(5):896–902. https:// doi. org/ 10. 1016/j. eneco. 2011. 05. 009 Lescaroux F (2011) Dynamics of final sectoral energy demand and aggregate energy intensity. Energy Policy 39(1):66–82. https:// doi. org/ 10. 1016/j. enpol. 2010. 09. 010 Li J, Lin B (2016) Inter-factor/inter-fuel substitution, carbon intensity, and energy-related CO2 reduction: empirical evidence from China. Energy Econ 56:483–494. https:// doi. org/ 10. 1016/j. eneco. 2016. 04. 001 Li DH, Yang L, Lam JC (2012) Impact of climate change on energy use in the built environment in different climate zones–a review. Energy 42(1):103–112. https:// doi. org/ 10. 1016/j. energy. 2012. 03. 044 Li J, Yang L, Long H (2018) Climatic impacts on energy consumption: intensive and extensive margins. Energy Econ 71:332–343. https:// doi. org/ 10. 1016/j. eneco. 2018. 03. 010 Liddle B (2017) Modeling disaggregated energy consumption: considering nonlinearity, asymmetry, and heterogeneity by analyzing US state-level data. International Association for Energy Economics (IAEE) Energy Forum 1st Quarter2017, 35–39 Lin B, Wesseh KP (2013) Estimates of inter-fuel substitution possibilities in Chinese chemical industry. Energy Econ 40:560–568. https:// doi. org/ 10. 1016/j. eneco. 2013. 08. 017 Lin B, Du K (2015) Energy and CO2 emissions performance in China’s regional economies: do market-oriented reforms matter? Energy Policy 78:113–124. https:// doi. org/ 10. 1016/j. enpol. 2014. 12. 025 Liu Y, Gao Y, Hao Y, Liao H (2016) The relationship between residential electricity consumption and income: a piecewise linear model with panel data. Energies 9(10):831. https:// doi. org/ 10. 3390/ en910 0831 Luzzati T, Orsini M (2009) Investigating the energy-environmental Kuznets curve. Energy 34(3):291–300. https:// doi. org/ 10. 1016/j. energy. 2008. 07. 006 Ministerio de Industria, Energía y Turismo (2019) Informes, Estadísticas y Balances Energéticos. Ministerio para la Transición Ecológica, Secretaría de Estado de Energía, Ministerio de Industria, Energía y Turismo, Madrid Mohammadi H, Ram R (2017) Convergence in energy consumption per capita across the US states, 1970–2013: an exploration through selected parametric and non-parametric methods. Energy Econ 62:404–410. https:// doi. org/ 10. 1016/j. eneco. 2016. 07. 002 Moral-Carcedo J, Vicens-Otero J (2005) Modelling the non-linear response of Spanish electricity demand to temperature variations. Energy Econ 27(3):477–494. https:// doi. org/ 10. 1016/j. eneco. 2005. 01. 003 Moral-Carcedo J, Pérez-García J (2015) Temperature effects on firms’ electricity demand: an analysis of sectorial differences in Spain. Applied En 142:407–425. https:// doi. org/ 10. 1016/j. apene rgy. 2014. 12. 064 Nie HG, Kemp R, Xu JH, Vasseur V, Fan Y (2018) Drivers of urban and rural residential energy consumption in China from the perspectives of climate and economic effects. J Clean Prod 172:2954– 2963. https:// doi. org/ 10. 1016/j. jclep ro. 2017. 11. 117 Nguyen-Van P (2010) Energy consumption and income: a semiparametric panel data analysis. Energy Econ 32(3):557–563. https:// doi. org/ 10. 1016/j. eneco. 2009. 08. 017 Pablo-Romero MP, De Jesús J (2016) Economic growth and energy consumption: the energy-environmental Kuznets curve for Latin America and the Caribbean. Renew Sust Energ Rev 60:1343– 1350. https:// doi. org/ 10. 1016/j. rser. 2016. 03. 029 Pablo-Romero MP, Sánchez-Braza A (2015) Productive energy use and economic growth: energy, physical and human capital relationships. Energy Econ 49:420–429. https:// doi. org/ 10. 1016/j. eneco. 2015. 03. 010 Pablo-Romero MP, Sánchez-Braza A (2017) Residential energy environmental Kuznets curve in the EU-28. Energy 125:44–54. https:// doi. org/ 10. 1016/j. energy. 2017. 02. 091 Pablo-Romero MP, Sánchez-Braza A, Sánchez-Rivas J (2017a) Relationships between hotel and restaurant electricity consumption and tourism in 11 European Union countries. Sustainability 9(11):2109. https:// doi. org/ 10. 3390/ su911 2109 Pablo-Romero MP, Cruz L, Barata E (2017b) Testing the transport energy-environmental Kuznets curve hypothesis in the EU27 countries. Energy Econ 62:257–269. https:// doi. org/ 10. 1016/j. eneco. 2017. 01. 003 Pablo-Romero MP, Sánchez-Braza A, Galyan A (2019) Relationship between economic growth and residential energy use in transition economies. Clim Dev 11(4):338–354. https:// doi. org/ 10. 1080/ 17565 529. 2018. 14427 89 Pablo-Romero MP, Pozo-Barajas R, Molleda-Jimena G (2021) Residential energy environmental Kuznets curve extended with nonlinear temperature effects: a quantile regression for Andalusian (Spain) municipalities. Environ Sci Pollut Res 28:48984–48999. https:// doi. org/ 10. 1007/ s11356- 021- 13608-z Pesaran MH (2004) General diagnostic tests for cross section dependence in panels. Cambridge Working Papers WP0435, Faculty of Economics, University of Cambridge, Cambridge Pesaran MH (2007) A simple panel unit root test in the presence of cross-section dependence. J Appl Econ 22(2):265–312. https:// doi. org/ 10. 1002/ jae. 951 Rahman MS, Junsheng H, Shahari F, Aslam M, Masud MM, Banna H, Liya M (2015) Long-run relationship between sectoral productivity and energy consumption in Malaysia: an aggregated and disaggregated viewpoint. Energy 86:436–445. https:// doi. org/ 10. 1016/j. energy. 2015. 04. 049 Rehermann F, Pablo-Romero MP (2018) Economic growth and transport energy consumption in the Latin American and Caribbean countries. Energy Policy 122:518–527. https:// doi. org/ 10. 1016/j. enpol. 2018. 08. 006 43111Environmental Science and Pollution Research (2023) 30:43096–43112 Román-Collado R, Colinet MJ (2018) Are labour productivity and residential living standards drivers of the energy consumption changes? Energy Econ 74:746–756. https:// doi. org/ 10. 1016/j. eneco. 2018. 07. 030 Romero-Jordán D, Peñasco C, Del Río P (2014) Analysing the determinants of household electricity demand in Spain. An econometric study. Int J Electr Power Energy Syst 63:950–961. https:// doi. org/ 10. 1016/j. ijepes. 2014. 06. 005 Salim R, Yao Y, Chen GS (2017) Does human capital matter for energy consumption in China? Energy Econ 67:49–59. https:// doi. org/ 10. 1016/j. eneco. 2017. 05. 016 Sbia R, Shahbaz M, Ozturk I (2017) Economic growth, financial development, urbanisation and electricity consumption nexus in UAE. Econ Res-Ekon Istraz 30(1):527–549. https:// doi. org/ 10. 1080/ 13316 77X. 2017. 13057 92 Schulze M, Nehler H, Ottosson M, Thollander P (2016) Energy management in industry – a systematic review of previous findings and an integrative conceptual framework. J Clean Prod 112(5):3692– 3708. https:// doi. org/ 10. 1016/j. jclep ro. 2015. 06. 060 Silva S, Soares I, Pinho C (2020) Climate change impacts on electricity demand: the case of a Southern European country. Utilities Policy 67:101115. https:// doi. org/ 10. 1016/j. jup. 2020. 101115 Stern DI, Gerlagh R, Burke PJ (2017) Modeling the emissions−income relationship using long-run growth rates. Environ Dev Econ 22(6):699–724. https:// doi. org/ 10. 1017/ S1355 770X1 70001 09 Suri V, Chapman D (1998) Economic growth, trade and energy: implications for the environmental Kuznets curve. Ecol Econ 25(2):195–208. https:// doi. org/ 10. 1016/ S0921- 8009(97) 00180-8 Tiba S, Omri A (2017) Literature survey on the relationships between energy, environment and economic growth. Renew Sust Energ Rev 69:1129–1146. https:// doi. org/ 10. 1016/j. rser. 2016. 09. 113 U.S. Energy Information Administration (2016) International Energy Outlook 2016. Office of Energy Analysis, U.S. Department of Energy, Washington DC. UNFCCC (2015) Adoption of the Paris agreement - COP21. United Nations Framework Convention on Climate Change, París Van Heddeghem W, Lambert S, Lannoo B, Colle D, Pickavet M, Demeester P (2014) Trends in worldwide ICT electricity consumption from 2007 to 2012. Comput Commun 50:64–76. https:// doi. org/ 10. 1016/j. comcom. 2014. 02. 008 Wooldridge J (2002) Econometric analysis of cross section and panel data. MIT Press, Cambridge Yoo SH, Lee JS (2010) Electricity consumption and economic growth: a cross-country analysis. Energy Policy 38(1):622–625 Zhang C, Xu J (2012) Retesting the causality between energy consumption and GDP in China: evidence from sectoral and regional analyses using dynamic panel data. Energy Econ 34(6):1782–1789 Zilio M, Recalde M (2011) GDP and environment pressure: the role of energy in Latin America and the Caribbean. Energy Policy 36(12):7941–7949 Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 43112 Environmental Science and Pollution Research (2023) 30:43096–43112