Urban Low-Carbon Consumption Performance Assessment: A Case Study of Yangtze River Delta Cities, China

Abstract

Urban low carbonization has been an essential element in China’s carbon peak and carbon neutrality strategies. An assessment of urban low-carbon performance could provide valuable information for monitoring and guiding the low-carbon transition in cities. However, due to cross-regional carbon transfer, the actual level of achievement would be masked, if the assessment was based only on a production-based index such as carbon emission intensity (CEI). Focusing, instead, on consumption-based low-carbon performance, this study calculated levels of urban carbon consumption intensity (CCI) based on city-level carbon footprint accounting, investigated the patterns and drivers of changes in CCI of 26 Yangtze River Delta (YRD) cities from 2012 to 2015, and conducted a comparative analysis of CEI and CCI data from both static and dynamic viewpoints. It was found that the CCI of YRD cities decreased from 1.254 to 1.153 over the period. Cities at higher economic levels were found to have lower CCI values. Decomposition results show that shifts in production structure, intensity of emissions and changing consumption patterns contributed to the decline in CCI of the YRD area. Richer cities were found to show greater declines in CCI due to decarbonizing structures in production and consumption. The comparative results show that although the CEI and CCI of cities were generally correlated in both static level and dynamic change, the net carbon transfer impacted the correlation sensitivity between various cities. Finally, our findings provide practical guidance on achieving coordinated emission reductions at an inter-city level from both production and consumption perspectives.

1. Introduction

Carbon emission reduction has become a matter of global consensus to address climate change and achieve the sustainable development of human beings [1]. As home to most of the world’s population and productive assets, cities account for 67–76% of global energy consumption and 71–76% of CO₂ emissions from energy combustion [2]. Especially in China, the largest developing country, urbanization has served as a major driver for economic growth and cities produce 85% of CO₂ emissions in the country [3]. In pursuit of the goals of carbon peak and carbon neutrality, urban carbon mitigation has become a critical issue in China [4].

How to coordinate economic development with emission reduction in cities is one of the predicaments facing the world. Kinzig and Kammen [5] proposed the “low-carbon economy” as a new urban development mode to resolve the tensions between economic development and ecological protection. Therefore, developing the “low-carbon city” has become one of the most important strategies to tackle the dilemma caused by the desire for both economic growth and climate mitigation in cities [6]. The National Development and Reform Commission (NDRC) of China has also introduced a low-carbon city pilot project to advance low carbonization in the country’s cities [7]. This program aims to promote the low-carbon transition in the areas of economic development, industrialization, urbanization, and residential lifestyle.

The evaluation and decomposition of low-carbon performance is valuable as it could provide significant information for monitoring and guiding the process of low-carbon development [8,9]. In this regard, carbon emission intensity (CEI), referring to the ratio of CO₂ emissions to GDP, is an index that reflects the level of low-carbon development in a country or region [10]. CEI has been widely applied to evaluate coordinated economic–environmental development and set carbon mitigation targets [11]. For example, Wang et al. [12] used the DMSP/OLS night-time light imagery to estimate temporal–spatial differences in CEI levels of 341 cities in China from 1992 to 2013. Xu and Xu [13] investigated the CEI of heavy industry in China’s 30 provinces from 2005 to 2019 by using a semiparametric model. Feng et al. [14] examined the impact of a low-carbon city pilot project on a city-level CEI by using a matched difference-in-differences model.

At the same time, tracking the drivers behind changes in CEI could provide directional guidance for formulating climate mitigation policies. This is an area of investigation that has attracted considerable attention [15]. In this regard, index decomposition analysis (IDA), structural decomposition analysis (SDA) and production-theoretical decomposition analysis (PDA) are all popular techniques for investigating the factors influencing changes in CEI. For example, Robaina and Neves [16] used the IDA method to identify the factors influencing CEI in the transportation sector of EU-27 countries in the period 2008–2018. Chen et al. [17] combined IDA and PDA models to investigate the determinants behind changes in China’s CEI in 2005–2020. Pan et al. [18] used the IDA model to decompose the change in CEI of 29 provinces and cities in China between 1998 and 2019. Yan et al. [19] applied the SDA model to investigate the determinants behind changes in the CEI in China from 2002 to 2012. A common finding of these studies is that energy efficiency improvement is the main driver of reductions in CEI levels from both demand and supply sides.

There is an emerging trend in that cities are becoming increasingly interconnected with the rapid development of inter-city transportation, leading to increased urban agglomerations that cross city boundaries and contain multiple urban areas [20]. Urban agglomeration involves close links between cities in terms of labor resources, technological elements, and production processes [21]. Through cross-regional trade in goods and services, water resources are virtually redistributed across regions. The Yangtze River Delta (YRD) urban agglomeration, located on the southeastern coast of China, consists of 26 cities in the provincial regions of Jiangsu, Zhejiang, Anhui and Shanghai. As one of the richest and most populous regions in China, the YRD agglomeration was home to a population of 132.6 million and produced 20.7% of China’s total GDP, according to figures for 2019. However, the rapid urbanization and modernization of the region resulted in high levels of energy consumption and carbon emissions. According to the latest statistics, in 2018, CO₂ emissions in the YRD amounted to 1.7 Gt, accounting for 16.6% of China’s total carbon emissions [2]. Against the background of China’s pledge to achieve carbon peak by 2030, and carbon neutrality by 2060, the YRD region is confronted with the dual task of achieving economic development and emission reduction. For this reason, the YRD Urban Agglomeration Development Plan was initiated to encourage low-carbon development and improve urban low-carbon performance in the region.

Previous studies have investigated low-carbon performance based on the production-based CEI [22]. However, the widespread activities of trade have created inter-regional CO₂ flows from consumers to producers [23] and ignoring these cross-regional carbon links may result in a failure to achieve emission reduction targets. While the CEI may effectively reflect low-carbon performance from the production perspective, such estimates could be biased as some regions achieve their targets for CEI reductions through outsourcing emissions to other regions [24]. With regards to the measurement of consumption-based emissions, the indicator of carbon footprint (CF), developed by Wiedmann and Minx [25], has been widely adopted to measure the total CO₂ emissions directly or indirectly induced by an individual or a region [26]. The CF indicator is capable of supplementing traditional methods of analysis of residents’ needs by combining production- and consumption-based perspectives and contributes to the advancement of knowledge of how humans exert pressure on the environment [27]. The method of multi-regional input–output analysis (MRIO) has been applied in several previous studies to investigate regional CFs and inter-regional CO₂ transfers at global [28,29,30,31], national [32,33,34,35], provincial [36,37,38], and city [20,39,40] levels.

Although previous studies have accounted for regional consumption-based emissions, they have rarely shed light on the comparison of emissions embodied in final consumption on a per unit basis, so as to measure consumption-based LCC performance effectively. In fact, shifts in residential lifestyle and consumption patterns are increasingly regarded as a significant component in addressing global climate change [41]. Consumption-based climate mitigation efforts such as shifting the products consumed by residents towards options with a lower environmental impact have been employed to reduce emissions [42]. In this regard, carbon consumption intensity (CCI), which expresses the CF embodied in one unit of final consumption, may offer an effective instrument to measure the performance of LCCs. Investigating the patterns and drivers of CCI could help to guide the process of consumption transformation and provide directional guidance for formulating climate mitigation policies on the consumption side. However, few studies have investigated the CCIs of various regions, let alone revealed the underlying mechanisms behind changes in CCI levels.

To this end, this study constructed a CCI indicator based on CF accounting and investigated the patterns, drivers, and implications of the CCIs of 26 YRD cities from 2012 to 2015. Targeted implications were provided for various regions to sketch appropriate and directed policy methods. To summarize, the novel elements of this study include the following: (1) a “city-region” nested MRIO table including 26 YRD cities was constructed to conduct a three-scale carbon-extended IO analysis of the urban agglomeration; (2) an urban CCI indicator was proposed for assessing the LCC performance of the 26 YRD cities; (3) a new structural decomposition model for CCI was constructed to explore the determinants behind the change in the CCIs of YRD cities; and (4) the multiple mechanisms behind China’s CCI in terms of pattern characteristics and influencing factors were revealed and thereby identify an effective low-carbon transformation path for YRD cities from both production and consumption perspectives. The rest of the current study is structured as follows: in Section 2, the methods and materials are introduced. Estimates, determinants, and implications of the CCIs of YRD cities are presented in Section 3. Section 4 concludes the study and provides policy implications.

2. Materials and Methods

As shown in Figure 1, following the research route of evaluation–decomposition–comparison analysis, this study combined MRIO, CCI indicator, SDA and correlation analysis techniques to reveal the patterns and drivers of changes in the CCI of the YRD area, to quantify the contributions of different cities, and compare the production- and consumption-based low-carbon performances of cities from both static and dynamic viewpoints. This section introduces the definition and decomposition of the CCI, and the data sources used.

2.1. Low-Carbon Consumption Performance

The MRIO model was applied to estimate emissions embodied in trade by considering regional disparities and technological differences [43]. Based on the horizontal accounting balance, we can derive the formula as follows:

$$X=AX+Y$$

(1)

where $A$ stands for the direct consumption matrix; $X$ denotes the total output vector; and $Y$ stands for the final demand vector.

Solving for $X$ yields:

$$X={\left(I-A\right)}^{-1}Y=L\times Y$$

(2)

where $L={\left(I-A\right)}^{-1}$ is the Leontief inverse matrix, indicating the demand of total output for satisfying one unit of final consumption.

Based on environmental input–output analysis, the CO₂ emissions triggered by final demand (i.e., $CF$) can be expressed mathematically as:

$$CF=EE+HH=D\times L\times Y+HH$$

(3)

where $EE$ denotes the embodied emissions; $D$ denotes the column vector of carbon emission coefficient; and $HH$ denotes household direct emissions.

As part of the quest for better measures of carbon emissions intensity, we suggest using an indicator defined as ‘carbon consumption intensity’ (i.e., carbon emissions per unit of final consumption) to normalize the carbon footprint into a common scale:

$$\mathrm{CCI}=CF/y$$

(4)

where $\mathrm{CCI}$ denotes carbon consumption intensity, indicating the level of LCC performance. Higher CCI means more CO₂ required for satisfying one unit of final demand, thereby indicating a lower level of LCC performance and vice versa. $y$ denotes the sum of final demand.

2.2. Structural Decomposition Model

For Equation (3), we further decompose the final demand vector $Y$ into two components (i.e., average consumption structure and total consumption volume) to quantify the impacts from different final consumption factors:

$$Y=S\times y$$

(5)

where $S$ denotes the consumption patterns vector; $y$ denotes the total consumption volume. Therefore, Equation (3) can be transformed to:

$$CF=EE+HH=DLSy+HH$$

(6)

Substituting Equation (6) into Equation (4) yields:

$$\mathrm{CCI}=\frac{CF}{EE}\frac{EE}{y}=kDLS$$

(7)

where $k$ denotes the ratio of carbon footprint to embodied emissions, indicating the composition structure of emissions.

As shown in Equation (7), the four factors of emission composition, emission intensity, production structure and consumption patterns, could fully explain the CCI changes in a given period. A total difference of Equation (7) generates Equation (8):

$$\Delta \mathrm{CCI}=\Delta kDLS+k\Delta DLS+kD\Delta LS+kDL\Delta S$$

(8)

where $\Delta$ denotes the difference operator.

Equation (8) converts four multiplicative terms in Equation (7) into four additive terms, and each additive term corresponds to the contribution of each factor. Due to the absence of a unique decomposition solution for Equation (8), the average of the polar decompositions was applied to approximate the average of all decompositions [44]. Solving Equation (8) with the so-called polar decompositions yields:

$$\begin{array}{ll}\hfill \Delta \mathrm{CCI}& =\frac{1}{2}\times \left[\Delta k{D}_0{L}_0{S}_0+\Delta k{D}_1{L}_1{S}_1\right]+\frac{1}{2}\times \left[k_1\Delta D{L}_0{S}_0+k_0\Delta D{L}_1{S}_1\right]\hfill \\ & +\frac{1}{2}\times \left[k_1{D}_1\Delta L{S}_0+k_0{D}_0\Delta L{S}_1\right]+\frac{1}{2}\times \left[k_1{D}_1{L}_1\Delta S+k_0{D}_0{L}_0\Delta S\right]\hfill \\ & =\Delta k+\Delta D+\Delta L+\Delta S\hfill \end{array}$$

(9)

where $\Delta \mathrm{CCI}$ represents the CCI changes; $\Delta k$, $\Delta D$, $\Delta L$, and $\Delta S$ are the contributions of emission composition, emission intensity, production structure, and consumption patterns, respectively.

2.3. Data Sources

For the MRIO table, we compiled two nested MRIO tables concentrating on YRD cities from the China city-level MRIO tables for 2012 and 2015 which were obtained from the CEADs [45] (https://www.ceads.net/ (accessed on 27 June 2021)). The nested tables describe the economic transaction data for 1386 sectors of 33 regions (26 YRD cities and the other 7 regions) in China (excluding Macao, Hong Kong, and Taiwan). The 42 sectors in the MRIO table were aggregated into 27 sectors. The detailed industry classification can be found in Supplementary Materials Table S1. Additionally, the economic values in the MRIO tables were deflated to a 2000 constant price in order to eliminate the statistical impacts of the price factor.

For the emission inventory, we obtained production-based CO₂ emission inventories for 30 provinces in 2012 and 2015 from the CEADs [46,47]. Production-based CO₂ emission inventories for the 26 YRD cities in 2012 and 2015 were estimated using China’s city statistics and IPCC guidelines for national greenhouse gas accounting, in which emissions are computed by multiplying activity levels by corresponding emission factors. The 45 sectors in provincial emission inventories were also aggregated into 27 sectors (See Supplementary Materials Table S1).

3. Results and Discussion

This section presents evaluation, decomposition and comparison results to reveal the patterns and drivers of changes in the CCI of the YRD, to quantify the contributions of different cities, and compare the production- and consumption-based low carbon performances of cities from both static and dynamic viewpoints.

3.1. Carbon Consumption Intensity Estimates

Equations (3) and (4) were employed to compute the CCI scores of 26 YRD cities in 2012 and 2015, and the estimates are presented in

Table 1. Carbon consumption intensity estimates of cities.

Cities	2012	2015	Mean	Rank	Cities	2012	2015	Mean	Rank
Shanghai	1.051	0.979	1.015	1	Huzhou	1.296	1.262	1.279	13
Nanjing	1.180	1.037	1.109	5	Shaoxing	1.659	1.616	1.637	24
Wuxi	1.137	1.004	1.071	2	Jinhua	1.534	1.495	1.515	21
Changzhou	1.185	1.060	1.123	6	Zhoushan	1.112	1.065	1.088	4
Suzhou	1.144	1.002	1.073	3	TaizhouZJ	1.385	1.360	1.372	17
Nantong	1.267	1.177	1.222	8	Hefei	1.535	1.457	1.496	20
Yancheng	1.266	1.211	1.239	11	Wuhu	1.643	1.585	1.614	23
Yangzhou	1.323	1.220	1.271	12	Maanshan	1.682	1.738	1.710	25
Zhenjiang	1.178	1.130	1.154	7	Tongling	1.689	1.758	1.724	26
TaizhouJS	1.270	1.204	1.237	10	Anqing	1.315	1.343	1.329	16
Hangzhou	1.301	1.277	1.289	14	Chuzhou	1.549	1.634	1.591	22
Ningbo	1.284	1.308	1.296	15	Chizhou	1.405	1.465	1.435	19
Jiaxing	1.277	1.172	1.225	9	Xuancheng	1.400	1.352	1.376	18

. It can be observed that the cities with the best LCC performance, that is, the lowest CCI, were, in rank order, Shanghai, Wuxi, Suzhou, Zhoushan and Nanjing, the average CCI scores of which were 1.015, 1.071, 1.073, 1.088 and 1.109, respectively. On the contrary, the worst LCC performers were, in rank order, Tongling, Maanshan, Shaoxing, Wuhu and Chuzhou, the average CCI scores of which were 1.724, 1.710, 1.637, 1.614 and 1.591, respectively. Evidently, there were significant differences of LCC performance in different cities. More specifically, the LCC performance of cities located in the inland province (i.e., Anhui) was significantly lower than those cities in coastal provinces such as Shanghai, Jiangsu and Zhejiang.

The significant spatial disparities of LCC performance between YRD cities may be related to differences in urban development level. As shown in Figure 2, cities with higher GDP per capita (e.g., Wuxi, Shanghai, Nanjing, Suzhou and Hangzhou) were found to have lower CCI values, while cities with lower GDP per capita (e.g., Anqing, Xuancheng, Chuzhou and Chizhou) had higher CCI values. To support this finding, a correlation analysis between CCI and GDP per capita was conducted for the sample of 26 YRD cities. As shown in Supplementary Materials Table S2, the CCIs of cities in the YRD are found to show a significantly negative correlation with their GDP per capita, verifying the impact of urban economic development on CCI levels.

In fact, due to differences in their stages of economic development, cities in different locations show significant differences in urban infrastructure construction [48]. Cities in the coastal provinces with a higher development level have a better building and transportation infrastructure. They therefore have less demand for infrastructure construction and have less need for emission-intensive products such as steel, cement and other building materials [49]. In contrast, less developed cities in Anhui province are at the stage of rapid industrialization and urbanization, in which infrastructure construction is the main task, resulting in a higher demand for energy- and emission-intensive commodities in comparison with more economically advanced cities [50]. As a result, the spatial differences in consumption preferences caused by urban development demand are the primary reason for the differences of LCC performance between cities.

3.2. Determinants of Changes in Carbon Consumption Intensity

Equation (9) was used to decompose the changes in the CCI of the YRD area from 2012 to 2015, and the results are illustrated in Figure 3. The CCI value of the YRD was found to decrease by 8.1% in the period (Figure 3, grey bar), which was mainly the result of shifts in production structure, emission intensity and consumption patterns. More specifically, these three factors exerted downward influences of 4.6%, 2.2% and 1.6% (Figure 3, green, blue and purple bars), respectively. The effects of emission composition had little impact on the change in CCI (Figure 3, yellow bar), indicating that the emissions from households accounted for a minor and stable proportion of the regional CF.

The significant decline in CCI caused by changes in production structure was due to the implementation of sustainable strategies and low-carbon policies in the new normal era [51]. Several plans have been promulgated to advance the integrated development of the YRD cities, including “YRD Regional Planning” in 2010, the “YRD City Cluster Development Plan” in 2016, and the “YRD Regional Integrated Development Plan” in 2019 [52]. Policy measures such as eliminating backward and outdated production capacity, have significantly improved the production technology and factor productivity of various sectors. As a result, the volume and type of intermediate products consumed by various industry sectors have evolved and become more efficient, and the production structure of the YRD also shifted towards a lower carbon and greener direction.

The three main determinants behind the change in CCI were then grouped into two categories: technical improvement (shift in emission intensity) and structural adjustment (shift in production structure and consumption patterns). Figure 4 illustrates contributions to changes in CCI in terms of technical improvement and structural adjustment among 26 YRD cities. The negative 45-degree red line divides outcomes of CCI comparisons in terms of technical improvement and structural adjustment among 26 cities during 2012–2015. The orange square points above the line denote cities with a net CCI increase, while the blue circular points below the line represent the cities with a net CCI reduction. More specifically, 6 of the 26 YRD cities (that is, Tongling, Chuzhou, Maanshan, Anqing, Chizhou and Ningbo) were found to have a net increase in CCI during the study period, while the remaining 20 cities achieved a net decline in CCI.

Considering the contribution of technical improvements at a city level, it can be seen that most points are located in the two quadrants on the left, indicating that most YRD cities have improved their technical level in the period. In fact, due to the wide promotion of advanced energy-saving and emission-reducing technologies, higher technical levels and more efficient applications in the energy sector and in other industrial sectors have brought about a substantial improvement in the period [53]. In particular, a few economically poor cities such as Tongling, Chuzhou, TaizhouZJ, Wuhu, Xuancheng and Maanshan, were found to have made a larger contribution in terms of technical improvements, but their carbonizing structural adjustment largely offset or even outweighed the beneficial effects of technological advancement. This may be attributed to the extensive economic development and rapid infrastructure construction in these cities, with consequent higher production and consumption of emission-intensive products and lower factor utilization efficiency [54].

In contrast, 17 of the 26 YRD cities are situated in the two lower quadrants, indicating that the structures of these cities in terms of production and consumption have become lower carbon. In 2012, China entered “the new normal” development stage, the main characteristics of which are high-quality development with inclusive and sustainable growth. In fact, by further eliminating backward capacity, improving production technologies and restructuring the industrial structure, the quantity and type of intermediate products used by various sectors has evolved and become more efficient [55]. Meanwhile, due to the increasingly promoted green economy and lifestyle, shifts in consumption patterns, such as increased purchasing of energy efficient and low-carbon products, have also contributed to the decline in CCI. As a result, structural adjustments in production and consumption have played a significant role in reducing regional CCI levels, especially in economically advanced cities in coastal provinces such as Suzhou, Wuxi and Nanjing (Figure 4).

3.3. Carbon Consumption Intensity vs. Carbon Emission Intensity

Figure 5 compares CEI and CCI scores among 26 YRD cities. It can be seen that developed cities (e.g., Shanghai, Wuxi, Suzhou and Nanjing) had the lowest scores for both CEI and CCI, while less developed cities (e.g., Tongling, Maanshan, Huhu and Chuzhou) had relatively higher scores, particularly in the case of CEI. Additionally, as shown by the trend line (red line, Figure 5), the CCI was found to be positively correlated with the CEI, but the correlation sensitivity was found to vary largely across different cities. More specifically, the slope of the trend line presented a decreasing trend from the developed cities to the less developed cities. This result means the developed cities should aim to lower carbonization of consumption (decline in CCI) with less emphasis on emission reduction efforts (decline in CEI), while the less developed cities should seek a substantial reduction in emissions so as to decarbonize their consumption.

The differences in the correlation sensitivity between CEI and CFI among various cities are mainly due to the “carbon leakage” phenomenon caused by the spatial differences in urban industrial specialization and development stages. Figure 6 shows the carbon balance (the difference between production- and consumption-based emissions) of the 26 YRD cities. Developed cities such as Shanghai, Hangzhou, Nanjing, Hefei and Suzhou were found to have a net carbon outflow (blue shading), while less developed cities such as Tongling, Ningbo, Huzhou and Maanshan had a net carbon inflow (red shading). Developed megacities tended to outsource their CO₂ emissions to the inland regions by transferring heavy industry and importing emission-intensive products [40]. In contrast, less developed cities need to maintain backward production capacity and increase their CO₂ emissions to meet the final consumption demands of other regions. In particular, the integrated development of the YRD has promoted the establishment of metropolitan areas, including the Nanjing, Hangzhou and Hefei urban circles, which not only strengthen the exchanges between the core and surrounding cities in the metropolitan areas, but also aggravate the carbon transfer between them (Figure 6).

In view of the above-mentioned facts, a correlation analysis between the carbon balance and the difference between CEI and CCI was conducted for the sample of 26 YRD cities. As shown in Supplementary Materials Table S3, the differences between the CEIs and CCIs of cities were found to be significantly correlated with their carbon balance. This implies that the developed megacities are shirking their responsibility to reduce emissions while outsourcing carbon-intensive industries to the less developed small cities; that is, they achieved CO₂ emissions reduction targets at the expense of their supplying regions.

Figure 7 further compares the CEI variation rate ($V_{\mathrm{CEI}}$) and CCI variation rate ($V_{\mathrm{CCI}}$) among the 26 YRD cities during 2012–2015. It can be seen that most of the cities, such as Suzhou, Wuxi, Hefei, Wuhu, TaizhouJS, Zhenjiang, Xuancheng, Shaoxing, Jinhua, Anqing, Tongling, are located close to the 45-degree line, implying a positive correlation between the CEI and CCI variations. To support this finding, a correlation analysis between CEI and CCI variations was conducted for the sample of 26 YRD cities. As shown in Supplementary Materials Table S4, CCI variations among YRD cities were found to be significantly correlated with their CEI variation. This is understandable as the majority of products and services consumed by cities are provided by themselves, leading inevitably to a close link between production and consumption inside the city.

Additionally, the line divides the cities into two groups: the points above the line ($V_{\mathrm{CEI}}\le V_{\mathrm{CCI}}$) and the points below the line ($V_{\mathrm{CEI}}\ge V_{\mathrm{CCI}}$). For the cities in the former category, their CCI variation rates were greater than the corresponding CEI variation rates (here refers to signed value). For example, Shanghai was found to have a decline of 6.8% in CCI, but its CEI variation was as low as −16.8%. A similar situation was also observed in cities such as Hangzhou, Nanjing and Nantong, most of which have a higher level of economic development. For the cities in the second category, CCI variation rates were less than their corresponding CEI variation rates (here refers to signed value). For example, Jiaxing reported a reduction of 2.0% in CEI, but its CCI experienced a decline of 8.3%. The same case was true for cities such as Huzhou, Changzhou and Chizhou.

Obviously, although there was a significant correlation between the CEI and CCI variations, the difference between the two variations also revealed significant spatial disparity characteristics. Richer cities (e.g., Shanghai, Hangzhou, Nanjing and Suzhou) had greater reductions in CCI than in CEI during 2002–2015, while less developed cities (e.g., Huzhou, Changzhou, Yangzhou and Chizhou) experienced the opposite situation. This result may be attributed to the shift in carbon transfer patterns (i.e., the change in net carbon flow of cities). If the net carbon outflow (inflow) of a city increased (decreased) from 2012 to 2015, its CEI reduction would be larger than the CCI reduction, and vice versa. The correlation analysis indicates that the difference between changes in CEI and CCI variations was significantly correlated with changes in net carbon flows of cities (see Supplementary Materials Table S5).

4. Conclusions and Policy Implications

This study constructed an indicator of CCI to assess urban LCC performance, and then investigated the patterns, drivers, and implications of CCI in YRD cities between 2012 and 2015. Cities with higher levels of economic development were found to have lower CCI values and better LCC performance. Decomposition results show that shifts in production structure, emission intensity and consumption patterns contributed to the decline in CCI of the YRD area. Richer cities were found to have a greater decline in CCI due to decarbonizing structure in production and consumption. Correlation results show that the CCI variation of cities in the YRD was significantly correlated with their CEI variation. Moreover, the carbon leakage from megacities to small cities led to the difference between CEI and CCI variations among the various cities.

Based on these findings, policy implications regarding production- and consumption-based emission reduction can be formulated. From the production perspective, the carbon market should be further developed to incorporate more industries and enterprises into the trading system. Participation in emission trading would motivate firms to reduce emissions to achieve an emission permits surplus, as extra revenues can then be obtained by selling the excess permits in the market. In addition, the mechanism would reduce the incentive of firms to substitute domestic production with imports and thus resolve the carbon leakage issue. Additionally, the ‘‘border carbon adjustment mechanism’’ might be an alternative means to address carbon leakage by placing a levy on imports in proportion to their carbon content.

At the same time, technological innovation and structure adjustment efforts should be made to decrease the emission intensity of heavy industry, especially in less developed cities. Notably, the national urbanization rate is expected to reach 65% by 2025, as set out in the “China 14th Five-Year Plan”. Hence, the deployment of green infrastructure in advance, as well as the reduction in emission intensity of building materials and transportation infrastructure, is essential to avoid a substantial increase in CO₂ emissions in the future.

From the consumption perspective, efforts should be made to guide low-carbon consumption behaviors and develop low-carbon consumption habits in the public, as well as encouraging citizens to enjoy a low-carbon and greener lifestyle. Beneficial policies such as choosing public transportation instead of using private cars, using energy-efficient appliances, and avoiding excessive and unnecessary consumption should be vigorously promoted to drive the low-carbon transformation of consumption patterns. Further, policies such as tax incentives and sales subsidies could be gradually implemented to restrict demand for emission-intensive products.

This study certainly has some limitations. Due to the data availability of the China city-level MRIO table, the timeframe for the study was restricted to the period 2012–2015. In the future, with available data for the years after 2015, the study period can be further extended to reveal more up-to-date findings about CCI in YRD cities. In addition, the empirical theory applied in this study of the YRD region can be extended to the whole country to further investigate the impact of “Low-carbon City Pilot Initiate” on urban LCC performance.