基于SBM-DEA与STWR模型的中国能源碳排放效率时空异质性分析

SPATIOTEMPORAL HETEROGENEITY ANALYSIS OF ENERGY CARBON EMISSION EFFICIENCY IN CHINA BASED ON SBM-DEA AND STWR MODEL

中国能源碳排放效率的时空异质性分析,是研究及制定区域能源碳排放效率提升策略,加速实现"双碳"目标的关键之一。提出一种将SBM-DEA(slack based measure-data envelopment analysis)与时空加权回归(spatiotemporal weighted regression,STWR)模型结合的能源碳排放时空异质性分析框架,先基于SBM-DEA模型计算省域能源碳排放效率指标(energy carbon emission efficiency index,ECEI),再通过STWR模型分析2012—2019年该效率指标与对外开放程度、城镇化水平、科技经费投入和煤炭消费占比等主要驱动力之间的时空非平稳性关系,并基于动态时间规整算法(dynamic time warping,DTW)计算各变量对应省域回归系数时间序列之间的相似度,结合Elbow方法运用K-Medoids进行聚类。结果表明:中国能源碳排放量总体呈增长趋势,但ECEI并未提高。1)对外开放程度对ECEI的影响自西向东呈阶梯分布,正向影响强度整体呈西部地区>中部地区>东部地区。2)城镇化水平的影响大部分为正向,程度先增后减,并于2015年达到最大。中国南部城镇化水平与能源碳排放效率呈"U"形关系,负向影响程度先增后减。3)科技经费投入与能源碳排放效率主要呈正相关,沿海地区相对稳定,湖北、湖南等地区正向影响逐渐增强,而东北三省和四川等地区却呈负相关。4)2013—2017年各地区煤炭消费占比负向影响集中于中部地区并呈外延趋势,且程度逐渐减弱。河北、河南及陕西一带煤炭消费占比的负向影响较大,且呈"W"形波动变化。此分析框架综合环境、资源消耗和社会价值等多个评价维度,能更科学合理地测度ECEI,且其首次引入STWR模型,能有效地探析其与各主要驱动因素的时空异质性。通过省域系数时序聚类来辅助识别碳排放效率的时空模式,可为科学、动态协调区域能源消费和CO_(2)排放提供参考。

Analysing the spatiotemporal heterogeneity of China’s energy carbon emission efficiency is one of the keys to researching and formulating regional energy carbon emission efficiency improvement strategies and accelerating the realization of the Double Carbon goal.This study proposed a framework for analyzing spatiotemporal heterogeneity of energy carbon emissions based on the combination of the SBM-DEA(slack-based measure-data envelopment analysis)and the spatiotemporal weighted regression(STWR)model.Firstly,the energy carbon emission efficiency index(ECEI)was calculated based on SBM-DEA model.Then,the spatiotemporal non-stationary relationships between the efficiency index and its main driving forces from 2012 to 2019,i.e.,the degree of opening to the outside world,the level of urbanization,the investment in science and technology,and the proportion of coal consumption were built by using STWR.Furthermore,based on the dynamic time warping algorithm(dynamic time warping,DTW),the similarity between the time series of different coefficients corresponding to each independent variable,which was generated by the STWR model,was calculated.The K-Medoids were employed to cluster based on the similarity with using the Elbow method to determine the optimal cluster number of K.The results show that China’s energy carbon emissions are generally increasing,but the ECEI has not improved.1)Among them,the degree of openness to the ECEI presents a ladder distribution from the western to the eastern,and the overall positive impact intensity is western regions>central regions>eastern regions.2)Most of the impact of urbanization level is positive,the degree increases first and then decreases,and reaches the maximum in 2015.The urbanization level in southern China has a U-shaped relationship with energy and carbon emission efficiency,and the degree of negative impact first increases and then decreases.3)The investment in science and technology is mainly positively correlated with the efficiency of energy and carbon emissions.The coastal areas are relatively stable,and the positive impact of Hubei and Hunan is gradually increasing,while the three northeastern provinces(Heilongjiang,Jilin,and Liaoning)and Sichuan are negatively correlated.4)From 2013 to 2017,the proportion of coal consumption in each region had a negative impact on energy carbon emission efficiency,concentrated in the central region and showed an extension trend,and its influence gradually was weakened.The proportion of coal consumption in Hebei,Henan,and Shaanxi has a large negative impact on energy carbon emission efficiency,showing a W shape.The proposed analysis framework can conduct a multi-dimensional evaluation of environmental impact,resource consumption,and social value,and measure China’s carbon emission efficiency more scientifically and reasonably.It is the first time to introduce the STWR model,which can be used to explore the sub-stationarity relationship between the energy carbon emission efficiency value and the main driving factors and its change over time.Clustering of time-series coefficients can help identify the spatiotemporal pattern of carbon emission efficiency and support decision-making to coordinate regional energy consumption and carbon dioxide emissions dynamically and rationally.