Research on Measurement and Improvement Path of Total-Factor Carbon Emission Eciency in China's Power Industry: A Perspective of Technological Heterogeneity

: Improving the total-factor carbon emission efficiency of the power industry (TCEPI) is of 7 great significance for realizing the low-carbon development of power industry and promoting the 8 transformation of society to green development. Considering the technological heterogeneity of 9 different regions in China, this paper adopted the Meta-frontier Global Malmquist-Luenberger (MGML) 10 index to measure TCEPI in 30 provinces from 2003 to 2017, and then analyzed the dynamic evolution 11 and regional differences of TCEPI. Finally, the two-step system GMM model was used to explore the 12 influencing factors of TCEPI. The results showed that: (1) During the survey period, the average 13 annual growth rate of TCEPI in China was 4.2%, and average values of TCEPI in all provinces were 14 greater than 1. The innovation effect was the key to TCEPI growth, while the catch-up effect and 15 leading effect were not significant. (2) There was obvious technological heterogeneity in the three 16 regions of China. TCEPI showed a decreasing trend from the western to eastern and central regions, 17 with average annual growth rates of 5.69%, 3.66% and 2.89%, respectively, and the driving factors of 18 each region were different. Moreover, the technology gap among the regions was constantly narrowing. 19 (3) Both the economic development level and the R&D level had played a significant role in promoting 20 TCEPI, while the intensity of power consumption had hindered the rise of TCEPI to a large extent. 21 Based on the conclusions of this article, relevant policy recommendations were put forward to improve 22 TCEPI in China.

carbon peak by 2030 and carbon neutrality by 2060. Achieving this goal requires the joint efforts of all 37 sectors in China, especially the power industry, which must transition to low-carbon development. The 38 power industry is the basic support for social and economic development, but also the largest source of 39 carbon emissions, accounting for more than 40% of China's total carbon emissions. The carbon 40 emission reduction of the power industry has a direct impact on the progress of the overall carbon 41 emission reduction target. Consequently, it is necessary to conduct in-depth research on carbon 42 emission reduction of the power industry.

43
The process of power production has the characteristics of total factors so that the generation of 44 carbon emissions is not a single factor, but the result of multiple factors such as economic development, 45 energy consumption and improvement of living standards. Therefore, the key to carbon emission 46 reduction in the power industry is to improve the total-factor carbon emission efficiency of the power 47 industry (TCEPI). However, due to the imbalance of economic development level, technical condition 48 and resource endowment in different regions of China, there is significant technology gap in power 49 production, which indicates that to measure TCEPI accurately, technological heterogeneity needs be 50 taken into account. For this reason, TCEPI of 30 provinces in China was estimated from the perspective 51 of technological heterogeneity in this paper. Then the dynamic evolution and regional differences were 52 analyzed, and the influencing factors were discussed. This study is conducive to providing scientific 53 data and theoretical references for China to formulate and refine carbon emission reduction policies in 54 the power industry, so as to explore the low-carbon development path of the power industry and 55 promote the transformation of society to green development. The core of the Meta-frontier approach is to construct group frontier and meta frontier separately,

163
The best-practice gap change (BPC) index represents the change rate of technological progress in the 164 group during two periods, reflecting the closeness of the intertemporal frontier of the group, which can 165 be regarded as the "innovation effect". BPC>1 means technological progress, otherwise, it is regressive.

166
The technology gap change (TGC) index indicates the technology gap ratio change, reflecting the 167 change of the gap between the intertemporal frontier and the global frontier of the group during two 168 periods, which can be regarded as "leading "Effect". TGC>1 means that the gap between the 169 intertemporal frontier and the global intertemporal frontier is narrowed, and vice versa. autocorrelation between independent variables and random disturbance items may cause endogeneity 6 , 0 1 , 1 2 2 3 3 4 4 5 5 6 6 , ln ln In the formula, ln is the natural logarithm, i is the province, and t is the year. MGML is the index 188 of TCEPI, lnMGMLi,t-1 is the lagging first order of the dependent variable, and a is the cumulative 189 value. β0 is the constant term, Xj (j=1,...,6) is the independent variable, and βj represents corresponding 190 coefficient. εi,t is a random disturbance term.

191
The specific explanations of these variables are as follows. (1)   The descriptive statistics of the input-output variables are detailed in Table 1

243
As Table 2 shows, the growth rates of input-output variables in each region were quite different.

244
Due to the unbalanced development of each region, the power production has obvious technical 245 heterogeneity, so it is necessary to analyze all provinces in groups. In terms of input, the growth rates 246 of installed capacity across the country and the three regions were higher than energy consumption and 247 labor force, and the growth rate of each input variable in the western region was higher than that in

252
In conclusion, due to the variation in economic development level and resource endowments in 253 different regions, the growth rates of input-output variables are unbalanced, resulting large differences 254 in production technology and production frontiers in the regions. Therefore, it is necessary to take 255 regional technology gap into account to measure TCEPI more accurately.

379
However, the growth rate varied greatly among provinces. For instance, the average annual 380 growth rate in Yunnan was as high as 9.2%, and Hainan was only 0.15%. Accordingly, it is necessary to

405
In order to further clarify the mechanism of TCEPI growth, the dynamic panel regression model 406 was employed to test the influencing factors of TCEPI. lnMGML was treated as the dependent variable,

407
L.lnMGML, lnGDP, lnEPS, lnRD, lnEPS, lnECI, and lnEGI were treated as the independent variables, 408 and Stata 16 software was used for regression. The descriptive statistical results of the variables are 409 shown in Table 5.

441
The coefficient of lnECI was significantly negative, which indicates that the reduction of power

491
(2) Technological heterogeneity needs to be fully considered in different regions and reasonable as to promote coordinated regional development in the power industry.

502
(3) Based on the analysis of the influencing factors of TCEPI, the following measures can be 503 taken: While the economic development level is improving, it is necessary to pay attention to the 504 transformation of economic growth mode and the optimization of industrial structure to explore 505 scientific development models and sustainable development path; The power industry should increase Availability of data and materials The datasets used and/or analyzed during the current study are available 512 from the corresponding author on reasonable request.