Impact of Urban Mining on Energy Efficiency: Evidence from China

Abstract

Achievement of carbon neutrality requires lowering energy consumption in China. However, only a few studies have focused on the impact of urban mining on the reduction in energy consumption, and the impact of urban mining on reducing energy consumption remains unclear. This study explored the impact of urban mining on energy efficiency by adopting the panel data of 281 prefecture-level cities in China between 2003 and 2016 using the difference-in-difference method, and tested the setting up of urban mining pilot bases on reducing urban energy consumption per unit of gross domestic product (GDP) and channel mechanisms. The empirical findings show that urban mining pilot bases have significantly reduced energy consumption per unit of GDP by 3.67%. The instrumental variable method was used to overcome endogenous problems of the empirical results and related robustness tests. The verification results of the impact mechanism show that urban mining reduces the energy consumption per unit of GDP through three channels: by improving the degree of urban marketization, enhancing the harmony between the government and the market, and increasing the abundance of factor markets. The heterogeneity analysis shows that urban mining has a significant reduction effect on energy consumption per unit of GDP in all four regions of China; however, the energy-saving effect in the northeast is higher than that in the other regions. This study provides an empirical test and policy insights for the contribution of urban mining pilot bases in promoting China’s development of green industry and circular economy. It also offers a new path for cities in developing countries to promote energy conservation and achieve urban sustainability via urban mining.

1. Introduction

Promoting energy efficiency and fostering energy conservation are two key tasks for achieving green development and have received attention worldwide [1,2,3,4]. Improving energy efficiency is an effective way to balance economic development and energy consumption [5]. In the context of carbon neutrality, promoting the resource recycling industry and reducing energy consumption is crucial [6,7,8]. China’s rapid industrial development has solved the problem of food, clothing, and employment for its population of 1.4 billion. However, continued industrialization is increasing the volume of industrial pollutants that are being discharged into the natural environment, leading to serious environmental pollution [9]. The annual increasing rate of municipal solid waste (MSW) has reached 10% and exceeded 4 billion metric tons [10]. Moreover, solid wastes, mainly composed of industrial solid waste (ISW), broken household appliances, abandoned electronic products, and scrapped automobiles, are accumulating in urban areas and cities [11]. Although industries are the greatest energy consumer, cities are the primary areas promoting industrial development. A large number of cities in China are contaminated by various types of solid wastes, which severely damage the ecological environment and harm the health of urban residents [12].

Rapid industrialization and urbanization have greatly increased China’s demand for mineral resources, mainly those obtained via virgin mining. Compared with virgin mining, which generates massive environmental waste [13], urban mining (UM) has lower environmental consequences because the process of recovering recyclable urban wastes requires fewer inputs and results in less pollution [4,13,14,15,16]. UM refers to the recycling of steel, non-ferrous metals, rare metals, and other resources produced and stored in urban areas. Cities should be treated as urban mines to recover the materials stored there and to maximize material efficiency. Recyclable resources can be recovered and reused to replace the exploitation of primary mineral resources, thereby avoiding high energy consumption and high emissions, and helping to form a closed-loop supply chain in a circular sustainable economy to ensure resource availability [6]. Continuous accumulation of solid e-product waste in cities can be recovered as a secondary raw resource via urban mining that replaces primary minerals and energy [17,18,19].

Urban mining has been widely studied and is regarded as a major strategy for climate mitigation because it lowers the demand for primary materials and reduces carbon emissions [14,15,20,21,22]. Urban mining in the buildings sector mainly focuses on concrete and metal recycling [15,20,23]. Most of prior research has focused on the urban mining of electronic waste, automobiles, vehicles, and electrical and electronic products and equipment [3,21,24]. The existing literature also analyzes UM from the perspective of its economic performance [4,25,26,27]. Concerns on the potential risks of recycling, including the dispersal and accumulation of hazardous substances in the recycled materials, have also been raised [28,29,30]. Recycling activities should not only focus on where the waste is routed, but also on the quality and utilization of recycled materials [29,31]. In addition, concerns on potential risks of recycling were also expressed.

As the material basis for the long-term development of a country, energy sources including coal, oil, and natural gas are vital for promoting social progress. China’s energy consumption has been dominated by coal for a long period of time. Therefore, improving energy efficiency is key to solving the contradiction between ecological environmental protection and economic green development in China. Fostering energy efficiency and reducing carbon emissions are two key strategies for reaching carbon-neutrality targets in China [8]. The literature on energy efficiency in China mainly focuses on three aspects: energy efficiency calculations, factors that influence energy efficiency, and regional differences in energy efficiency [1,32,33,34].

The Urban Mining Pilot Base (UMPB) policy was implemented in China when the country faced the dual challenge of severe resource shortage and heavy environmental pollution in urban areas. To realize the diversified recovery, centralized treatment, and large-scale utilization of recyclable resources through urban mining, the National Development and Reform Commission (NDRC) and the Ministry of Finance (MOF) initiated the UMPB in 2010. The core target of the UMPB policy is to promote the resource recycling industry, improve resource utilization efficiency, and transform the traditional linear economic growth model of “resource-product-waste” into a circular model of “resource-product-waste-renewable resources” to achieve sustainable development.

Panel data modeling [35] and case studies have been used to investigate the effects of environmental regulations and policies on solid wastes treatment in urban areas. The difference-in-difference-in-differences (DDD) framework [36,37], the regression discontinuity design (RDD) [11], the difference-in-differences (DID) method, and instrumental variables have been used to analyze the impact of governmental regulations on pollution reduction and environmental protection. The literature on solid waste generation and recycling reveal the application of widely varying modeling techniques, including regression modeling [38], time series analyses [39], system dynamics [40], spatial panel modeling [11,25], and input–output analyses.

Existing studies have concluded factors affecting energy efficiency, including government will [41,42], economies of scale [5], energy taxes [32], urbanization [5,43], industrial structure [5,42,43,44], level of economic development [5,42,44,45,46], and technological progress [32,42,43,44]. It has also been argued that the degree of openness to the outside world [44], the level of marketisation [42], foreign direct investment [42], and industrial agglomeration can also have an impact on urban energy efficiency [47]. Few studies have included UM as an impacting factor on energy efficiency. The existing literature on UM is limited to qualitative research, and few have studied the potential impact of UM on improving energy efficiency [23]. Based on the assumption that UM can improve urban energy efficiency via reducing energy consumption per unit of GDP, a time-varying DID model, which has been widely used to conduct policy evaluation, was adopted by treating the establishment of UMPBs as quasi-natural experiments to empirically analyze their impact on energy efficiency, and we expect UMPBs can provide a new pathway to help achieve urban sustainability and carbon neutrality. To the best of our knowledge, this study is the first to discover the impact of UMPB policy on fostering energy efficiency using a time-varying DID model. Propensity score matching (PSM) was adopted to avoid sample selection bias.

2. Materials and Methods

2.1. Sample and Data Selection

Based on the panel data from 2003–2016 of 281 prefecture-level cities across China, this paper sets the years 2003–2009 as the pre-period, and the years 2010–2016 as the post-period of the quasi-natural experiment. From 2010–2015, the NDRC and MOF of China approved 49 UMPBs in 27 provinces to efficiently recover recyclable industrial solid waste formed by industrial production. The number and location of UMPBs in China varied from year to year between 2010 and 2015, and the economic development of UMPBs also varies. However, most UMPBs are concentrated in the east and along the southeast coast (Figure 1).

Therefore, 49 UMPB cities were set as the treated group, and the remaining 232 cities were set as the control group. Since the China Energy Statistical Yearbook only provides energy consumption data at the provincial level, the authors obtained city-level energy consumption data by fitting night light data and provincial energy consumption data. The gross regional product (GDP) and per capita GDP of cities were deflated using 2003 as the base period.

We measure energy efficiency at city level. At present, China Energy Statistical Yearbook only provides energy consumption data at provincial-level, the Chinese government only publishes energy consumption data of major cities, and the China City Statistical Yearbook only discloses annual data of consumption of electricity, gas, and liquefied petroleum gas of prefecture-level cities. Due to data constraints, we assumed the proportion of total energy consumption at city level is the same as that of the provincial level. Therefore, the authors obtained the city-level energy consumption data based on provincial energy consumption data. Following [48], the energy consumption units are uniformly converted into 10,000 tons of standard coal.

$$CE{I}_{it}=\frac{P{E}_{it}+P{G}_{it}+P{L}_{it}}{PE{E}_{it}}$$

(1)

$$CC{E}_{it}=\frac{C{E}_{it}+C{G}_{it}+C{L}_{it}}{CE{I}_{it}}$$

(2)

$$Unit{E}_{it}=\frac{CC{E}_{it}}{GD{P}_{it}}$$

(3)

where i standards for the city or province and t refers to the year in model (1), (2), and (3). In model (1), CEI_it represents the conversion coefficient of energy consumption of city i in year t; PE_it, PG_it, and PL_it represent the consumption of electricity, gas, and liquefied petroleum gas of province i in year t; and PEE_it represents the electricity consumption of the city i in year t. In model (2), CCE_it represents the total energy consumption of the city i in year t; CE_it, Cg_it, and Cl_it stand for the consumption of electricity, gas, and liquefied petroleum gas of city i in year t. In model (3), UnitE_it refers to energy consumption per unit GDP of city i in year t; GDP_it is the gross domestic product of city i in year t.

The Marketisation Index (we obtained the marketization index from the China Market Index Database (https://cmi.ssap.com.cn/instruction, accessed on 7 November 2022)) measures the process of marketisation in China’s provinces using a large amount of standardized data. Based on 17 secondary indices, the index consists of five primary indices, each reflecting a specific aspect of marketisation in China: government–market relationship; development of the non-state economy; development of product markets; development of factor markets; development of market intermediary organizations; and rule of law environment. Based on the provincial marketization index, according to [49], we calculated 3 primary indices, including government–market relationship, development of factor markets, and development of market intermediary organizations.

2.2. Variable Definition and Data Description

To measure the impact of UMPB on fostering energy efficiency in urban China, this paper uses urban energy consumption per unit of GDP as the dependent variable. Based on the existing literature that studies the factors affecting energy efficiency, the control variables selected are the per capita GDP of each city [44,45], secondary industry share [5], industrial structure [44], population density [43], and innovation capability [42] of each city (

Table 1. Dependent and independent variables.

Variables	Meaning	Calculation	Unit
lngdpenergy	urban energy consumption per unit of GDP	Logarithm of city energy consumption volume divided by the local GDP
pgdp	Per capita GDP of each city	Gross regional product deflated using 2003 as the base period	RMB 10,000
indgdp	Secondary industry as a share of total GDP of each city to measure energy demand	The total output value of the secondary industry divided by GDP value of each city	%
indstruc	Industrial structure of each city	Proportion of industrial output value above designated size in the total GDP value of each region	%
popudens	Population size for energy consumption demand at city level	Population size at the end of the year (10,000 persons) divided by the area of the jurisdiction (sq.km)	10,000 persons/sq.km
patent	Scientific research and innovation capabilities of each city	The number of invention patents in each city	piece

).

The descriptive statistics are presented in

Table 2. Descriptive statistics.

Variable	N	mean	sd	min	p50	max
lngdpenergy	3934	0.193	0.837	−1.807	0.152	4.137
pgdp	3934	8.407	0.738	6.067	8.411	10.673
indgdp	3934	0.295	0.672	−4.003	0.399	3.207
indstruc	3934	0.488	0.111	0.09	0.492	0.91
popudens	3934	5.707	0.886	1.681	5.825	7.889
patent	3934	6.069	1.828	0.693	5.951	11.535

Note: N stands for sample size, mean for average value, sd for standard deviation, min for minimum, max for maximum, and P50 for median.

. The mean value of energy consumption per unit of GDP per unit of GDP (lngdpenergy) was 0.193, the standard deviation was 0.837, the minimum value was −1.807, the maximum value was 4.137, and the median value was 0.152. This shows that there are significant differences in the energy consumption per unit of GDP among cities in the research time interval between 2003 and 2016.

2.3. Model Design

We evaluated the impact of the UMPBs on energy consumption per unit of GDP per unit of GDP of each experimental city using the DID method and set up the following model (1):

$$Y_{it}={\alpha }_0+{\beta }_1\ast \left(trea{t}_{it}\times pos{t}_{it}\right)+{\beta }_2{X}_{it}+{\gamma }_t+{\theta }_i+Cit{y}_i\times Yea{r}_t+{\epsilon }_{it}$$

(4)

In model (4), i and t represent city and year, respectively; Y is energy consumption per unit of GDP of each city; treat is a city grouping variable (a UMBP city has a value 1, if not it is 0); and post is a time grouping variable, with a value of 0 for 2003–2009 and 1 for 2010–2016. In contrast to the traditional DID model of a single period, this study adopts a multi-period DID, so the multiplication item $trea{t}_{it}\times pos{t}_{it}$ captures the influence of the UMPB in different periods on explained variables. X is the control variable group; ${\gamma }_t$ is the time fixed effect;${\theta }_i$ is the city fixed effect that does not change with time, and $Cit{y}_i\times Yea{r}_t$ is the individual time effect of each city, which aims to control the unobservable changes in each city over time. ${\epsilon }_{it}$ represents the random error term. This study mainly observed the cross-term coefficient of to estimate the impact of the pilot policy of setting up UMPB on the energy efficiency of cities. We used Stata version 16.0 to perform empirical analysis.

3. Results

3.1. Empirical Results

3.1.1. Trend Chart of Energy Consumption per Unit of GDP

By drawing a trend graph of the change in energy consumption per unit of GDP among UMPBs and non-UMPBs, this study found that from 2003–2016, energy consumption per unit of GDP in both groups showed a downward trend from an average 0.4 to −0.1. Before 2010, the energy consumption per unit of GDP of both groups exhibited a relatively similar downward trend, with non-UMPBs changes from 0.43 to 0.2 and UMPBs dropping from 0.4 to 0.09. In 2010, energy consumption per unit of GDP of UMPBs was slightly lower than that of non-UMPBs. From 2010–2013, the energy consumption per unit of GDP of both groups gradually converged, the gap narrowed from 0.11 to 0.1. However, since 2014, the energy consumption per unit of GDP of UMPBs has shown a faster downward trend, from 0.09 to −0.2, than that of non-UMPBs which dropped from 0.2 to −0.05. We assume the decrease in the energy consumption per unit of GDP of UMPBs after 2014 is triggered by UMPB policies. The inherent logical relationship between the UMPB and the reduction in energy consumption per unit of GDP is shown in Figure 2.

3.1.2. DID Model Regression Results

The dynamic impact of the UMPB on energy consumption per unit of GDP is presented in

Table 3. DID results.

	Average Effect		Dynamic Effect
	gdpenergy (a1)	gdpenergy (a2)	gdpenergy (a3)
treat × post	−0.367 ***	−0.163 ***
	(0.02)	(0.03)
treat × 2010			−0.0722 ***
			(0.02)
treat × 2011			−0.0978 ***
			(0.02)
treat × 2012			−0.115 ***
			(0.03)
treat × 2013			−0.166 ***
			(0.03)
treat × 2014			−0.196 ***
			(0.03)
treat × 2015			−0.239 ***
			(0.04)
treat × 2016			−0.273 ***
			(0.04)
_cons	0.224 ***	5.195 ***	5.155 ***
	(0.00)	(0.92)	(0.90)
Control	No	Yes	Yes
N	3934.00	3934.00	3934.00
R-sq	0.11	0.50	0.51

Note: The values in parentheses are standard errors: ***, **, and * indicate the significance at the 1%, 5%, and 10% levels, respectively. The following tables are the same.

. Whether control variables are introduced (a2) or not (a1), the implementation of the UMPB significantly reduces energy consumption per unit of GDP by 3.67% among UMPB cities. In the analysis of dynamic effects (a3), from 2010–2016, the UMPB policy played a significant role in lowering the energy consumption per unit of GDP of UMPBs, and the effect of the pilot policy showed an increasing trend, from –0.0722 in 2010 to –0.273 in 2016. It can be empirically verified that UMPB policy has a continuous and stable effect on reducing energy consumption per unit of GDP of UMPBs.

3.1.3. Analysis of Dynamic Effects

The assumption of parallel trends is an important precondition for evaluating the policy effects using the DID method. This study tests whether the energy consumption per unit of GDP maintained a relatively stable trend before the implementation of the UMPB policy. Figure 3 shows a parallel trend in the changes in the urban energy consumption per unit of GDP before the implementation of the UMPB policy. Because benchmark regression results only reflect the average impact of the UMPB policy on energy consumption per unit of GDP, they cannot reflect the differences in the policy at different time intervals. We adopted the event study approach (ESA) to ensure that this study meets the assumption of parallel trends. Model (5) empirically analyzes the dynamic effect of the UMPB policy on the energy consumption per unit of GDP.

$$Y_{it}={\beta }_0+{\mathsf{\Sigma }}_{t=2003}^{2016}{\beta }_{t}trea{t}_{it}\times {\gamma }_t+\lambda X+{\gamma }_t+{\mu }_i+{\eta }_j+{\epsilon }_{it }$$

(5)

In Model (5), β_t represents a series of estimated values for the study interval 2003–2016. The other variables are the same as those in Formula (1). This research used 2010 as the base year when the pilot policy of the UMPB started. As shown in Figure 3, after controlling for those characteristics of cities and provinces that do not change with time, coefficient β_t was not significant between 2003 and 2009, indicating that there was no significant difference in unit GDP energy efficiency between UMPBs and non-UMPBs before the implementation of the pilot policy. This proves that the precondition of the parallel trend assumption is guaranteed. The estimated coefficient β_t became significant from the fourth year after 2010 (2014), indicating that the UMPB policy has a lagging effect on the reduction in urban energy consumption, which may be caused by the relatively low productivity of China’s overall industry, a large proportion of high-energy consumption industries, and relatively backward technology and equipment of high-energy consumption industries, which leads to high energy consumption per unit of GDP. Some cities have not made significant investments in energy-saving and consumption-reducing technological transformation, and, moreover, there is a time lag between the promulgation and implementation of energy-saving facilities. In addition, the energy efficiency of major energy-consuming equipment, such as industrial machinery, power generators, and industrial boilers used in engineering construction, remains low. Further energy is expended as cities light up urban areas to enhance their image and develop night economies.

3.1.4. Eliminate the Impact of Other Policies

China’s “Eleventh 5-Year Plan” (from year 2006–2010) aims to obtain the target of “energy conservation and emission reduction” by reducing the energy consumption per unit of GDP by approximately 20% and decreasing the total discharge of major pollutants by 10%. Since then, several decisions to strengthen energy conservation have been made, and a series of policies and measures to promote energy conservation and emission reduction have been formulated. Energy saving and emission reduction refer to saving both material and energy resources and reducing waste discharge and environmental hazards. Energy saving and emission reduction are usually carried out simultaneously; to obtain the net impact of the UMPB policy on energy saving, other possible causes must be excluded. Among them, new investments in the central and western regions of China are mainly concentrated in energy and resource fields, such as power output, which heavily relies on high energy consumption. In 2014, the NDRC deployed wind and solar energy development in 12 western provinces of China (Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, Inner Mongolia, and Guangxi). In terms of policies, in 2009 the Ministry of Finance, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, and the NDRC launched a financial subsidy pilot policy to purchase energy-saving and new energy vehicles (NEVs) in 13 cities (Beijing, Shanghai, Chongqing, Changchun, Dalian, Hangzhou, Jinan, Wuhan, Shenzhen, Hefei, Changsha, Kunming and Nanchang). In 2010, seven other pilot cities (including Tianjin, Haikou, Zhengzhou, Xiamen, Suzhou, Tangshan and Guangzhou) were added. In 2013, six further cities were set up as the third batch of pilot cities (Shenyang, Hohhot, Baotou, Chengdu, Nantong, and Xiangyang) to profit from the subsidy. From 2011–2014, a total of 30 cities (Beijing, Shenzhen, Chongqing, Hangzhou, Changsha, Guiyang, Jilin, and Xinyu in 2011; in 2013, Shijiazhuang, Tangshan, Tieling, Qiqihar, Tongling, Nanping, Jingmen, Shaoguan, Dongguan and Tongchuan; in 2014, Tianjin, Linfen, Baotou, Xuzhou, Liaocheng, Hebi City, Meizhou City, Nanning City, Deyang City, Lanzhou City, Haidong City, and Urumqi City) were established as comprehensive demonstration cities for energy conservation and emissions reduction benefiting from financial support. In addition, China has introduced several policies to adjust the structure of energy consumption in provinces with major coal energy consumption including Hebei, Shanxi, Inner Mongolia, Jiangsu, Shandong, and Henan, which will also affect the net effect of UMPB policies on energy conservation.

To determine the net impact of the UMPB policy on energy saving, this study excludes some sample cities in the above-mentioned provinces and areas. From the estimated results in

Table 4. UMPB’s impact on energy consumption per unit of GDP (other energy policies removed).

	gdpenergy	gdpenergy
treat × post	−0.304 ***	−0.110 ***
	(0.04)	(0.03)
_cons	0.146 ***	5.829 ***
	(0.01)	(0.51)
control	No	Yes
N	1847	1847
R-sq	0.04	0.50

, after excluding other policies that may affect the net effect of the UMPB policy on energy conservation, the UMPB policy was still found to have a significant effect on reducing the energy consumption per unit of GDP. Therefore, the empirical results above are robust.

3.1.5. Instrumental Variable Method

Another hypothetical premise of DID is the randomness of sample selection. Before setting UMPBs, the government considers endogenous indicators such as the industrial base and recycling capacity of candidate cities. Therefore, potential influencing factors may interfere with DID estimations and affect the final regression results. To address this problem, this study used the instrumental variable (IV) method to overcome potential endogenous influence. The choice of IVs needs to be related to endogenous variables but not to random disturbance items. In related research on environmental economics, the air circulation coefficient is often used as an instrumental variable in the model [36,50]. Therefore, this study selected the air circulation coefficient as an IV to overcome the endogeneity problem. The reasons are as follows: First, a large amount of industrial and domestic solid wastes is produced in cities during production and consumption. Long-term accumulation of solid waste causes serious soil, air, and water pollution, and a small air circulation coefficient causes the accumulation of pollutants and increases their concentrations. Under an environmental performance evaluation system, the weight of green GDP in the promotion and evaluation systems of local officials continuously encourages them to control pollution in their jurisdictions. Therefore, when the air circulation coefficient is low, local governments will be more inclined to adopt strict environmental regulations, including the use of command-based and market-based environmental regulations, to restrict production enterprises from adopting cleaner production technologies and the harmless discharge of pollutants. Similarly, in cities with low air circulation coefficients, local governments are more inclined to apply for the establishment of the UMPB to improve resource recycling and energy efficiency.

Second, the air circulation coefficient is not restricted by other economic variables but is determined by the geographic location and meteorological conditions of each city, conforming therefore to the exogenous assumption. The air circulation coefficient used in this study uses wind speed information from the European weather forecast center (ERA) dataset with the longitude and latitude of the Chinese city, and the wind speed and boundary layer height are then multiplied by the air circulation coefficient. Logarithmic processing was performed after sampling the average of this interval. The IV analysis was divided into two stages. As shown in

Table 5. UMPB on energy consumption per unit of GDP: IV estimation.

	First Stage	Second Stage
	treat × post	gdpenergy
iv × post	0.124 ***
	(0.00)
treat × post		−0.223 ***
		(0.02)
_cons	−0.864 ***	5.080 ***
	(0.23)	(0.28)
control	yes	yes
N	3934.00	3934.00
R-sq	0.57	0.30

, in the first-stage regression process, the cross-term coefficient of the IV and the policy implementation time variable (post) are significant, indicating that the IV satisfies the correlation. In the second stage, the DID coefficient (post × treat) had a significant negative impact on energy consumption per unit of GDP, indicating that the use of the IV method to eliminate the endogeneity problem of the policy implementation group sample selection, and UMPB policy has significantly increased energy consumption per unit of GDP, which is consistent with the results obtained by the above regression models and verifies the robustness of the conclusions.

3.1.6. Verification of the Impact Mechanism

The above regression results verify UMPB’s impact on reducing energy consumption per unit of GDP; however, further analysis is needed to explore the mechanisms of internal impact. This study sought to verify these mechanisms from the perspectives of market development, government–market relationships, and factor abundance.

First, enterprises with high recycling technology in UMPBs are the main players in resource recycling. However, imperfect competition still exists among enterprises in UMPBs. A small number of enterprises apply for quotas that exceed their recycling capacity and damage the interests of other enterprises. Creating an approval system with sufficient market competitiveness under government control is an important prerequisite to ensuring that the pilot policy effectively increases energy consumption per unit of GDP.

Second, market dominance and appropriate government intervention can effectively improve energy efficiency. Local governments favor low-efficiency state-owned recycling enterprises. Fair competition is jeopardized among enterprises due to policy inclination resulting from investment promotion or the attraction of leading state-owned enterprises (SOEs) in the recycling industry. Therefore, the effective coordination of the relationship between the government and the market is also an important factor in ensuring the efficient operation of the UMPB.

Third, market imperfections and government overcontrol often cause factor market prices to deviate, leading to non-optimal allocation of production factors. Eliminating factor market distortions can improve energy efficiency. Waste recycling enterprises are generally small-scale and lack up to date technology. Although the state has issued a series of preferential policies to encourage and support their development, the vast majority of waste recycling enterprises are still unprofitable, and, constrained by finances, have no ability to adopt new technologies. Designing tax incentives and fiscal subsidy policies that match the UMPB policy will better foster energy consumption per unit of GDP.

Based on the above derivation, this study assumes that UMPB’s impact on reducing energy consumption per unit of GDP is channeled by mediators, such as the level of marketization, government-market relationship, and the degree of factor market development. To test the impact mechanism of these mediators, we designed the following model:

$$\begin{gathered} Y_{it}={\alpha }_0+{\beta }_1\ast \left(trea{t}_{it}\times pos{t}_{it}\times mediato{r}_{it}\right)+{\beta }_2\left(trea{t}_{it}\times pos{t}_{it}\right)+{\beta }_{3}Moderato{r}_{it}+{\beta }_4{X}_{it}+{\gamma }_t+{\theta }_i \\ +Cit{y}_j\times Yea{r}_t+{\epsilon }_{it} \end{gathered}$$

(6)

In Formula (6), the explained variable Y is the energy efficiency per unit GDP of each city, and the mediators refer to the adjustment variables, including the marketization level index (market), government–market relationship harmony index (govern), and factor market index (gactor). This part mainly focuses on the coefficient of the interaction term coefficient β1, while the other variables in Formula (6) are consistent with Formula (4).

Table 6. Verification of the impact mechanism.

	gdpenergy
	Market	Govern	Gactor
treat × post × mediator	−0.221 ***	−0.232 ***	−0.229 ***
	(0.02)	(0.02)	(0.02)
Moderator	−1.159 ***	−0.406 ***	−0.581 ***
	(0.05)	(0.06)	(0.04)
_cons	−1.066 ***	−2.120 ***	−2.270 ***
	(0.15)	(0.17)	(0.14)
control	yes	yes	yes
N	3934	3934	3934
R-sq	0.36	0.29	0.33

shows that UMPBs are affected by the degree of market development of each city in reducing energy consumption per unit of GDP. Specifically, the coefficients of the crossover terms such as marketization level, the governments–market relationship, and factor market abundance degree significantly reduce energy consumption per unit of GDP.

3.1.7. Analysis of Heterogeneity: Regional Differences

Energy consumption structure in China is quite different among regions, and the northeast and midwest areas have been particularly inefficient for a long time. Following the official regional division, this study identifies the differential impact of the UMPB on the energy efficiency per unit of GDP among regions.

Table 7. Regional heterogeneity test.

	East	Middle	Northeast	West	East	Middle	Northeast	West
	gdpenergy				gdpenergy
treat × post	−0.2516 ***	−0.2606 ***	−0.2745 ***	−0.2207 ***	−0.3348 ***	−0.4182 ***	−0.4658 ***	−0.1940 **
	(0.03)	(0.05)	(0.10)	(0.07)	(0.02)	(0.04)	(0.12)	(0.08)
_cons	6.4638 ***	13.3618 ***	12.6450 ***	11.86	−0.0600 ***	0.1720 ***	0.5881 ***	0.4973 ***
	(1.69)	(2.35)	(2.61)	(8.57)	(0.00)	(0.00)	(0.00)	(0.00)
control	yes	yes	yes	yes	no	no	no	no
N	1400.00	1246.00	812.00	476.00	1400.00	1246.00	812.00	476.00
R-sq	0.34	0.44	0.36	0.20	0.14	0.09	0.09	0.02

indicates that the UMPB has a significant impact on the reduction in energy consumption per unit of GDP in all four regions.

Specifically, the UMPB policy has the greatest effect on reducing energy consumption per unit of GDP in the northeast area, followed by the central, eastern, and western regions. The reason for these differences may be that the northeast is an old industrial base, and the original characteristics of high energy consumption and high pollution have long restricted the improvement of its energy efficiency. Being an old industrial base has inherent advantages in terms of integrating the heavy technical equipment required by the recycling industry. Under the pressure of the national energy conservation and emission reduction policy, the old industrial base has begun to change its own energy consumption structure. The demand for industrial raw materials has gradually shifted to secondary resources obtained from recycling rather than from virgin mining.

The central region had a sound manufacturing foundation. Although most of the manufacturing industry in central China is still at the low end of the value chain, the UMPB’s impact on the energy consumption per unit of GDP reduction in the central region is slightly higher than that in the eastern region.

The eastern coastal areas have fewer mineral resources than those in the northeast, central, and western regions. Rapid industrial development requires a large quantity of resources, forcing the eastern region to shift its demand for raw materials to secondary resources earlier than other regions. Before 2000, the eastern coastal areas obtained relatively cheap raw materials for industrial development by importing waste materials, such that industrial products in the eastern area had a greater competitive advantage in exports. Compared with other regions, the energy consumption structure of the eastern region is less dependent on coal; therefore, the UMPB policy has a smaller impact on energy consumption per unit of GDP than the central and northeastern regions.

As the per capita GDP in western China is relatively small, the energy consumption per unit of GDP is larger than that of the other regions. Owing to the relatively lagging industrial development, weaker environmental regulations, and low entry barriers for industries with high energy consumption and high emissions, energy consumption per unit of GDP remains high in the western region, thus offsetting the reduction in energy consumption per unit of GDP in the western areas.

4. Discussion

This study analyzed the impact of UMPB from the perspective of fostering energy efficiency, explored the internal mechanism of marketization in reducing energy consumption per unit of GDP, and conducted an in-depth exploration of the heterogeneity of each region. From 2010–2015, UMPBs in China mainly treated recyclable solid waste generated by domestic economic activities and solid waste imported from overseas. However, since 2017, to further implement green development, China drastically reduced the total amount of solid waste imports, and completely banned it in 2020. Since then, more domestic solid waste that can be recycled and reused has been recycled in UMPBs in various provinces. Therefore, it is of great strategic significance to evaluate the role of UMPBs rationally and scientifically in reducing the urban energy consumption per unit of GDP and to explore reasonable methods to better achieve green development.

As a basic national policy of China, the primary purpose of resource conservation is to improve the efficiency of resource utilization and change the mode of economic development [41]. According to this development strategy, China has quantified the target of reducing energy consumption per unit GDP, written it into the national economic development plans, and adopted a series of economic, legal, and necessary administrative measures to achieve the target, such as the establishment of UMPBs, and use it as a binding indicator for the performance evaluation of governments at all levels to give consideration to economic development and ecological protection, and conduct regular assessment every year. The proposal of energy consumption per unit GDP as a binding assessment indicator is an important means to urge all regions and industries to pay more attention to improving the quality and efficiency of economic development, as well as scientific and sustainable development [42]. A correct understanding of the energy consumption index per unit GDP can not only help us understand the situation of energy conservation and consumption reduction, but also provide an important reference for analyzing and studying the quality of regional economic development. The significant negative relationship between UMPBs and the energy consumption per unit of GDP indicates that UMPBs has an actual effect on promoting energy efficiency. Both static and dynamic effects of UMPBs on energy consumption per unit of GDP support the results that UMPBs can reduce energy consumption per unit of GDP and improve energy efficiency.

The possible explanation is three-fold. First, with the goal of achieving carbon peaking and carbon neutrality, China has accelerated the adjustment toward energy structure optimization [41]. As the main carrier for industrial development, industrial cities in China require large amounts of industrial raw materials, leading to boosting demand for exploiting virgin mining, which may lead to high energy consumption and high pollution [42]. The establishment of UMPBs can effectively recycle ISW, release the dependence of urban industry on virgin mining and reduce energy consumption while achieving economic growth. Therefore, the deployment of UMPBs in various provinces in China has achieved the objective of saving resources and optimizing the ecological environment in line with the needs of China’s green economic development. Second, ageing industrial and residential electrical equipment in urban areas are causing excessive energy consumption due to relatively low energy efficiency. To further promote the renewal of old appliances with high energy consumption in urban area, the government has promoted a policy of “exchanging old appliances for new ones”, through which UMPBs are used to recycle electrical equipment that has been phased out in Chinese cities due to energy inefficiency. This promotes the use of new appliances with higher energy efficiency standards, thereby reducing urban energy consumption. Third, according to the dynamic effect regression result, the impact of UMPBs on reducing energy consumption per unit GDP increased annually. This finding is consistent with [42] that the concept of circular economy is becoming more widespread in China with the gradual implementation of circular economy policies.

The results of this study are consistent with the findings of previous studies from the perspective of urban mining’s roles in easing resource supply constraints and environmental sustainability [23,51,52,53]. Unlike existing studies, we attempted to explore a new path to improve energy consumption per unit of GDP in urban areas from the perspective of UMPBs. In addition, this study provides an assessment of the UMPB policy implemented in China since 2010, and to provide policy support to promote establishment of UMPBs in other cities in China. Additionally, this study provides a new experience for cities in developing countries to promote energy conservation and emission reduction, enhance energy efficiency, and promote green development via the establishment of UMPBs.

5. Conclusions and Policy Implications

This study considered 281 prefecture-level cities as samples between 2003 and 2016 and applied the DID model to study the impact of the UMPB policy on urban energy consumption per unit of GDP. The main conclusions are as follows:

(1)

The UMPB policy can significantly reduce energy consumption per unit of GDP by 3.67% on average. Based on the parallel trend test, IV method, and removal of other energy policy influences that may affect the net effect of the UMPB on energy consumption per unit of GDP, the research conclusions are still valid, indicating that the research results are robust.

(2)

In the process of exploring the influence channels of UMPB on energy consumption per unit of GDP, this study found that the degree of marketization, the level of the local government–market relationship, and the development and abundance of factor markets are mediators in reducing energy consumption per unit of GDP.

(3)

Through the heterogeneity analysis of UMPB’s different impacts on energy consumption per unit of GDP in different regions, this study finds that UMPB have reduced the energy consumption per unit of GDP of each city as a whole; however, regional differences are obvious. The UMPB’s impact on reducing unit GDP energy consumption weakened from the northeast (−0.2745%) to the central (−0.2606%), eastern (−0.2516%), and western (−0.2207%) areas.

Accordingly, the policy implications to promote the UMPB’s impact on fostering energy efficiency are as follows:

(1)

Build a top-level design for a green financial policy system involving governments, financial institutions, and environmental protection companies. Use the national recycling network and the recycling system spontaneously formed by society, reduce recycling costs, standardize recycling behavior, and improve recycling efficiency. Actively build a solid waste electronic trading platform, organically connect the advanced production lines of each UMPB with the source of solid waste, set up recycling and remanufacturing product standards, clarify the market access conditions for recycled products, link the in-depth cooperation between recycling companies and raw material processing companies, and form a closed industrial chain loop.

(2)

Design the intensity of environmental regulations in a more precise and scientific manner. Although the current environmental regulations in the central and western regions are relatively low, local governments should not blindly increase them or build high investment entry thresholds in accordance with their own development conditions when formulating environmental regulations. On one hand, environmental regulations that are inconsistent with the local reality will increase the burden on local enterprises and inhibit enterprises’ green innovation. On the other hand, high energy pollution enterprises will reduce investment in the western region, which will affect its economic development. The central and western regions can adopt a combination of command-based environmental regulatory tools and incentive-based environmental regulatory tools to steadily promote energy conservation to achieve green development.

Reducing the energy consumption per unit of GDP in urban areas and increasing energy efficiency are two important starting points for China’s energy conservation, emission reduction, and industrial green development. Through the construction of the UMPB, the renewable resource utilization capacity and recycling level of each city can be improved, and the dual constraints of resource shortage and ecological safety can be relieved.

However, our study is not without limitations. The China Energy Statistical Yearbook only publishes energy consumption data at the provincial level and for major cities. The China Urban Statistical Yearbook includes annual data on consumption of electricity, gas, and liquefied petroleum gas but does not contain total energy consumption data. The data of China Market Index Database does not include municipal marketization index data. Some of the secondary indices are no longer provided officially by the government after 2016. For example, the proportion of financial institutions’ loans to non-state-owned enterprises reflects the degree of marketization of credit fund distribution, which is an important indicator of the marketization of the financial industry (a secondary variable measuring the abundance of factor markets), but it cannot be obtained from statistical yearbooks after 2016. Due to data limitations, we set the study period from 2003–2016. We plan to follow up on this in future studies and look for more new evidence of UMPB’s impact on energy efficiency in China as soon as new data becomes available. In addition, the mechanism of UMPB’s impact on reducing energy consumption per unit of GDP of each city should be further explored.