Electricity consumption and adaptation to climate change: heterogeneity across regions and economic sectors in China

Electricity consumption and adaptation to climate change: heterogeneity across regions and economic sectors in China | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Electricity consumption and adaptation to climate change: heterogeneity across regions and economic sectors in China Hanyi Chen, Qingran Li, Xuebin Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3944484/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract There is growing empirical evidence that a warming climate will induce adaptation of the electricity system. Less understood is the extent to which climate change could exert on electricity demand given heterogeneity in consumption patterns and the latent mechanisms driving these patterns. We statistically estimate an asymmetric U-shaped temperature response function of city-level daily electricity consumptions using data in years 2018–2019 and examine heterogeneous responses across regions and economic sectors. Benefiting from the high-frequency electricity consumption data that covers 92 Chinese cities and multiple economic sectors, our findings speak to both the intensive margin as well as the extensive margin adaptation to climate change. We find that access to district heating can explain the asymmetry slopes in temperature-load responses. We also find that although the marginal load responses are statistically significant for all sectors in both high (> 25.6°C) and low (< 6.6°C) temperature ranges, the tertiary (service) sector load is more sensitive to temperature changes. Taking account of the tertiarization trend, we predict about 66% of the cities will experience more than 3% increase in their summer daily electricity consumption before year 2040. This will likely require substantial investments to expand power grid capacity and to build up energy storage. electricity consumption climate change adaptation China Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction The COP26 global climate summit ended with a clear consensus that significant and immediate carbon mitigation efforts are required to prevent catastrophic climate change. Future emissions pathway, though started to bend downward due to clean energy policies and technologies, is still not enough to avoid temperature rise. Assuming nations will actually follow through on their climate pledges, the carbon abatement would be far from assured to achieve the 1.5 degree goal (Meinshausen et al., 2022 ). The need to mitigate carbon emissions has become more pressing as extreme temperatures driven by climate change will influence economic activities (Cronin et al., 2018 ), which is reflected by surging energy consumptions (Auffhammer et al., 2017 ; Perera et al., 2020 ; van Ruijven et al., 2019 ) and increasing capital investments to meet the demand (Khan et al., 2021 ). Less clear, however, is the extent to which climate change could exert on energy demand given regional heterogeneity in consumption patterns and the latent mechanisms driving these patterns. More extreme temperatures (hotter summer and/or colder winter) are likely to boost electricity demand for heating and cooling, which depends on the climate scenarios, electrification rates, urbanization degrees, and various socio-economic attributes (Auffhammer, 2022 ; Garrido-Perez et al., 2021 ; Gupta, 2016 ; Li et al., 2019 ; Propato et al., 2021 ; Teng et al., 2022 ; van Ruijven et al., 2019 ; C. Zhang et al., 2019 ). However, due to data availability and reliability issues, empirical studies in the developing economy context were quite sparse. Most studies use regional aggregates (Chen et al., 2021 ; De Cian & Sue Wing, 2019 ; Fan et al., 2019 ; Propato et al., 2021 ; Teng et al., 2022 ; C. Zhang et al., 2019 ) to estimate the temperature-load response functions in lack of data with higher temporal resolutions (daily or weekly load), which tend to aggregate away the high-frequency variations and can potentially bias the load responses (Ghanem & Smith, 2021 ). When studies are available, they tend to focus on adaptive responses of residential electricity consumptions (Davis & Gertler, 2015 ; Li et al., 2019 ; Teng et al., 2022 ; S. Zhang et al., 2022 ), where responses to extreme weather events are most intense (Li et al., 2019 ). This suggests a knowledge gap in understanding the climate-adaptive responses of other economic sectors (e.g. industry, commercial, agriculture). Closing this knowledge gap is critical given non-residential electricity consumptions are the major driver of global electricity demand, which is about 73% between 2018 and 2019 with China and United States being the top two electricity consuming countries (IEA, 2021 ). If electricity consumptions in these economic sectors are sensitive to temperature variability, this will then increase the vulnerability of the economy to climate extremes, requiring intensive investments to enhance power supply resilience and reliability. Given the total gross electricity production in non-OECD countries surpassed OECD countries after 2010 and continued to grow rapidly (IEA, 2021 ), the confluence of this growth and temperature variability is likely to amplify the impact of climate shocks. This would introduce disparities in countries’ abilities to adapt to climate change and enlarge the gap in requisite investments associated with electrification and mitigation (Pachauri et al., 2022 ). In order to design efficient and equitable national and international climate policies, increasing empirical evidence in the developing and/or emerging economy context is needed. Our work address this issue. We assess the electricity-climate responses of various economic sectors in China — a fast-growing economy with increasing electricity demand and over 40% produced by coal in 2020 (IEA, 2022 ). Using a high-frequency daily electricity consumption data set from 92 Chinese cities, our efforts emphasize the spatial heterogeneity (regions and cities) in temperature-load response functions. We subsequently examine the influence of tertiarization — increasing shares of the tertiary sector in economic outputs — on electricity demand projections under different climate change scenarios. We extrapolate implications of our findings for the long-term power sector resource planning and for regional and/or sectoral policy designs to adapt to climate change. Our work builds upon and contributes to the existing literature of climate adaptation in energy consumption. First, this study examines the temperature response heterogeneity across different economic sectors and regions using high-frequency electricity consumption data, whereas previous studies have examined diversity in a limited set (e.g. rural and urban residential) with high-frequency data (Li et al., 2019 ) or have examined geo-diversity (e.g. city/provincial panel) with lower temporal resolutions (Chen et al., 2021 ; Fan et al., 2019 ; Teng et al., 2022 ; S. Zhang et al., 2022 ) and aggregated sectoral consumptions (Propato et al., 2021 ). The sectoral-distinction is important. Electricity consumption patterns vary with economic sectors. The temperature elasticities in these sectoral electricity consumptions are associated with the portion which heating and cooling account for in the total end-use consumptions. Modeling load responses to temperature variability, if not discriminating the sectors, is likely to introduce bias to electricity demand projections. This is particularly true for regions where the economy is heavily oriented towards service sectors, as service-sector outputs usually exhibit higher temperature elasticities. Our study accounts for the sectoral-distinction, permitting examinations on the electricity demand robustness to tertiarization, which has become a major force driving economic growth and transformation in middle- and low-income countries (Nayyar et al., 2021 ). Our findings therefore add to the literature on extensive margin of climatic impacts (Auffhammer & Mansur, 2014 ), by addressing the marginal impact via tertiarization, whereas many past studies examined how energy consumptions respond to short-term weather shocks (the intensive margin). Moreover, this study contributes to empirical evidences of spatial diversity in climatic impacts. The north versus south distinction is critical as climate change not only lead to increasing global surface temperature, but also alters the patterns of different climate zones, thus resulting to notable shifts in electricity consumption patterns (Wenz et al., 2017 ). Our analysis reveals a significant north-south sensitivity difference to temperature, especially in the cold temperature range. This difference can be attributed to the extent to which district heating is available, though mostly powered by fossil fuels in China (e.g. natural gas, residual heat from coal power plants). Hence, modeling the spatial-heterogeneous temperature responses helps to refine predictions of heating energy consumptions, thus informing future energy management and pollution controls. 2. Methodology 2.1 Data Our analyses primarily utilize two data sources: electricity consumption data and the weather data with daily variations. Data on electricity consumption originate from the China Electric Power Research Institute, a research center affiliated to the State Grid Corporation of China. The metered consumption data (in GWhs) are aggregated at the city and sector level. For 92 Chinese cities (see SI Appendix , Table S1 for the list of cities), we obtain daily electricity consumption data of 20 non-residential economic sectors and the residential sector. The two-year time series ranges from January 1, 2018 to December 31, 2019. We also attain daily peak load data for 11 of the 92 cities. SI Appendix , Fig. S1 illustrates the variation of daily total electricity load by sector (secondary and tertiary) and the residential consumption load. For the two-year time period (2018–2019), the temperature and other weather variables are from the Global Surface Summary of Day (GSOD) database accessible via National Centers for Environmental Information (NOAA, n.d.), including daily mean temperature, dew point (to compute relative humidity), precipitation, and wind speed. To integrate this data into the city panel, we match the city with its nearest weather station. 2.2 Regression Model Our baseline model follows Li et al. ( 2019 ), which uses splines to model the nonlinear relationship between temperature and electricity consumption. Our econometric model is specified by the log-linear equation below for city \(i\) and day \(t\) : $$\text{ln}\left(E{C}_{it}\right)=\alpha +\sum {\beta }_{j}{f}_{j}\left(tem{p}_{it}\right)+\gamma Weathe{r}_{it}+\theta nonWorkda{y}_{t}+{\eta }_{i,m,y}+{\epsilon }_{it}.$$ The dependent variable is the natural logarithm of a city’s daily electricity consumption 1 , and \(tem{p}_{it}\) is the city’s daily mean temperature. To control for non-temperature confounders, the \(Weathe{r}_{it}\) variables include daily mean precipitation, wind speed, and relative humidity. The dummy variables \(nonWorkda{y}_{t}\) and \({\eta }_{i,m,y}\) control for non-workdays and the city-month-year fixed effects. This is because electricity consumption load tends to be lower in weekends and holidays, and the cities and industries can face different electricity price in same month of the different years. Standard errors are clustered at the city-week level 2 . Summary statistics of these variables are provided in SI Appendix , Table S2. The functions \({f}_{j}(.)\) are linear splines with a number of defined temperature thresholds or knots, \({k}_{j}\) , which are specified as: $${f}_{0}\left(tem{p}_{it}\right)=tem{p}_{it}, \text{a}\text{n}\text{d} {f}_{j}\left(tem{p}_{it}\right)=\left\{\begin{array}{c}tem{p}_{it} if tem{p}_{it}>{k}_{j}\\ 0 if tem{p}_{it}\le {k}_{j}\end{array}\right. \text{f}\text{o}\text{r} j\ge 1.$$ This approach models a smooth response function and permits changing slopes, implying flexible variability of the responses in different temperature ranges. This flexibility increases with the number of knots used to model the spline function. A data-driven algorithm is used to select the level of flexibility for the spline function ( SI Appendix , Model Selection), i.e., the number of equally spaced quantile knots for the linear spline. The result is a choice of \(j=4\) , or four knots. To explore heterogeneity by geographic region, we divide the city panel to North-heating and South-non-heating groups 3 . For all cities in the North or South group, we apply the baseline city-panel regression model with linear temperature splines, and compare the regional heterogeneity result from variation in heating infrastructures. To explore heterogeneity by economic sector, we estimate the baseline city-panel regression model for each economic sector. The summary statistics of cities’ sectoral electricity consumptions are provided in SI Appendix , Table S3. 2.3 Load Projections To develop future temperature portfolios for each city under different climate change scenarios, we obtain temperature data under two climate Representative Concentration Pathways (RCPs) — RCP4.5 and RCP8.5 — from the NASA Earth Exchange Global Daily Downscaled Projections (NASA, n.d.). The projections are derived from 21 General Circulation Model (GCM) runs and have been downscaled to a spatial resolution of 0.25 degrees (~ 25 x 25 km). For each of the 21 GCMs, daily maximum and minimum temperatures are downloaded for historical years 2001–2005 and two forecasted time periods, 2051–2055 and 2091–2095. To match the temperature data with our city panel, for each city, we calculate the average temperature of the four closest coordinates to the centroid of the city. The city-wide load projection procedure follows Li et al. ( 2019 ), and adopts the common assumption in the literature (Auffhammer et al., 2017 ; Davis & Gertler, 2015 ; Li et al., 2019 ) that the fixed effects and other non-temperature variables remain the same for the past and future. For each city in our sample, we estimate the city-specific temperature-load responses using the baseline model and compute its average future electricity load changes across the temperature scenarios from 21 GCMs under two climate RCPs (RCP4.5 and RCP8.5). For each city and scenario, we first compute the annual profile of average daily temperatures in a historical period (2001–2005) and two forecast periods (2051–2055 and 2091–2095). We then calculate the percentage change in electricity consumption of city \(i\) in \(t\) th day of the year as: $$\frac{{\Delta }E{C}_{it}}{E{C}_{it}}=\frac{\text{exp}\left(\sum {\beta }_{j}{f}_{j}\left({T}_{it}^{{\prime }}\right)\right)}{\text{exp}\left(\sum {\beta }_{j}{f}_{j}\left({T}_{it}\right)\right)}-1=\text{exp}\left(\sum {\beta }_{j}\left({f}_{j}\left({T}_{it}^{{\prime }}\right)-{f}_{j}\left({T}_{it}\right)\right)\right)-1,$$ where \({T}_{it}^{{\prime }}\) is from the future temperature profile and \({T}_{it}\) is from the historical temperature profile. The city’s electricity consumption in day \(t\) takes various weight ( \({w}_{it}=\frac{E{C}_{it}}{{\sum }_{t=1}^{365}E{C}_{it}}\) ) in its annual load. We apply the consumption weights (computed using 2019 data) when aggregating the percentage change in \(E{C}_{it}\) and calculate the percentage change in the city’s annual electricity consumption. The 92 cities in our study account for 37.3% of the national population and 46.7% of the gross domestic product (GDP) in 2019. We collected historical shares of tertiary sector in GDP 4 (2003–2019) from China City Statistical Yearbook (National Bureau of Statistics of China (NBSC), 2022 ). We use the average annual growth rate of the tertiary sector to reflect the city’s tertiarization trend, and extrapolate this trend to predict future tertiary GDP shares (in years 2024, 2029, 2034, 2039). Cities are grouped into three tiers based on their tertiary GDP share in 2019: low (below 43.8%), middle (43.8–53.0%), and high (above 53.0%) 5 . A city is assumed to move to a higher tier if its predicted GDP share in some future year exceeds the threshold of that tier. The projected city classifications based on the tertiarization trend are shown in SI Appendix , Fig. S2. We estimate the temperature-load responses function for cities in each tertiarization rank tier in 2019 respectively. The results show that (all-sector) electricity consumptions of cities with higher tertiarization level are more sensitive to temperature variations ( SI Appendix , Table S5). We then apply these estimates according to the predicted future tertiarization level of each city, assuming when cities move to a higher tertiarization tier, their electricity consumptions will be more sensitive to temperature variations. Following the same projection procedure described above, we calculate the percentage change in the city’s annual electricity consumption under two climate RCPs. 3. Results 3.1 Nonlinear temperature-load responsiveness Figure 1 represents the average responses across all cities, i.e. a U-shape temperature response function 6 . We find this U-shape response function to be asymmetric. The function is relatively flat over 14.1–20.2°C (57–68°F), implying minimal needs for heating or cooling. Between 20°C and 26°C, increasing temperature induces a positive and significant load response. This load response is more abrupt when temperature rises beyond 25.6°C (78°F). Table 1 Percentage change in daily electricity load in response to 1 °C temperature increase. Temperature Range Load Change 25.6 °C 2.12% We calculate the slope in each spline segment as the estimated percentage change in daily electricity load in response to 1°C temperature increase (Table 1). Contrast to the intensity of responses in the high temperature range, the electricity consumption is less sensitive when daily mean temperature drops 25.6°C would lead to 2.12% increase in daily electricity consumption, while a 1°C decrease for temperature < 7°C would lead to 0.96% increase in electricity consumption. This finding implies that hot days will likely require more generation capacity and/or energy storage to cope with abrupt load changes in response to extreme temperature events. We compare our results with Li et al. ( 2019 ) which used temperature splines to estimate the response function for residential electricity consumptions in the Yangtze River Delta using 2015–2016 data. We find our estimates present a similarly U-shaped curve but with flatter slopes. For example, marginal load change corresponding to 1°C temperature increase is 2.12% (> 25.6°C) in our result versus 14.5% (> 25°C) in the previous study. Applying our analysis respectively to the cities’ residential loads (rural and urban), we find compared with the average responses (Fig. 1), the residential sector’s daily electricity consumption is much more sensitive to temperatures, and the urban residential load responds more intensively compared with rural in the low and high temperature ranges ( SI Appendix , Fig. S4). However, the residential sector only accounts for approximately 16% of the total electricity consumption. Hence extrapolating implications without discriminating the sectoral heterogeneity will likely overestimate the climate impact on future electricity demand. 3.2 Seasonal discrepancy associated with regional heating availability We also examine the daily peak electricity load, which tends to drive the requisite peak generation capacity investment and peak management strategies (Auffhammer et al., 2017 ; Khan et al., 2021 ). We apply our baseline model on electricity consumptions — daily peak and aggregated daily sum — of 11 cities for the two-year period, and compare the temperature-load response functions (see SI Appendix , Table S4 and Fig. S3). Different from Auffhammer et al ( 2017 ) which found the peak load responses are substantially stronger than that of average load to temperature increases, we do not find significant differences in hot days (> 25°C), but relatively stronger peak responses in cold days (< 8°C). Though the robustness of this result could be affected by data limitations, our finding implies that mechanisms driving the heating and cooling electricity demand might vary in different regions. We make some speculations to explain this seasonal discrepancy finding. Smart thermostats are not widely adopted in China. Changing the HVAC settings usually require behavioral actions. In hot days, the industries and businesses might have already maximized their cooling capacity, thus overcooling the space and driving up the average electricity consumptions. In cold days, alternative heating devices (e.g. district heating with hot water, non-electric stoves) are available. If electric heating were only used to meet the residual demand, i.e., to cope with extreme cold, then peak electricity consumption will be significantly affected more than the average daily consumption. To examine the speculation that heating mechanism variation might attribute to the seasonal disparity in temperature responses, we estimate separate response functions for two groups of cities: north with district heating and south without (or limited) district heating. Figure 2 confirms that the South cities, where non-electric district heating is not widely available, respond much stronger to cold temperatures than the North cities. The hot-day (daily mean temperature > 20°C) electricity load responses, on the other hand, do not demonstrate significant differences between the two regional groups. 3.3 Temperature sensitivity is higher for the tertiary sector In addition to estimating the all-sector total electricity load responses, we explore the extent to which tertiarization drives these responses by examining the sectoral heterogeneity and predicting future electricity loads under hypothetical tertiarization scenarios. We estimate the panel regression model with temperature splines using city-level electricity consumption data aggregated by economic sectors. Contrast to the primary and secondary sectors, which are agriculture, raw material extraction and manufacturing, the tertiary sector provides services to other businesses and to consumers directly. Our study contains 15 tertiary sectors (a complete list is provided in SI Appendix , Table S3) which account for 29.71% electricity consumption in our sample for year 2019. Our finding demonstrates sectoral heterogeneity in how electricity demand varies with temperature. The marginal load responses are statistically significant for all sectors in both high (> 25.6°C) and low (< 6.6°C) temperature ranges. However, we find the tertiary electricity loads 7 more sensitive to temperature variations (Fig. 3 and SI Appendix , Fig. S5) than the non-tertiary loads. This is because electricity consumptions in the primary and secondary sectors are majorly driven by production activities rather than heating and cooling demands. For example, the agriculture and food production exhibits obvious seasonal patterns. We find the agricultual loads on hot days (> 20°C) are far more sensitive to temperature variations than on cooler days. This is likely driven by the intensified use of electric pumps for irrigation when temperature is high. 3.4 Load projection under climate change Load projection exhibits regional and seasonal variations. We apply the estimated temperature-load response function to project changes in daily electricity consumption of each city in our sample under two Representative Concentration Pathways: RCP4.5 and RCP8.5. The RCP8.5 pathway represents the climate scenario with little mitigation effort and a failure to control warming by 2100. The RCP4.5 pathway corresponds to a low-emission pathway with modest climate change. We find regional and seasonal variations in the predicted daily load changes (Fig. 4 ). The hot-day electricity consumption is predicted to increase more significantly and with greater variations, especially under RCP8.5 (comparing two columns in Fig. 4 a). Regarding the change in load from the mid-term (2051–2055) to the more long-term future (2091–2095), we find a similar pattern for the two climate RCPs. In the long run, some cities might experience moderate declines in daily electricity consumptions, which mostly occur in the winter season, implying reduced heating demand due to a warmer climate (Fig. 4 b upper panel and SI Appendix , Fig. S6). Most cities, however, will have pronounced increases in the summer peak load (Fig. 4 b lower panel and SI Appendix , Fig. S6). Coping with the new patterns in future electricity demand requires expanding the generation and transmission capacities, which might result in increasing electricity prices. Load projection adapts to the tertiarization trend. Cities with higher tertiary sector GDP shares are more sensitive to temperature variations ( SI Appendix , Table S5). We therefore predict the tertiarization trend of the cities according to their historical growth rate (2003–2019) of the tertiary sector GDPs. We then examine how the confluence of tertiarization and climate change affects load projections. We find in the near-term future of RCP8.5 almost all the cities in our sample would experience some positive changes in the summer load, while having a reduced winter load due to warming (Fig. 5 ). This confirms the finding that summer electricity consumptions will be the primary driver of the future annual peak load. Provided a majority of the cities (over 66%) would have significant increase in their summer peak load, the requisite grid capacity expansion and energy storage investment would have to occur much sooner than predictions made in absence of the tertiarization trend. Given the tertiary sector consumes less electricity overall compared to the primary and secondary sectors, we questioned whether tertiarization could lead to a net negative change in aggregate electricity consumption, i.e., the energy intensity impact overshadows the climate impact. Examining the historical GDPs as well as the productivity of the three economic sectors, we find cities in our sample had experienced growth in the tertiary sector without crowding out productions in other sectors significantly. This justifies the assumption in our approach that tertiarization would increase load sensitivity to temperature rather than shrink the aggregated load. We note that this approach risks missing potential impacts on other extensive margins, such as energy efficiency improvements. Another caveat is that the tertiarization trend might not proceed linearly as we predicted. Therefore, our result could represent an upper-bound estimate of the climate-adaptive electricity consumptions, which aims to demonstrate the challenges in annual peak load management under climate change even with a good understanding about the intensive effects. 4. Conclusion This study adds upon the growing literature of quantifying the long-term climate impact on future electricity demand. Benefiting from the high-frequency electricity consumption data that covers a wide geo-scale and multiple economic sectors, our findings can speak to both the intensive margin as well as extensive margin adaptation to climate change. To estimate the intensive margin, an asymmetric U-shaped load-temperature response function is identified, while holding the capital stock, technology, and socioeconomic factors constant. Distinguishing from most previous studies, which focused on the magnitude of responses, we further examine the potential mechanism that drives the asymmetric shape in terms of regional disparities in heating infrasture. Following this effort, we recommend implications extrapolated for the winter heating load be treated with caution. District heating, especially in urban areas of northern China, is supplied by Combined Heat and Power (CHP) plants, most of which are coal-fired and are expected to retire as China pushes its carbon neutrality plan (Cui et al., 2021 ). This creates uncertainty in forecasting winter electricity demand as the district heating infrastructure evolves from CHP-dominated to a diverse profile of electrification and zero-carbon heating options (e.g. heat pumps). Future studies will need to provide empirical evidence as well as forecast on the timing and efficiency of heating infrastructure transforms across regions. This study contributes to quantify the extensive margin of climate impact on electricity consumption via changes in economic structure. Our finding demonstrates the heterogeneity between industrial sectors in how electricity demand varies with temperature, while highlighting the needs to better understand mechanisms driving such heterogeneity. China, and many other developing economies, are experiencing a rapid tertiarization process, with expanding shares of employment and economic growth driven by the production of services (CEPR, 2022). Our finding about the load-temperature sensitivity in the tertiary sector implies greater variance of the aggregate electricity load demand as economic activities shift to producing and consuming services. Our simulation shows that if cities continue their current trend of terterisation then intensive increases in summer electricity demand are predicted to occur within the next decade. This would require expanding capacity investments in flexible generations (e.g. peaker plants) and grid-scale energy storage, imposing challenges and opportunities to renewable energy development. While our work adds noval findings to the climate adapatation linearture on energy demand, we recognize several limitations of this study which are left to future studies to resolve. First, our data only contains electricity consumptions in urban cities (and aggregated rural residential loads). Rural areas are likely to differ in their industrial layout, building charateristics, socio-demographic factors, and the electrification rates, all of which can reshape the temperature-load responses (intensive margin) as well as long-term climate implications (extensive margin). Second, our study assumes stability in population distribution across regions, and finds electricity demand reductions in some cities experiencing warming climate, i.e., climate zones shift north. Existing evidence suggests occurance of climate-induced migration over time in China, which is found historically due to displaced agricultural activities (Gray et al., 2020 ). Our results could be interpreted as a lower bound if future climte change drives more people to move to cooler regions. Finally, the climate impact on electricity demand may not linearly convert to damages, which are associated with population distribution, adaptation strategies, and losses from extreme events etc. Future efforts in quantifying the confluence of these factors to climate damages are critical for informing mitigation and adaptation policies. Declarations Ethics approval and consent to participate This study uses secondary data that is de-identified and aggregated to city level. There is no human and animal participants to this study. Consent for publication All authors are consent to publish this manuscript. Author contributions All authors conceived and designed research. Data preparation was performed by Xuebin Wang. Data analysis and visualization were performed by Hanyi Chen and Qingran Li. The first draft of the manuscript was written by Qingran Li, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Competing Interests The authors have no relevant financial or non-financial interests to disclose. Funding There is no funding source to disclose. Data Availability Records of weather variables were obtained from Global Surface Summary of the Day (GSOD) database accessible via National Centers for Environmental Information: https://www.ncei.noaa.gov/. The temperature data under climate Representative Concentration Pathways was from the NASA Earth Exchange Global Daily Downscaled Projections, accessible through: https://www.nccs.nasa.gov/services/data-collections/land-based-products/nex-gddp. The city-level tertiary sector GDP share came from China City Statistical Yearbook, which were provided in the SI Appendix. The daily electricity consumption data is from the China Electric Power Research Institute (CEPRI). Due to protection of privacy, they can be made available upon reasonable request and with permission from the CEPRI. References Auffhammer, M. (2022). Climate Adaptive Response Estimation: Short and long run impacts of climate change on residential electricity and natural gas consumption. Journal of Environmental Economics and Management , 114 , 102669. https://doi.org/10.1016/j.jeem.2022.102669 Auffhammer, M., Baylis, P., & Hausman, C. H. (2017). Climate change is projected to have severe impacts on the frequency and intensity of peak electricity demand across the United States. Proceedings of the National Academy of Sciences , 114 (8), 1886–1891. https://doi.org/10.1073/pnas.1613193114 Auffhammer, M., & Mansur, E. T. (2014). Measuring climatic impacts on energy consumption: A review of the empirical literature. Energy Economics , 46 , 522–530. https://doi.org/10.1016/j.eneco.2014.04.017 CEPR. (2022, October 24). Tertiarisation like China . CEPR. https://cepr.org/voxeu/columns/tertiarisation-china Chen, H., Yan, H., Gong, K., & Yuan, X.-C. (2021). How will climate change affect the peak electricity load? Evidence from China. Journal of Cleaner Production , 322 , 129080. https://doi.org/10.1016/j.jclepro.2021.129080 Cronin, J., Anandarajah, G., & Dessens, O. (2018). Climate change impacts on the energy system: A review of trends and gaps. Climatic Change , 151 (2), 79–93. https://doi.org/10.1007/s10584-018-2265-4 Cui, R. Y., Hultman, N., Cui, D., McJeon, H., Yu, S., Edwards, M. R., Sen, A., Song, K., Bowman, C., Clarke, L., & others. (2021). A plant-by-plant strategy for high-ambition coal power phaseout in China. Nature Communications , 12 (1), 1468. Davis, L. W., & Gertler, P. J. (2015). Contribution of air conditioning adoption to future energy use under global warming. Proceedings of the National Academy of Sciences , 112 (19), 5962–5967. https://doi.org/10.1073/pnas.1423558112 De Cian, E., & Sue Wing, I. (2019). Global Energy Consumption in a Warming Climate. Environmental and Resource Economics , 72 (2), 365–410. https://doi.org/10.1007/s10640-017-0198-4 Fan, J.-L., Hu, J.-W., & Zhang, X. (2019). Impacts of climate change on electricity demand in China: An empirical estimation based on panel data. Energy , 170 , 880–888. https://doi.org/10.1016/j.energy.2018.12.044 Garrido-Perez, J. M., Barriopedro, D., García-Herrera, R., & Ordóñez, C. (2021). Impact of climate change on Spanish electricity demand. Climatic Change , 165 (3), 50. https://doi.org/10.1007/s10584-021-03086-0 Ghanem, D., & Smith, A. (2021). What Are the Benefits of High-Frequency Data for Fixed Effects Panel Models? Journal of the Association of Environmental and Resource Economists , 8 (2), 199–234. https://doi.org/10.1086/710968 Gray, C., Hopping, D., & Mueller, V. (2020). The changing climate-migration relationship in China, 1989–2011. Climatic Change , 160 (1), 103–122. https://doi.org/10.1007/s10584-020-02657-x Gupta, E. (2016). The effect of development on the climate sensitivity of electricity demand in india. Climate Change Economics , 07 (02), 1650003. https://doi.org/10.1142/S2010007816500032 IEA. (2021). Electricity Information: Overview . https://www.iea.org/reports/electricity-information-overview/electricity-consumption IEA. (2022). IEA Statistics: Countries & Regions—China . https://www.iea.org/countries/china Khan, Z., Iyer, G., Patel, P., Kim, S., Hejazi, M., Burleyson, C., & Wise, M. (2021). Impacts of long-term temperature change and variability on electricity investments. Nature Communications , 12 (1), 1643. https://doi.org/10.1038/s41467-021-21785-1 Li, Y., Pizer, W. A., & Wu, L. (2019). Climate change and residential electricity consumption in the Yangtze River Delta, China. Proceedings of the National Academy of Sciences , 116 (2), 472–477. https://doi.org/10.1073/pnas.1804667115 Meinshausen, M., Lewis, J., McGlade, C., Gütschow, J., Nicholls, Z., Burdon, R., Cozzi, L., & Hackmann, B. (2022). Realization of Paris Agreement pledges may limit warming just below 2 °C. Nature , 604 (7905), 7905. https://doi.org/10.1038/s41586-022-04553-z NASA. (n.d.). NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) . NASA Center for Climate Simulation. Retrieved March 3, 2023, from https://www.nccs.nasa.gov/services/data-collections/land-based-products/nex-gddp National Bureau of Statistics of China (NBSC). (2022). Part II: Statistical Data of Cities at Prefecture Level and Above. In China City Statistical Yearbook 2021 (1st edition). China Statistics Press. https://www.chinayearbooks.com/china-city-statistical-yearbook-2021.html Nayyar, G., Hallward-Driemeier, M., & Davies, E. (2021). At Your Service?: The Promise of Services-Led Development . World Bank Publications. NOAA. (n.d.). National Centers for Environmental Information (NCEI) . Retrieved March 3, 2023, from https://www.ncei.noaa.gov/ Pachauri, S., Pelz, S., Bertram, C., Kreibiehl, S., Rao, N. D., Sokona, Y., & Riahi, K. (2022). Fairness considerations in global mitigation investments. Science , 378 (6624), 1057–1059. https://doi.org/10.1126/science.adf0067 Perera, A. T. D., Nik, V. M., Chen, D., Scartezzini, J.-L., & Hong, T. (2020). Quantifying the impacts of climate change and extreme climate events on energy systems. Nature Energy , 5 (2), 2. https://doi.org/10.1038/s41560-020-0558-0 Propato, T. S., de Abelleyra, D., Semmartin, M., & Verón, S. R. (2021). Differential sensitivities of electricity consumption to global warming across regions of Argentina. Climatic Change , 166 (1), 25. https://doi.org/10.1007/s10584-021-03129-6 Teng, M., Liao, H., Burke, P. J., Chen, T., & Zhang, C. (2022). Adaptive responses: The effects of temperature levels on residential electricity use in China. Climatic Change , 172 (3), 32. https://doi.org/10.1007/s10584-022-03374-3 van Ruijven, B. J., De Cian, E., & Sue Wing, I. (2019). Amplification of future energy demand growth due to climate change. Nature Communications , 10 (1), 2762. https://doi.org/10.1038/s41467-019-10399-3 Wenz, L., Levermann, A., & Auffhammer, M. (2017). North–south polarization of European electricity consumption under future warming. Proceedings of the National Academy of Sciences , 114 (38), E7910–E7918. https://doi.org/10.1073/pnas.1704339114 Zhang, C., Liao, H., & Mi, Z. (2019). Climate impacts: Temperature and electricity consumption. Natural Hazards , 99 (3), 1259–1275. https://doi.org/10.1007/s11069-019-03653-w Zhang, S., Guo, Q., Smyth, R., & Yao, Y. (2022). Extreme temperatures and residential electricity consumption: Evidence from Chinese households. Energy Economics , 107 , 105890. https://doi.org/10.1016/j.eneco.2022.105890 Footnotes EC is the all-sector electricity consumption at the city level for the average effect. We also estimated the same model with EC being the city’s sectoral electricity consumption to examine heterogeneity across sectors. As temperature could vary over time and space over a larger scale, we also estimated the model with standard errors clustered at the province-month level. The results are consistent. Fifteen cities fall into the North-heating group: Beijing, Tianjin, Xi’an, Baoji, Xianyang, Weinan, Yan’an, Ankang, Lanzhou, Tianshui, Wuwei, Zhangye, Pingliang, Dixi, and Pangnan. See SI Appendix, Table S1 for cities’ tertiary GDP share in 2019. The classification thresholds are the 25th and 75th percentiles of the data. The coefficients of the temperature spline are found to be statistically significant (SI Appendix, Table S4 column 1). Higher sensitivity to temperature does not necessarily lead to significant changes in total load, as the tertiary sectors usually consume less electricity than the non-tertiary sectors. For our load project, we used the city-wide estimate grouped by tertiarization tiers and applied consumption weight when computing changes in total electricity consumption. Supplementary Files SIAppendix.pdf Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3944484","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":275073435,"identity":"55051505-7ef8-4a21-bb32-ba16c723f623","order_by":0,"name":"Hanyi Chen","email":"","orcid":"","institution":"Dongbei University of Finance and Economics","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Hanyi","middleName":"","lastName":"Chen","suffix":""},{"id":275073436,"identity":"328093ba-8e36-4a4e-8180-1fa4c0a82ac6","order_by":1,"name":"Qingran Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAvUlEQVRIiWNgGAWjYDACdsbGBwwFDAwSEC4zEVqYGZsNGAxI08LAJkGaFr7DzG0VPwzs8iX7Dx97wFBhndhASIvkYca2mz0GyZazJdLSDRjOpBPWYgDUcpvBgNlAToLHTIKx7TBxWooZDOoN5PjPALX8I1ILM5A0kGbIAWppIEIL0C/Nkj0Gxw0kZ6SlSSQcSzcmqIXvePvDDz8qqg0kzh8+JvGhxlqWoBaGA8icBILKMbSMglEwCkbBKMAGABnLN78PoyLLAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0001-5506-3066","institution":"Clarkson University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Qingran","middleName":"","lastName":"Li","suffix":""},{"id":275073437,"identity":"afb21c6d-735e-4544-a268-7a425c21b671","order_by":2,"name":"Xuebin Wang","email":"","orcid":"","institution":"Shanghai University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xuebin","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2024-02-10 00:08:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3944484/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3944484/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":51818328,"identity":"ff486463-390e-4b3c-93f4-42e44f89fe4d","added_by":"auto","created_at":"2024-02-29 15:31:06","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":314837,"visible":true,"origin":"","legend":"\u003cp\u003eResponsiveness of the city-level daily electricity consumption to daily mean temperature.\u003cstrong\u003e \u003c/strong\u003eThe light blue histogram illustrates the temperature distribution. The temperature spline knots are 6.6, 14.1, 20.2, and 25.6 °C. The y axis plots the natural log of electricity consumption with the lowest value normalized to zero. The pink line depicts the response function derived using the estimated spline coefficients in our baseline regression model. The gray shade represents the 95% confidence interval, using standard errors clustered at the city-week level.\u003c/p\u003e","description":"","filename":"floatimage1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/6e5269e480e1c85d3ed499f5.jpg"},{"id":51818327,"identity":"282dced4-c19f-434c-b99c-b7dc18b4e1d3","added_by":"auto","created_at":"2024-02-29 15:31:06","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":352033,"visible":true,"origin":"","legend":"\u003cp\u003eThe north-south discrepancy in daily load responses to daily mean temperature. The y axis plots the natural log of electricity consumption with the lowest value normalized to zero. The pink line is the response function estimated using data for the “South” cities. The blue line is the response function estimated using data for the “North” cities. The temperature spline knots are the same as in Fig. 1. The gray shade represents the 95% confidence interval, using standard errors clustered at the city-week level.\u003c/p\u003e","description":"","filename":"floatimage2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/ebfabd0b537f841065524e1d.jpg"},{"id":51818329,"identity":"65db0f6b-24eb-457a-b5fc-f86a6e39850e","added_by":"auto","created_at":"2024-02-29 15:31:06","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":91802,"visible":true,"origin":"","legend":"\u003cp\u003eHeterogeneity in temperature-load responses from selected economic sectors. Panel a. (agriculture), panel b. (mining and quarrying) and panel c. (manufacturing) are response functions from three non-tertiary sectors. Panel d. (wholesale and retail trades), panel e. (accommodation and catering) and penal f. (real estate) are response functions from three tertiary sectors.\u003c/p\u003e","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/c102e0fb7457bf4112af340a.png"},{"id":51818331,"identity":"32ac54d5-4297-456a-a296-940870d99bb0","added_by":"auto","created_at":"2024-02-29 15:31:06","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":543856,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted changes in daily electricity consumption under two climate RCPs. Panel a. depicts the average percentage change in daily load across two 5-year periods (2051-2055 and 2091-2095) comparing to year 2001. The gray lines are predictions for the 92 cities. The orange line is the average across all cities. Panel b. maps the regional variation of city’s load predictions for the 2091-2095 future period in a winter month (December) and a summer month (July). The warm colors illustrate positive load changes. The cool colors illustrate negative load changes.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/89dfac4ed5e2ec0b43c2233d.png"},{"id":51818700,"identity":"b9ef6b73-5125-4984-9f7e-bc9df1fdf39e","added_by":"auto","created_at":"2024-02-29 15:39:06","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":50274,"visible":true,"origin":"","legend":"\u003cp\u003eLoad projection with predicted tertiarization trend under climate RCP8.5. Each bar depicts the predicted percentage of cities experiencing increase (warm colors) or decrease (cool colors) in their future daily electricity loads in year 2029 and 2039 for a winter month (December) and a summer month (July). Projection under climate RCP4.5 is shown in \u003cem\u003eSI Appendix\u003c/em\u003e, Fig. S7.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/af46fe36ffebc5d557d0cdb5.png"},{"id":53400735,"identity":"ff81f29c-1268-4e60-80d7-62f2dbbcc22c","added_by":"auto","created_at":"2024-03-25 14:29:13","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1041148,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/dc33ce6b-6981-418e-a813-c39e9964964a.pdf"},{"id":51818332,"identity":"058cc8e6-70a0-4c8f-84d9-fd18ba06ebe2","added_by":"auto","created_at":"2024-02-29 15:31:07","extension":"pdf","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":4760330,"visible":true,"origin":"","legend":"","description":"","filename":"SIAppendix.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3944484/v1/e4d810941cde85e294ee2d17.pdf"}],"financialInterests":"","formattedTitle":"Electricity consumption and adaptation to climate change: heterogeneity across regions and economic sectors in China","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe COP26 global climate summit ended with a clear consensus that significant and immediate carbon mitigation efforts are required to prevent catastrophic climate change. Future emissions pathway, though started to bend downward due to clean energy policies and technologies, is still not enough to avoid temperature rise. Assuming nations will actually follow through on their climate pledges, the carbon abatement would be far from assured to achieve the 1.5 degree goal (Meinshausen et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The need to mitigate carbon emissions has become more pressing as extreme temperatures driven by climate change will influence economic activities (Cronin et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), which is reflected by surging energy consumptions (Auffhammer et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Perera et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; van Ruijven et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) and increasing capital investments to meet the demand (Khan et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLess clear, however, is the extent to which climate change could exert on energy demand given regional heterogeneity in consumption patterns and the latent mechanisms driving these patterns. More extreme temperatures (hotter summer and/or colder winter) are likely to boost electricity demand for heating and cooling, which depends on the climate scenarios, electrification rates, urbanization degrees, and various socio-economic attributes (Auffhammer, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Garrido-Perez et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Gupta, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Propato et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Teng et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; van Ruijven et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; C. Zhang et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). However, due to data availability and reliability issues, empirical studies in the developing economy context were quite sparse. Most studies use regional aggregates (Chen et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; De Cian \u0026amp; Sue Wing, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Fan et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Propato et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Teng et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; C. Zhang et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) to estimate the temperature-load response functions in lack of data with higher temporal resolutions (daily or weekly load), which tend to aggregate away the high-frequency variations and can potentially bias the load responses (Ghanem \u0026amp; Smith, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). When studies are available, they tend to focus on adaptive responses of residential electricity consumptions (Davis \u0026amp; Gertler, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Teng et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; S. Zhang et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), where responses to extreme weather events are most intense (Li et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). This suggests a knowledge gap in understanding the climate-adaptive responses of other economic sectors (e.g. industry, commercial, agriculture).\u003c/p\u003e \u003cp\u003eClosing this knowledge gap is critical given non-residential electricity consumptions are the major driver of global electricity demand, which is about 73% between 2018 and 2019 with China and United States being the top two electricity consuming countries (IEA, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). If electricity consumptions in these economic sectors are sensitive to temperature variability, this will then increase the vulnerability of the economy to climate extremes, requiring intensive investments to enhance power supply resilience and reliability. Given the total gross electricity production in non-OECD countries surpassed OECD countries after 2010 and continued to grow rapidly (IEA, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), the confluence of this growth and temperature variability is likely to amplify the impact of climate shocks. This would introduce disparities in countries\u0026rsquo; abilities to adapt to climate change and enlarge the gap in requisite investments associated with electrification and mitigation (Pachauri et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In order to design efficient and equitable national and international climate policies, increasing empirical evidence in the developing and/or emerging economy context is needed.\u003c/p\u003e \u003cp\u003eOur work address this issue. We assess the electricity-climate responses of various economic sectors in China \u0026mdash; a fast-growing economy with increasing electricity demand and over 40% produced by coal in 2020 (IEA, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Using a high-frequency daily electricity consumption data set from 92 Chinese cities, our efforts emphasize the spatial heterogeneity (regions and cities) in temperature-load response functions. We subsequently examine the influence of tertiarization \u0026mdash; increasing shares of the tertiary sector in economic outputs \u0026mdash; on electricity demand projections under different climate change scenarios. We extrapolate implications of our findings for the long-term power sector resource planning and for regional and/or sectoral policy designs to adapt to climate change.\u003c/p\u003e \u003cp\u003eOur work builds upon and contributes to the existing literature of climate adaptation in energy consumption. First, this study examines the temperature response heterogeneity across different economic sectors and regions using high-frequency electricity consumption data, whereas previous studies have examined diversity in a limited set (e.g. rural and urban residential) with high-frequency data (Li et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) or have examined geo-diversity (e.g. city/provincial panel) with lower temporal resolutions (Chen et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Fan et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Teng et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; S. Zhang et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and aggregated sectoral consumptions (Propato et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The sectoral-distinction is important. Electricity consumption patterns vary with economic sectors. The temperature elasticities in these sectoral electricity consumptions are associated with the portion which heating and cooling account for in the total end-use consumptions. Modeling load responses to temperature variability, if not discriminating the sectors, is likely to introduce bias to electricity demand projections. This is particularly true for regions where the economy is heavily oriented towards service sectors, as service-sector outputs usually exhibit higher temperature elasticities. Our study accounts for the sectoral-distinction, permitting examinations on the electricity demand robustness to tertiarization, which has become a major force driving economic growth and transformation in middle- and low-income countries (Nayyar et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Our findings therefore add to the literature on extensive margin of climatic impacts (Auffhammer \u0026amp; Mansur, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), by addressing the marginal impact via tertiarization, whereas many past studies examined how energy consumptions respond to short-term weather shocks (the intensive margin).\u003c/p\u003e \u003cp\u003eMoreover, this study contributes to empirical evidences of spatial diversity in climatic impacts. The north versus south distinction is critical as climate change not only lead to increasing global surface temperature, but also alters the patterns of different climate zones, thus resulting to notable shifts in electricity consumption patterns (Wenz et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Our analysis reveals a significant north-south sensitivity difference to temperature, especially in the cold temperature range. This difference can be attributed to the extent to which district heating is available, though mostly powered by fossil fuels in China (e.g. natural gas, residual heat from coal power plants). Hence, modeling the spatial-heterogeneous temperature responses helps to refine predictions of heating energy consumptions, thus informing future energy management and pollution controls.\u003c/p\u003e"},{"header":"2. Methodology","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Data\u003c/h2\u003e \u003cp\u003eOur analyses primarily utilize two data sources: electricity consumption data and the weather data with daily variations. Data on electricity consumption originate from the China Electric Power Research Institute, a research center affiliated to the State Grid Corporation of China. The metered consumption data (in GWhs) are aggregated at the city and sector level. For 92 Chinese cities (see \u003cem\u003eSI Appendix\u003c/em\u003e, Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e for the list of cities), we obtain daily electricity consumption data of 20 non-residential economic sectors and the residential sector. The two-year time series ranges from January 1, 2018 to December 31, 2019. We also attain daily peak load data for 11 of the 92 cities. \u003cem\u003eSI Appendix\u003c/em\u003e, Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e illustrates the variation of daily total electricity load by sector (secondary and tertiary) and the residential consumption load. For the two-year time period (2018\u0026ndash;2019), the temperature and other weather variables are from the Global Surface Summary of Day (GSOD) database accessible via National Centers for Environmental Information (NOAA, n.d.), including daily mean temperature, dew point (to compute relative humidity), precipitation, and wind speed. To integrate this data into the city panel, we match the city with its nearest weather station.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Regression Model\u003c/h2\u003e \u003cp\u003eOur baseline model follows Li et al. (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), which uses splines to model the nonlinear relationship between temperature and electricity consumption. Our econometric model is specified by the log-linear equation below for city \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(i\\)\u003c/span\u003e\u003c/span\u003e and day \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(t\\)\u003c/span\u003e\u003c/span\u003e:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\text{ln}\\left(E{C}_{it}\\right)=\\alpha +\\sum {\\beta }_{j}{f}_{j}\\left(tem{p}_{it}\\right)+\\gamma Weathe{r}_{it}+\\theta nonWorkda{y}_{t}+{\\eta }_{i,m,y}+{\\epsilon }_{it}.$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe dependent variable is the natural logarithm of a city\u0026rsquo;s daily electricity consumption\u003csup\u003e1\u003c/sup\u003e, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(tem{p}_{it}\\)\u003c/span\u003e\u003c/span\u003e is the city\u0026rsquo;s daily mean temperature. To control for non-temperature confounders, the \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(Weathe{r}_{it}\\)\u003c/span\u003e\u003c/span\u003e variables include daily mean precipitation, wind speed, and relative humidity. The dummy variables \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(nonWorkda{y}_{t}\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({\\eta }_{i,m,y}\\)\u003c/span\u003e\u003c/span\u003e control for non-workdays and the city-month-year fixed effects. This is because electricity consumption load tends to be lower in weekends and holidays, and the cities and industries can face different electricity price in same month of the different years. Standard errors are clustered at the city-week level\u003csup\u003e2\u003c/sup\u003e. Summary statistics of these variables are provided in \u003cem\u003eSI Appendix\u003c/em\u003e, Table S2.\u003c/p\u003e \u003cp\u003eThe functions \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({f}_{j}(.)\\)\u003c/span\u003e\u003c/span\u003e are linear splines with a number of defined temperature thresholds or knots, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({k}_{j}\\)\u003c/span\u003e\u003c/span\u003e, which are specified as:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$${f}_{0}\\left(tem{p}_{it}\\right)=tem{p}_{it}, \\text{a}\\text{n}\\text{d} {f}_{j}\\left(tem{p}_{it}\\right)=\\left\\{\\begin{array}{c}tem{p}_{it} if tem{p}_{it}\u0026gt;{k}_{j}\\\\ 0 if tem{p}_{it}\\le {k}_{j}\\end{array}\\right. \\text{f}\\text{o}\\text{r} j\\ge 1.$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThis approach models a smooth response function and permits changing slopes, implying flexible variability of the responses in different temperature ranges. This flexibility increases with the number of knots used to model the spline function. A data-driven algorithm is used to select the level of flexibility for the spline function (\u003cem\u003eSI Appendix\u003c/em\u003e, Model Selection), i.e., the number of equally spaced quantile knots for the linear spline. The result is a choice of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(j=4\\)\u003c/span\u003e\u003c/span\u003e, or four knots.\u003c/p\u003e \u003cp\u003eTo explore heterogeneity by geographic region, we divide the city panel to North-heating and South-non-heating groups\u003csup\u003e3\u003c/sup\u003e. For all cities in the North or South group, we apply the baseline city-panel regression model with linear temperature splines, and compare the regional heterogeneity result from variation in heating infrastructures. To explore heterogeneity by economic sector, we estimate the baseline city-panel regression model for each economic sector. The summary statistics of cities\u0026rsquo; sectoral electricity consumptions are provided in \u003cem\u003eSI Appendix\u003c/em\u003e, Table S3.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Load Projections\u003c/h2\u003e \u003cp\u003eTo develop future temperature portfolios for each city under different climate change scenarios, we obtain temperature data under two climate Representative Concentration Pathways (RCPs) \u0026mdash; RCP4.5 and RCP8.5 \u0026mdash; from the NASA Earth Exchange Global Daily Downscaled Projections (NASA, n.d.). The projections are derived from 21 General Circulation Model (GCM) runs and have been downscaled to a spatial resolution of 0.25 degrees (~\u0026thinsp;25 x 25 km). For each of the 21 GCMs, daily maximum and minimum temperatures are downloaded for historical years 2001\u0026ndash;2005 and two forecasted time periods, 2051\u0026ndash;2055 and 2091\u0026ndash;2095. To match the temperature data with our city panel, for each city, we calculate the average temperature of the four closest coordinates to the centroid of the city.\u003c/p\u003e \u003cp\u003eThe city-wide load projection procedure follows Li et al. (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), and adopts the common assumption in the literature (Auffhammer et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Davis \u0026amp; Gertler, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) that the fixed effects and other non-temperature variables remain the same for the past and future. For each city in our sample, we estimate the city-specific temperature-load responses using the baseline model and compute its average future electricity load changes across the temperature scenarios from 21 GCMs under two climate RCPs (RCP4.5 and RCP8.5). For each city and scenario, we first compute the annual profile of average daily temperatures in a historical period (2001\u0026ndash;2005) and two forecast periods (2051\u0026ndash;2055 and 2091\u0026ndash;2095). We then calculate the percentage change in electricity consumption of city \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(i\\)\u003c/span\u003e\u003c/span\u003e in \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(t\\)\u003c/span\u003e\u003c/span\u003e\u003csup\u003eth\u003c/sup\u003e day of the year as:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\frac{{\\Delta }E{C}_{it}}{E{C}_{it}}=\\frac{\\text{exp}\\left(\\sum {\\beta }_{j}{f}_{j}\\left({T}_{it}^{{\\prime }}\\right)\\right)}{\\text{exp}\\left(\\sum {\\beta }_{j}{f}_{j}\\left({T}_{it}\\right)\\right)}-1=\\text{exp}\\left(\\sum {\\beta }_{j}\\left({f}_{j}\\left({T}_{it}^{{\\prime }}\\right)-{f}_{j}\\left({T}_{it}\\right)\\right)\\right)-1,$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({T}_{it}^{{\\prime }}\\)\u003c/span\u003e\u003c/span\u003e is from the future temperature profile and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({T}_{it}\\)\u003c/span\u003e\u003c/span\u003e is from the historical temperature profile. The city\u0026rsquo;s electricity consumption in day \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(t\\)\u003c/span\u003e\u003c/span\u003e takes various weight (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({w}_{it}=\\frac{E{C}_{it}}{{\\sum }_{t=1}^{365}E{C}_{it}}\\)\u003c/span\u003e\u003c/span\u003e) in its annual load. We apply the consumption weights (computed using 2019 data) when aggregating the percentage change in \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(E{C}_{it}\\)\u003c/span\u003e\u003c/span\u003e and calculate the percentage change in the city\u0026rsquo;s annual electricity consumption.\u003c/p\u003e \u003cp\u003eThe 92 cities in our study account for 37.3% of the national population and 46.7% of the gross domestic product (GDP) in 2019. We collected historical shares of tertiary sector in GDP\u003csup\u003e4\u003c/sup\u003e (2003\u0026ndash;2019) from China City Statistical Yearbook (National Bureau of Statistics of China (NBSC), \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). We use the average annual growth rate of the tertiary sector to reflect the city\u0026rsquo;s tertiarization trend, and extrapolate this trend to predict future tertiary GDP shares (in years 2024, 2029, 2034, 2039). Cities are grouped into three tiers based on their tertiary GDP share in 2019: low (below 43.8%), middle (43.8\u0026ndash;53.0%), and high (above 53.0%)\u003csup\u003e5\u003c/sup\u003e. A city is assumed to move to a higher tier if its predicted GDP share in some future year exceeds the threshold of that tier. The projected city classifications based on the tertiarization trend are shown in \u003cem\u003eSI Appendix\u003c/em\u003e, Fig. S2.\u003c/p\u003e \u003cp\u003eWe estimate the temperature-load responses function for cities in each tertiarization rank tier in 2019 respectively. The results show that (all-sector) electricity consumptions of cities with higher tertiarization level are more sensitive to temperature variations (\u003cem\u003eSI Appendix\u003c/em\u003e, Table S5). We then apply these estimates according to the predicted future tertiarization level of each city, assuming when cities move to a higher tertiarization tier, their electricity consumptions will be more sensitive to temperature variations. Following the same projection procedure described above, we calculate the percentage change in the city\u0026rsquo;s annual electricity consumption under two climate RCPs.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n\u003ch2\u003e3.1 Nonlinear temperature-load responsiveness\u003c/h2\u003e\n\u003cp\u003eFigure\u0026nbsp;1 represents the average responses across all cities, i.e. a U-shape temperature response function\u003csup\u003e6\u003c/sup\u003e. We find this U-shape response function to be asymmetric. The function is relatively flat over 14.1\u0026ndash;20.2\u0026deg;C (57\u0026ndash;68\u0026deg;F), implying minimal needs for heating or cooling. Between 20\u0026deg;C and 26\u0026deg;C, increasing temperature induces a positive and significant load response. This load response is more abrupt when temperature rises beyond 25.6\u0026deg;C (78\u0026deg;F).\u003c/p\u003e\n\u003ctable border=\"1\"\u003e\u003ccaption\u003e\n\u003cp\u003eTable 1\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePercentage change in daily electricity load in response to 1 \u0026deg;C temperature increase.\u003c/p\u003e\n\u003c/caption\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e\u003cstrong\u003eTemperature Range\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"68\"\u003e\n\u003cp\u003e\u003cstrong\u003eLoad Change\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e\u0026lt; 6.6 \u0026deg;C\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"68\"\u003e\n\u003cp\u003e-0.96%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e6.6-14.1 \u0026deg;C\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"68\"\u003e\n\u003cp\u003e-0.92%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e14.1-20.2 \u0026deg;C\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"68\"\u003e\n\u003cp\u003e-0.07%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e20.2-25.6 \u0026deg;C\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"68\"\u003e\n\u003cp\u003e1.01%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd width=\"122\"\u003e\n\u003cp\u003e\u0026gt; 25.6 \u0026deg;C\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd width=\"68\"\u003e\n\u003cp\u003e2.12%\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eWe calculate the slope in each spline segment as the estimated percentage change in daily electricity load in response to 1\u0026deg;C temperature increase (Table\u0026nbsp;1). Contrast to the intensity of responses in the high temperature range, the electricity consumption is less sensitive when daily mean temperature drops\u0026thinsp;\u0026lt;\u0026thinsp;14.1\u0026deg;C. A 1\u0026deg;C increase for temperature\u0026thinsp;\u0026gt;\u0026thinsp;25.6\u0026deg;C would lead to 2.12% increase in daily electricity consumption, while a 1\u0026deg;C decrease for temperature\u0026thinsp;\u0026lt;\u0026thinsp;7\u0026deg;C would lead to 0.96% increase in electricity consumption. This finding implies that hot days will likely require more generation capacity and/or energy storage to cope with abrupt load changes in response to extreme temperature events.\u003c/p\u003e\n\u003cp\u003eWe compare our results with Li et al. (\u003cspan class=\"CitationRef\"\u003e2019\u003c/span\u003e) which used temperature splines to estimate the response function for residential electricity consumptions in the Yangtze River Delta using 2015\u0026ndash;2016 data. We find our estimates present a similarly U-shaped curve but with flatter slopes. For example, marginal load change corresponding to 1\u0026deg;C temperature increase is 2.12% (\u0026gt;\u0026thinsp;25.6\u0026deg;C) in our result versus 14.5% (\u0026gt;\u0026thinsp;25\u0026deg;C) in the previous study. Applying our analysis respectively to the cities\u0026rsquo; residential loads (rural and urban), we find compared with the average responses (Fig.\u0026nbsp;1), the residential sector\u0026rsquo;s daily electricity consumption is much more sensitive to temperatures, and the urban residential load responds more intensively compared with rural in the low and high temperature ranges (\u003cem\u003eSI Appendix\u003c/em\u003e, Fig. S4). However, the residential sector only accounts for approximately 16% of the total electricity consumption. Hence extrapolating implications without discriminating the sectoral heterogeneity will likely overestimate the climate impact on future electricity demand.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n\u003ch2\u003e3.2 Seasonal discrepancy associated with regional heating availability\u003c/h2\u003e\n\u003cp\u003eWe also examine the daily peak electricity load, which tends to drive the requisite peak generation capacity investment and peak management strategies (Auffhammer et al., \u003cspan class=\"CitationRef\"\u003e2017\u003c/span\u003e; Khan et al., \u003cspan class=\"CitationRef\"\u003e2021\u003c/span\u003e). We apply our baseline model on electricity consumptions \u0026mdash; daily peak and aggregated daily sum \u0026mdash; of 11 cities for the two-year period, and compare the temperature-load response functions (see \u003cem\u003eSI Appendix\u003c/em\u003e, Table S4 and Fig. S3). Different from Auffhammer et al (\u003cspan class=\"CitationRef\"\u003e2017\u003c/span\u003e) which found the peak load responses are substantially stronger than that of average load to temperature increases, we do not find significant differences in hot days (\u0026gt;\u0026thinsp;25\u0026deg;C), but relatively stronger peak responses in cold days (\u0026lt;\u0026thinsp;8\u0026deg;C). Though the robustness of this result could be affected by data limitations, our finding implies that mechanisms driving the heating and cooling electricity demand might vary in different regions.\u003c/p\u003e\n\u003cp\u003eWe make some speculations to explain this seasonal discrepancy finding. Smart thermostats are not widely adopted in China. Changing the HVAC settings usually require behavioral actions. In hot days, the industries and businesses might have already maximized their cooling capacity, thus overcooling the space and driving up the average electricity consumptions. In cold days, alternative heating devices (e.g. district heating with hot water, non-electric stoves) are available. If electric heating were only used to meet the residual demand, i.e., to cope with extreme cold, then peak electricity consumption will be significantly affected more than the average daily consumption.\u003c/p\u003e\n\u003cp\u003eTo examine the speculation that heating mechanism variation might attribute to the seasonal disparity in temperature responses, we estimate separate response functions for two groups of cities: north with district heating and south without (or limited) district heating. Figure\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e confirms that the \u003cem\u003eSouth\u003c/em\u003e cities, where non-electric district heating is not widely available, respond much stronger to cold temperatures than the \u003cem\u003eNorth\u003c/em\u003e cities. The hot-day (daily mean temperature\u0026thinsp;\u0026gt;\u0026thinsp;20\u0026deg;C) electricity load responses, on the other hand, do not demonstrate significant differences between the two regional groups.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n\u003ch2\u003e3.3 Temperature sensitivity is higher for the tertiary sector\u003c/h2\u003e\n\u003cp\u003eIn addition to estimating the all-sector total electricity load responses, we explore the extent to which tertiarization drives these responses by examining the sectoral heterogeneity and predicting future electricity loads under hypothetical tertiarization scenarios. We estimate the panel regression model with temperature splines using city-level electricity consumption data aggregated by economic sectors. Contrast to the primary and secondary sectors, which are agriculture, raw material extraction and manufacturing, the tertiary sector provides services to other businesses and to consumers directly. Our study contains 15 tertiary sectors (a complete list is provided in \u003cem\u003eSI Appendix\u003c/em\u003e, Table S3) which account for 29.71% electricity consumption in our sample for year 2019.\u003c/p\u003e\n\u003cp\u003eOur finding demonstrates sectoral heterogeneity in how electricity demand varies with temperature. The marginal load responses are statistically significant for all sectors in both high (\u0026gt;\u0026thinsp;25.6\u0026deg;C) and low (\u0026lt;\u0026thinsp;6.6\u0026deg;C) temperature ranges. However, we find the tertiary electricity loads\u003csup\u003e7\u003c/sup\u003e more sensitive to temperature variations (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e and \u003cem\u003eSI Appendix\u003c/em\u003e, Fig. S5) than the non-tertiary loads. This is because electricity consumptions in the primary and secondary sectors are majorly driven by production activities rather than heating and cooling demands. For example, the agriculture and food production exhibits obvious seasonal patterns. We find the agricultual loads on hot days (\u0026gt;\u0026thinsp;20\u0026deg;C) are far more sensitive to temperature variations than on cooler days. This is likely driven by the intensified use of electric pumps for irrigation when temperature is high.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n\u003ch2\u003e3.4 Load projection under climate change\u003c/h2\u003e\n\u003cp\u003e\u003cstrong\u003eLoad projection exhibits regional and seasonal variations.\u003c/strong\u003e We apply the estimated temperature-load response function to project changes in daily electricity consumption of each city in our sample under two Representative Concentration Pathways: RCP4.5 and RCP8.5. The RCP8.5 pathway represents the climate scenario with little mitigation effort and a failure to control warming by 2100. The RCP4.5 pathway corresponds to a low-emission pathway with modest climate change. We find regional and seasonal variations in the predicted daily load changes (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e). The hot-day electricity consumption is predicted to increase more significantly and with greater variations, especially under RCP8.5 (comparing two columns in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003ea).\u003c/p\u003e\n\u003cp\u003eRegarding the change in load from the mid-term (2051\u0026ndash;2055) to the more long-term future (2091\u0026ndash;2095), we find a similar pattern for the two climate RCPs. In the long run, some cities might experience moderate declines in daily electricity consumptions, which mostly occur in the winter season, implying reduced heating demand due to a warmer climate (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eb upper panel and \u003cem\u003eSI Appendix\u003c/em\u003e, Fig. S6). Most cities, however, will have pronounced increases in the summer peak load (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eb lower panel and \u003cem\u003eSI Appendix\u003c/em\u003e, Fig. S6). Coping with the new patterns in future electricity demand requires expanding the generation and transmission capacities, which might result in increasing electricity prices.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLoad projection adapts to the tertiarization trend.\u003c/strong\u003e Cities with higher tertiary sector GDP shares are more sensitive to temperature variations (\u003cem\u003eSI Appendix\u003c/em\u003e, Table S5). We therefore predict the tertiarization trend of the cities according to their historical growth rate (2003\u0026ndash;2019) of the tertiary sector GDPs. We then examine how the confluence of tertiarization and climate change affects load projections. We find in the near-term future of RCP8.5 almost all the cities in our sample would experience some positive changes in the summer load, while having a reduced winter load due to warming (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e). This confirms the finding that summer electricity consumptions will be the primary driver of the future annual peak load. Provided a majority of the cities (over 66%) would have significant increase in their summer peak load, the requisite grid capacity expansion and energy storage investment would have to occur much sooner than predictions made in absence of the tertiarization trend.\u003c/p\u003e\n\u003cp\u003eGiven the tertiary sector consumes less electricity overall compared to the primary and secondary sectors, we questioned whether tertiarization could lead to a net negative change in aggregate electricity consumption, i.e., the energy intensity impact overshadows the climate impact. Examining the historical GDPs as well as the productivity of the three economic sectors, we find cities in our sample had experienced growth in the tertiary sector without crowding out productions in other sectors significantly. This justifies the assumption in our approach that tertiarization would increase load sensitivity to temperature rather than shrink the aggregated load. We note that this approach risks missing potential impacts on other extensive margins, such as energy efficiency improvements. Another caveat is that the tertiarization trend might not proceed linearly as we predicted. Therefore, our result could represent an upper-bound estimate of the climate-adaptive electricity consumptions, which aims to demonstrate the challenges in annual peak load management under climate change even with a good understanding about the intensive effects.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Conclusion","content":"\u003cp\u003eThis study adds upon the growing literature of quantifying the long-term climate impact on future electricity demand. Benefiting from the high-frequency electricity consumption data that covers a wide geo-scale and multiple economic sectors, our findings can speak to both the intensive margin as well as extensive margin adaptation to climate change. To estimate the intensive margin, an asymmetric U-shaped load-temperature response function is identified, while holding the capital stock, technology, and socioeconomic factors constant. Distinguishing from most previous studies, which focused on the magnitude of responses, we further examine the potential mechanism that drives the asymmetric shape in terms of regional disparities in heating infrasture. Following this effort, we recommend implications extrapolated for the winter heating load be treated with caution. District heating, especially in urban areas of northern China, is supplied by Combined Heat and Power (CHP) plants, most of which are coal-fired and are expected to retire as China pushes its carbon neutrality plan (Cui et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). This creates uncertainty in forecasting winter electricity demand as the district heating infrastructure evolves from CHP-dominated to a diverse profile of electrification and zero-carbon heating options (e.g. heat pumps). Future studies will need to provide empirical evidence as well as forecast on the timing and efficiency of heating infrastructure transforms across regions.\u003c/p\u003e \u003cp\u003eThis study contributes to quantify the extensive margin of climate impact on electricity consumption via changes in economic structure. Our finding demonstrates the heterogeneity between industrial sectors in how electricity demand varies with temperature, while highlighting the needs to better understand mechanisms driving such heterogeneity. China, and many other developing economies, are experiencing a rapid tertiarization process, with expanding shares of employment and economic growth driven by the production of services (CEPR, 2022). Our finding about the load-temperature sensitivity in the tertiary sector implies greater variance of the aggregate electricity load demand as economic activities shift to producing and consuming services. Our simulation shows that if cities continue their current trend of terterisation then intensive increases in summer electricity demand are predicted to occur within the next decade. This would require expanding capacity investments in flexible generations (e.g. peaker plants) and grid-scale energy storage, imposing challenges and opportunities to renewable energy development.\u003c/p\u003e \u003cp\u003eWhile our work adds noval findings to the climate adapatation linearture on energy demand, we recognize several limitations of this study which are left to future studies to resolve. First, our data only contains electricity consumptions in urban cities (and aggregated rural residential loads). Rural areas are likely to differ in their industrial layout, building charateristics, socio-demographic factors, and the electrification rates, all of which can reshape the temperature-load responses (intensive margin) as well as long-term climate implications (extensive margin). Second, our study assumes stability in population distribution across regions, and finds electricity demand reductions in some cities experiencing warming climate, i.e., climate zones shift north. Existing evidence suggests occurance of climate-induced migration over time in China, which is found historically due to displaced agricultural activities (Gray et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Our results could be interpreted as a lower bound if future climte change drives more people to move to cooler regions. Finally, the climate impact on electricity demand may not linearly convert to damages, which are associated with population distribution, adaptation strategies, and losses from extreme events etc. Future efforts in quantifying the confluence of these factors to climate damages are critical for informing mitigation and adaptation policies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study uses secondary data that is de-identified and aggregated to city level. There is no human and animal participants to this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors are consent to publish this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors conceived and designed research. Data preparation was performed by Xuebin Wang. Data analysis and visualization were performed by Hanyi Chen and Qingran Li. The first draft of the manuscript was written by Qingran Li, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no relevant financial or non-financial interests to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThere is no funding source to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRecords of weather variables were obtained from Global Surface Summary of the Day (GSOD) database accessible via National Centers for Environmental Information: https://www.ncei.noaa.gov/. The temperature data under climate Representative Concentration Pathways was from the NASA Earth Exchange Global Daily Downscaled Projections, accessible through: https://www.nccs.nasa.gov/services/data-collections/land-based-products/nex-gddp. The city-level tertiary sector GDP share came from China City Statistical Yearbook, which were provided in the SI Appendix. The daily electricity consumption data is from the China Electric Power Research Institute (CEPRI). Due to protection of privacy, they can be made available upon reasonable request and with permission from the CEPRI.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAuffhammer, M. (2022). Climate Adaptive Response Estimation: Short and long run impacts of climate change on residential electricity and natural gas consumption. \u003cem\u003eJournal of Environmental Economics and Management\u003c/em\u003e, \u003cem\u003e114\u003c/em\u003e, 102669. https://doi.org/10.1016/j.jeem.2022.102669\u003c/li\u003e\n\u003cli\u003eAuffhammer, M., Baylis, P., \u0026amp; Hausman, C. H. (2017). Climate change is projected to have severe impacts on the frequency and intensity of peak electricity demand across the United States. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e, \u003cem\u003e114\u003c/em\u003e(8), 1886\u0026ndash;1891. https://doi.org/10.1073/pnas.1613193114\u003c/li\u003e\n\u003cli\u003eAuffhammer, M., \u0026amp; Mansur, E. T. (2014). Measuring climatic impacts on energy consumption: A review of the empirical literature. \u003cem\u003eEnergy Economics\u003c/em\u003e, \u003cem\u003e46\u003c/em\u003e, 522\u0026ndash;530. https://doi.org/10.1016/j.eneco.2014.04.017\u003c/li\u003e\n\u003cli\u003eCEPR. (2022, October 24). \u003cem\u003eTertiarisation like China\u003c/em\u003e. CEPR. https://cepr.org/voxeu/columns/tertiarisation-china\u003c/li\u003e\n\u003cli\u003eChen, H., Yan, H., Gong, K., \u0026amp; Yuan, X.-C. (2021). How will climate change affect the peak electricity load? Evidence from China. \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e, \u003cem\u003e322\u003c/em\u003e, 129080. https://doi.org/10.1016/j.jclepro.2021.129080\u003c/li\u003e\n\u003cli\u003eCronin, J., Anandarajah, G., \u0026amp; Dessens, O. (2018). Climate change impacts on the energy system: A review of trends and gaps. \u003cem\u003eClimatic Change\u003c/em\u003e, \u003cem\u003e151\u003c/em\u003e(2), 79\u0026ndash;93. https://doi.org/10.1007/s10584-018-2265-4\u003c/li\u003e\n\u003cli\u003eCui, R. Y., Hultman, N., Cui, D., McJeon, H., Yu, S., Edwards, M. R., Sen, A., Song, K., Bowman, C., Clarke, L., \u0026amp; others. (2021). A plant-by-plant strategy for high-ambition coal power phaseout in China. \u003cem\u003eNature Communications\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), 1468.\u003c/li\u003e\n\u003cli\u003eDavis, L. W., \u0026amp; Gertler, P. J. (2015). Contribution of air conditioning adoption to future energy use under global warming. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e, \u003cem\u003e112\u003c/em\u003e(19), 5962\u0026ndash;5967. https://doi.org/10.1073/pnas.1423558112\u003c/li\u003e\n\u003cli\u003eDe Cian, E., \u0026amp; Sue Wing, I. (2019). Global Energy Consumption in a Warming Climate. \u003cem\u003eEnvironmental and Resource Economics\u003c/em\u003e, \u003cem\u003e72\u003c/em\u003e(2), 365\u0026ndash;410. https://doi.org/10.1007/s10640-017-0198-4\u003c/li\u003e\n\u003cli\u003eFan, J.-L., Hu, J.-W., \u0026amp; Zhang, X. (2019). Impacts of climate change on electricity demand in China: An empirical estimation based on panel data. \u003cem\u003eEnergy\u003c/em\u003e, \u003cem\u003e170\u003c/em\u003e, 880\u0026ndash;888. https://doi.org/10.1016/j.energy.2018.12.044\u003c/li\u003e\n\u003cli\u003eGarrido-Perez, J. M., Barriopedro, D., Garc\u0026iacute;a-Herrera, R., \u0026amp; Ord\u0026oacute;\u0026ntilde;ez, C. (2021). Impact of climate change on Spanish electricity demand. \u003cem\u003eClimatic Change\u003c/em\u003e, \u003cem\u003e165\u003c/em\u003e(3), 50. https://doi.org/10.1007/s10584-021-03086-0\u003c/li\u003e\n\u003cli\u003eGhanem, D., \u0026amp; Smith, A. (2021). What Are the Benefits of High-Frequency Data for Fixed Effects Panel Models? \u003cem\u003eJournal of the Association of Environmental and Resource Economists\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(2), 199\u0026ndash;234. https://doi.org/10.1086/710968\u003c/li\u003e\n\u003cli\u003eGray, C., Hopping, D., \u0026amp; Mueller, V. (2020). The changing climate-migration relationship in China, 1989\u0026ndash;2011. \u003cem\u003eClimatic Change\u003c/em\u003e, \u003cem\u003e160\u003c/em\u003e(1), 103\u0026ndash;122. https://doi.org/10.1007/s10584-020-02657-x\u003c/li\u003e\n\u003cli\u003eGupta, E. (2016). The effect of development on the climate sensitivity of electricity demand in india. \u003cem\u003eClimate Change Economics\u003c/em\u003e, \u003cem\u003e07\u003c/em\u003e(02), 1650003. https://doi.org/10.1142/S2010007816500032\u003c/li\u003e\n\u003cli\u003eIEA. (2021). \u003cem\u003eElectricity Information: Overview\u003c/em\u003e. https://www.iea.org/reports/electricity-information-overview/electricity-consumption\u003c/li\u003e\n\u003cli\u003eIEA. (2022). \u003cem\u003eIEA Statistics: Countries \u0026amp; Regions\u0026mdash;China\u003c/em\u003e. https://www.iea.org/countries/china\u003c/li\u003e\n\u003cli\u003eKhan, Z., Iyer, G., Patel, P., Kim, S., Hejazi, M., Burleyson, C., \u0026amp; Wise, M. (2021). Impacts of long-term temperature change and variability on electricity investments. \u003cem\u003eNature Communications\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), 1643. https://doi.org/10.1038/s41467-021-21785-1\u003c/li\u003e\n\u003cli\u003eLi, Y., Pizer, W. A., \u0026amp; Wu, L. (2019). Climate change and residential electricity consumption in the Yangtze River Delta, China. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e, \u003cem\u003e116\u003c/em\u003e(2), 472\u0026ndash;477. https://doi.org/10.1073/pnas.1804667115\u003c/li\u003e\n\u003cli\u003eMeinshausen, M., Lewis, J., McGlade, C., G\u0026uuml;tschow, J., Nicholls, Z., Burdon, R., Cozzi, L., \u0026amp; Hackmann, B. (2022). Realization of Paris Agreement pledges may limit warming just below 2 \u0026deg;C. \u003cem\u003eNature\u003c/em\u003e, \u003cem\u003e604\u003c/em\u003e(7905), 7905. https://doi.org/10.1038/s41586-022-04553-z\u003c/li\u003e\n\u003cli\u003eNASA. (n.d.). \u003cem\u003eNASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP)\u003c/em\u003e. NASA Center for Climate Simulation. Retrieved March 3, 2023, from https://www.nccs.nasa.gov/services/data-collections/land-based-products/nex-gddp\u003c/li\u003e\n\u003cli\u003eNational Bureau of Statistics of China (NBSC). (2022). Part II: Statistical Data of Cities at Prefecture Level and Above. In \u003cem\u003eChina City Statistical Yearbook 2021\u003c/em\u003e (1st edition). China Statistics Press. https://www.chinayearbooks.com/china-city-statistical-yearbook-2021.html\u003c/li\u003e\n\u003cli\u003eNayyar, G., Hallward-Driemeier, M., \u0026amp; Davies, E. (2021). \u003cem\u003eAt Your Service?: The Promise of Services-Led Development\u003c/em\u003e. World Bank Publications.\u003c/li\u003e\n\u003cli\u003eNOAA. (n.d.). \u003cem\u003eNational Centers for Environmental Information (NCEI)\u003c/em\u003e. Retrieved March 3, 2023, from https://www.ncei.noaa.gov/\u003c/li\u003e\n\u003cli\u003ePachauri, S., Pelz, S., Bertram, C., Kreibiehl, S., Rao, N. D., Sokona, Y., \u0026amp; Riahi, K. (2022). Fairness considerations in global mitigation investments. \u003cem\u003eScience\u003c/em\u003e, \u003cem\u003e378\u003c/em\u003e(6624), 1057\u0026ndash;1059. https://doi.org/10.1126/science.adf0067\u003c/li\u003e\n\u003cli\u003ePerera, A. T. D., Nik, V. M., Chen, D., Scartezzini, J.-L., \u0026amp; Hong, T. (2020). Quantifying the impacts of climate change and extreme climate events on energy systems. \u003cem\u003eNature Energy\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e(2), 2. https://doi.org/10.1038/s41560-020-0558-0\u003c/li\u003e\n\u003cli\u003ePropato, T. S., de Abelleyra, D., Semmartin, M., \u0026amp; Ver\u0026oacute;n, S. R. (2021). Differential sensitivities of electricity consumption to global warming across regions of Argentina. \u003cem\u003eClimatic Change\u003c/em\u003e, \u003cem\u003e166\u003c/em\u003e(1), 25. https://doi.org/10.1007/s10584-021-03129-6\u003c/li\u003e\n\u003cli\u003eTeng, M., Liao, H., Burke, P. J., Chen, T., \u0026amp; Zhang, C. (2022). Adaptive responses: The effects of temperature levels on residential electricity use in China. \u003cem\u003eClimatic Change\u003c/em\u003e, \u003cem\u003e172\u003c/em\u003e(3), 32. https://doi.org/10.1007/s10584-022-03374-3\u003c/li\u003e\n\u003cli\u003evan Ruijven, B. J., De Cian, E., \u0026amp; Sue Wing, I. (2019). Amplification of future energy demand growth due to climate change. \u003cem\u003eNature Communications\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(1), 2762. https://doi.org/10.1038/s41467-019-10399-3\u003c/li\u003e\n\u003cli\u003eWenz, L., Levermann, A., \u0026amp; Auffhammer, M. (2017). North\u0026ndash;south polarization of European electricity consumption under future warming. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e, \u003cem\u003e114\u003c/em\u003e(38), E7910\u0026ndash;E7918. https://doi.org/10.1073/pnas.1704339114\u003c/li\u003e\n\u003cli\u003eZhang, C., Liao, H., \u0026amp; Mi, Z. (2019). Climate impacts: Temperature and electricity consumption. \u003cem\u003eNatural Hazards\u003c/em\u003e, \u003cem\u003e99\u003c/em\u003e(3), 1259\u0026ndash;1275. https://doi.org/10.1007/s11069-019-03653-w\u003c/li\u003e\n\u003cli\u003eZhang, S., Guo, Q., Smyth, R., \u0026amp; Yao, Y. (2022). Extreme temperatures and residential electricity consumption: Evidence from Chinese households. \u003cem\u003eEnergy Economics\u003c/em\u003e, \u003cem\u003e107\u003c/em\u003e, 105890. https://doi.org/10.1016/j.eneco.2022.105890\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e \u003cem\u003eEC\u003c/em\u003e is the all-sector electricity consumption at the city level for the average effect. We also estimated the same model with \u003cem\u003eEC\u003c/em\u003e being the city\u0026rsquo;s sectoral electricity consumption to examine heterogeneity across sectors.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e As temperature could vary over time and space over a larger scale, we also estimated the model with standard errors clustered at the province-month level. The results are consistent.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e Fifteen cities fall into the North-heating group: Beijing, Tianjin, Xi\u0026rsquo;an, Baoji, Xianyang, Weinan, Yan\u0026rsquo;an, Ankang, Lanzhou, Tianshui, Wuwei, Zhangye, Pingliang, Dixi, and Pangnan.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e See SI Appendix, Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e for cities\u0026rsquo; tertiary GDP share in 2019.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e The classification thresholds are the 25th and 75th percentiles of the data.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e The coefficients of the temperature spline are found to be statistically significant (SI Appendix, Table S4 column 1).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e Higher sensitivity to temperature does not necessarily lead to significant changes in total load, as the tertiary sectors usually consume less electricity than the non-tertiary sectors. For our load project, we used the city-wide estimate grouped by tertiarization tiers and applied consumption weight when computing changes in total electricity consumption.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"electricity consumption, climate change, adaptation, China","lastPublishedDoi":"10.21203/rs.3.rs-3944484/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3944484/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThere is growing empirical evidence that a warming climate will induce adaptation of the electricity system. Less understood is the extent to which climate change could exert on electricity demand given heterogeneity in consumption patterns and the latent mechanisms driving these patterns. We statistically estimate an asymmetric U-shaped temperature response function of city-level daily electricity consumptions using data in years 2018\u0026ndash;2019 and examine heterogeneous responses across regions and economic sectors. Benefiting from the high-frequency electricity consumption data that covers 92 Chinese cities and multiple economic sectors, our findings speak to both the intensive margin as well as the extensive margin adaptation to climate change. We find that access to district heating can explain the asymmetry slopes in temperature-load responses. We also find that although the marginal load responses are statistically significant for all sectors in both high (\u0026gt;\u0026thinsp;25.6\u0026deg;C) and low (\u0026lt;\u0026thinsp;6.6\u0026deg;C) temperature ranges, the tertiary (service) sector load is more sensitive to temperature changes. Taking account of the tertiarization trend, we predict about 66% of the cities will experience more than 3% increase in their summer daily electricity consumption before year 2040. This will likely require substantial investments to expand power grid capacity and to build up energy storage.\u003c/p\u003e","manuscriptTitle":"Electricity consumption and adaptation to climate change: heterogeneity across regions and economic sectors in China","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-02-29 15:31:01","doi":"10.21203/rs.3.rs-3944484/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"71062034-5398-40b7-bbcc-7b07a94d0fbe","owner":[],"postedDate":"February 29th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-09-25T22:22:33+00:00","versionOfRecord":[],"versionCreatedAt":"2024-02-29 15:31:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3944484","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3944484","identity":"rs-3944484","version":["v1"]},"buildId":"-HB7Z8yhvgn0wM9Nzuekk","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}