Can the Environmental Protection Tax Promote the Improvement of Energy Efficiency? Evidence from Prefecture-Level City Data in China

Abstract

The environmental protection tax (EPT) implemented in 2018 is a significant measure in China’s tax system reform, providing a valuable opportunity to encourage green development, promote resource conservation, and advance sustainable growth. This study, based on panel data from Chinese prefecture-level cities between 2006 and 2021, uses the 2018 EPT as a quasi-natural experiment and applies the difference-in-differences (DID) model to empirically examine the impact of the EPT on urban energy efficiency. The results show that the EPT effectively enhances urban energy efficiency. Strengthening environmental law enforcement and promoting technological innovation are identified as key pathways to improving energy efficiency. Heterogeneity analysis reveals that the reform has a more significant impact on energy efficiency in southern cities compared to northern cities, and the effect is more pronounced in large cities than in smaller ones. This study enriches the existing literature on EPT and the application of institutional theory, providing empirical evidence for the effectiveness of the reform, highlighting the importance of enhancing local government environmental enforcement, and promoting technological innovation in improving energy efficiency. It offers valuable theoretical and practical guidance for policymakers, contributing to the low-carbon transition and the achievement of sustainable economic growth.

1. Introduction

Over the past 40 years of reform and opening up, China’s economic development has achieved remarkable success. However, the rapid economic growth model has not only increased the burden on the ecological environment but has also resulted in China’s energy efficiency being higher than that of other economies worldwide. According to statistics, from 2005 to 2022, China’s energy intensity (primary energy consumption per unit of GDP) decreased from 2.44 kWh/USD to 1.65 kWh/USD, while during the same period, the energy intensity of the United States decreased from 1.83 kWh/USD to 1.36 kWh/USD, and the world average dropped from 1.58 kWh/USD in 2010 to 1.30 kWh/USD in 2022 (2011, PPP) [1,2]. This situation indicates that although China’s energy efficiency has improved significantly over the past few years, there remains considerable room for improvement compared to the United States and the world average. Excessive energy consumption has imposed a heavy burden on both the environment and the economy. Low energy efficiency has resulted in resource waste and environmental degradation, reducing the overall operational efficiency of the economy. As a result, improving energy efficiency through policy reforms and promoting sustainable development has become a key concern for the government.

Energy efficiency is a concept frequently discussed in both academic and practical contexts, yet its definition lacks uniformity, partly due to its complexity and the varying perspectives across different disciplines. Broadly, energy efficiency is often defined as “using less energy to produce the same number of services or useful output” [3,4]. This definition emphasizes achieving the same or higher levels of output while reducing energy consumption, thereby enhancing resource utilization. Specifically, energy efficiency encompasses all stages of energy use, from extraction and transmission to final consumption. Improvements in energy efficiency are regarded as a critical strategy for addressing climate change [5]. By reducing energy consumption, particularly through the decreased use of fossil fuels, enhanced energy efficiency can significantly lower greenhouse gas emissions, thus mitigating the threat of global warming. This reduction benefits environmental protection and contributes to the achievement of sustainable development goals [6,7]. Therefore, energy efficiency is not merely an issue within the realms of economics or engineering; it also involves interdisciplinary discussions spanning environmental science and policy research. In recent years, research on energy efficiency has primarily explored factors affecting energy efficiency, energy efficiency gaps, measures to improve energy efficiency, and related policies [8,9,10,11]. Regarding factors affecting energy efficiency, Wu et al. found that the development of the internet not only improves local green total-factor energy efficiency but also enhances the green total-factor energy efficiency in neighboring regions [12]. Issam et al. investigated the impact of various CSR (Corporate Social Responsibility) dimensions on energy efficiency in SMEs, discovering a strong correlation between the activity levels of social, environmental, and economic CSR dimensions and energy efficiency, with the environmental CSR dimension contributing the most to improving energy efficiency [13]. Hori et al. reached similar conclusions, suggesting that policies encouraging companies to improve CSR can effectively promote energy-saving activities [14]. Mette and Lene, in their empirical literature review on the driving factors of energy efficiency in manufacturing enterprises, pointed out that organizational and economic drivers are the main forces behind energy efficiency improvements, while they believe that policy tools and market drivers are relatively less important [9]. In contrast, Wu et al. argue that effective policies are crucial for improving energy efficiency and economic development [15]. Many scholars have studied the impact of relevant policies on energy efficiency. Yu and Zhang found that the low-carbon city pilot policy significantly improved carbon emission efficiency [16]. Song et al. studied the impact of green credit policies on energy-efficient utilization and found that regional energy-efficient utilization in China exhibits significant spatial effects, with green credit, environmental regulation, technological progress, and industrial structure having some impact on regional energy-efficient utilization, though the effect of credit scale was not significant [17]. However, there has been little research on the impact of the environmental protection tax (EPT) on energy utilization efficiency.

As a market-incentive environmental regulation, EPT aims to internalize the external costs of pollution into the production and operation costs of enterprises by assigning monetary value to the negative environmental impacts caused by production and consumption activities, allocating resources through market mechanisms. Unlike the pollution fee system, which is less operational and inconsistently enforced, EPT has established a more structural and legally binding framework. Research on EPT began relatively early in foreign countries. Since the 1980s, many developed nations have introduced environmental and green taxes to address industrial and corporate-level environmental challenges. Studies have shown that in countries such as Japan, Norway, and Finland, EPT policies have had significant impacts on industrial emission behaviors, encouraging companies to take more proactive measures to reduce pollutant emissions [18,19,20]. Notably, member states of the Organization for Economic Co-operation and Development (OECD) adopted the “polluter pays” principle as early as 1972, which mandates that polluters bear the cost of the pollution they cause, whether directly or indirectly [21]. This principle has driven the development of stricter emission controls and environmental protection policies. In contrast, research and implementation of EPT in China started relatively late and have gone through several stages of development and evolution. Since 1979, when the Standing Committee of the National People’s Congress passed the “Environmental Protection Law of the People’s Republic of China (Trial)”, which first introduced the pollutant discharge fee system, China has begun imposing economic penalties on pollutant emissions. The aim was to reduce pollution through economic means. In 1982, the State Council issued the “Interim Measures for the Collection of Pollutant Discharge Fees”, officially launching the implementation of the pollutant discharge fee system. In 2003, the “Regulations on the Collection and Use of Pollutant Discharge Fees” were enacted, detailing the objects, standards, usage, and management of pollutant discharge fees, further strengthening the effectiveness of the system. As environmental protection concepts deepened and legal systems improved, and to align with international practices, China gradually transitioned from the pollutant discharge fee system to an EPT system to further enhance the strength and effectiveness of environmental governance. The EPT, promulgated in 2016 and officially implemented on 1 January 2018, was born in this context. The EPT is levied on producers that discharge pollutants directly into the environment. The taxable pollutants include four categories: air pollutants, water pollutants, solid waste, and noise. Compared with the pollutant discharge fee system, the EPT has the characteristics of increasing the tax reduction and exemption thresholds, further standardizing the collection management procedures, and raising the upper limit of tax collection. The formal implementation of the EPT marked the end of the pollutant discharge fee system and the comprehensive launch of the EPT system. This shift aimed to further reduce pollutant emissions through more systematic and scientific taxation methods, promoting environmental protection and sustainable development.

Previous scholars have conducted multi-faceted studies on the impact of the EPT. At the enterprise level, research on corporate environmental performance and illegal emissions has indicated that the implementation of the environmental tax has spurred corporate environmental responsibility, subjecting polluting companies to stricter environmental controls, thereby increasing their environmental costs [22,23,24,25]. Regarding corporate economic performance, scholars generally believe that the policy significantly enhances corporate performance [26,27,28,29]. Moreover, some studies have found that since environmental taxes are considered local revenue, environmental corruption issues may intensify, leading to illegal emissions [22,30]. Some scholars innovatively used the synthetic control method to explore the impact of the EPT on corporate green total factor productivity (GTFP). GTFP is a measure of productivity that integrates environmental and resource considerations into traditional total factor productivity (TFP) assessments. Unlike conventional TFP, which focuses solely on economic outputs relative to inputs, GTFP accounts for environmental impacts, resource depletion, and pollution. This approach reflects a more comprehensive evaluation of productivity by balancing economic growth with ecological sustainability, making it consistent with sustainable development principles. Numerous studies finding that the implementation of the EPT significantly improves total factor productivity [31,32,33]. In terms of enterprise capacity reduction, some scholars found that its implementation can significantly solve the problem of overcapacity and improve overall operational efficiency [34,35]. At the urban level, existing studies have focused on the impact of environmental pollution. Some scholars used the DID method to take 2016 as the policy impact point, studied the impact of EPT on urban energy efficiency, and explored the role of green technology innovation and industrial structure upgrading [24]. Some scholars have innovatively used the regime-switching threshold regression (RSTR) threshold model to analyze the impact of the EPT on ecological footprints [36,37]. The threshold model is commonly used in economics, finance, and econometrics to study relationships between variables under different states. The RSTR model combines the threshold model and the Markov-switching model by introducing a threshold variable to distinguish between different states (or regimes), allowing the model to have different regression coefficients in each state. This type of model helps in understanding the nonlinear characteristics of relationships between variables and reflects the heterogeneity of the system under varying conditions. These studies collectively reveal the far-reaching impact of the EPT at different levels, providing an important theoretical foundation for subsequent research. While existing literature has covered many aspects of the impact of the EPT at the corporate and urban levels, much research has primarily focused on the corporate level, lacking a systematic analysis of the comprehensive effects on urban energy efficiency and environmental governance.

In China’s economic development, energy efficiency has always been a key factor influencing sustainable development [38]. Improving energy efficiency can not only significantly reduce energy consumption and lower corporate production costs but also reduce pollutant emissions, helping to alleviate environmental pressures [39]. However, despite the government adopting various measures over the past few decades to improve energy efficiency, there are still many challenges to improving energy efficiency at the prefecture-level city level, especially in regions concentrated with heavy industries. Investigating whether the reform of EPT can improve the energy efficiency of prefecture-level cities by reducing pollution and encouraging enterprises to adopt more energy-efficient technologies has become a research topic of great practical significance.

It is noteworthy that although the EPT was implemented simultaneously in all provinces and cities in 2018, there were significant differences in tax rates among cities. In 12 provinces, including Hebei, Jiangsu, and Shandong, the EPT rates were significantly higher than those of the pollutant discharge fee system. In contrast, in provinces and cities such as Hubei, Zhejiang, and Fujian, the EPT rates were largely equivalent to the pollutant discharge fee rates. Cities in provinces where tax rates increased may experience significant changes in aspects such as energy efficiency due to the implementation of the EPT, while other cities, where the EPT rates remained consistent with the discharge fee rates, might be less affected. Based on previous studies, many scholars have used the difference-in-differences (DID) method to examine the impact of the EPT on corporate environmental performance, financial performance, and illegal pollution emissions [22,28,40]. Additionally, some researchers have employed the SCM to investigate the effect of EPT on green total productivity [33]. While SCM offers a more refined construction of control groups, it requires high-quality data and is more challenging to apply in situations involving multiple policy interventions. Therefore, based on panel data from 282 prefecture-level cities from 2006 to 2021, the implementation of the EPT is regarded as a quasi-natural experiment, with cities that experienced tax rate increases serving as the treatment group and those with unchanged rates as the control group. The DID method is used to systematically analyze the impact of the EPT on urban energy efficiency. The results show that the policy significantly improves urban energy efficiency, and after conducting robustness tests such as parallel trend tests, placebo tests, and excluding the interference of other policies, the results remain significant. Mechanism tests reveal that the policy enhances urban energy efficiency by strengthening environmental law enforcement and promoting technological innovation. Analysis of the analysis shows that the policy’s effect on improving energy efficiency is more pronounced in southern cities than in northern cities, and its impact is more significant in large cities compared to small ones.

Compared to existing studies, this paper makes several marginal contributions. First, it expands the scope of research on the reform of EPT: previous research mainly focused on the impact of the reform on environmental pollution control and corporate performance, with limited attention to its effect on energy efficiency. This study introduces the reform of EPT into the field of energy efficiency research, systematically analyzing its role in improving energy efficiency at the urban level, filling the gap in research on the relationship between environmental policies and energy efficiency, and broadening the scope of research on the effects of related policies. Second, it reveals the internal mechanisms of energy efficiency improvement: through mechanism tests, this paper further explores the pathways through which the reform of EPT affects urban energy efficiency. The study finds that the implementation of the policy significantly improves urban energy efficiency by strengthening environmental law enforcement and promoting technological innovation, providing theoretical support for understanding the policy’s implementation mechanisms and helping explain how environmental policies improve energy efficiency through enforcement and innovation. Third, it reveals regional and scale differences in policy effects: through heterogeneity analysis, this paper reveals the differentiated impact of the reform across regions and city sizes. The study shows that the policy’s effect in southern cities is significantly higher than in northern cities, and its impact on improving energy efficiency is stronger in large cities than in small ones. These findings on regional and city-size differences provide new practical evidence for policymakers to implement targeted measures in different regions and types.

The rest of this paper is arranged as follows: Section 2 proposes research hypotheses through theoretical analysis; Section 3 introduces the research design; Section 4 presents empirical results and analysis; Section 5 provides a mechanism analysis; Section 6 conducts further heterogeneity analysis; Section 7 discusses the similarities and differences between the results of this study and other studies and Section 8 concludes and offers relevant suggestions.

2. Theoretical Analysis and Research Hypotheses

2.1. The Impact of the EPT on Energy Efficiency

From the perspective of environmental regulation theory, the implementation of the EPT exerts economic pressure on enterprises, prompting them to seek higher energy efficiency in their production processes. As energy demand continues to increase, the environmental costs faced by enterprises also rise. To reduce these costs, companies need to optimize their energy structure. In high-pollution industries, especially those reliant on high-pollution energy sources like coal, the policy’s implementation may lead to a reduction in their dependence on such energy sources. This not only helps reduce energy consumption but also effectively decreases pollutant emissions [41,42]. Therefore, while improving corporate environmental performance, the reform also provides strong support for enhancing overall energy efficiency in prefecture-level cities.

From the perspective of incentive theory, the cost pressure brought about by the policy encourages companies to take measures to reduce energy waste. Faced with rising environmental costs, companies are forced to pay more attention to managing their production processes. By optimizing production processes and improving resource utilization efficiency, companies can lower operational costs [30,43]. This shift not only enhances corporate economic benefits but also contributes to environmental benefits on a societal level. In responding to environmental regulations, companies often actively explore ways to improve resource use efficiency and reduce energy consumption, further driving energy efficiency improvements.

Based on the above theoretical analysis, the following hypothesis is proposed:

Hypothesis 1 (H1). 

The implementation of the EPT will significantly enhance the energy efficiency of prefecture-level cities.

2.2. Mechanism Analysis of the Impact of EPT on Energy Efficiency

In the field of environmental protection, “effective environmental enforcement” refers to actions taken by the government and relevant regulatory agencies to ensure that businesses and individuals comply with environmental regulations, reduce pollution, and conserve resources. This enforcement is facilitated through comprehensive policy tools, oversight mechanisms, and penalty structures [41,44]. It encompasses not only inspections and penalties but also education, guidance, and policy formulation to create a fully compliant environment. Enforcement serves as a safeguard for policy efficacy, particularly under the EPT framework, where the increase in non-compliance costs drives companies to enhance adherence, thereby achieving the policy’s intended outcomes. Policy enforcement effectiveness refers to the degree to which a policy is implemented and achieves its goals. Within the EPT framework, enforcement effectiveness is exhibited through strengthened regulatory rigor. The EPT, as a “green tax system”, aims to guide businesses toward energy conservation and emissions reduction through economic incentives [35,45]. Supported by effective enforcement, policy execution gains strength because companies not only pay environmental taxes but must also meet stricter environmental standards or face elevated compliance costs and legal risks. Compliance costs are the expenses companies incur to meet regulatory and policy requirements, including investments in new technology, process improvements, and employee training [35]. Legal risk, on the other hand, is the potential for legal penalties or reputational harm if companies fail to meet standards. By enhancing enforcement, companies face increased compliance costs and legal risks, prompting them to more carefully assess their energy usage. To avoid significant fines or legal repercussions, companies often opt for more cost-effective production methods, such as investing in clean technology and efficient equipment, which can reduce future compliance expenses [41,46]. According to incentive theory, companies respond to external threats by actively pursuing innovation to optimize processes and improve energy efficiency, thereby reducing operating costs in a high-risk environment. Incentive theory suggests that companies facing external threats or potential rewards are internally motivated to adapt to external conditions [28,47]. The pressure from environmental enforcement acts as a form of “negative incentive”, pushing companies to scrutinize and optimize their production processes to avoid adverse consequences from non-compliance.

The Porter Hypothesis, proposed by economist Michael Porter, posits that appropriate environmental regulations can stimulate corporate innovation, which does not weaken, but rather enhances, a firm’s productivity and competitiveness [36,40,48]. The core of the Porter Hypothesis is that environmental regulations provide an “external pressure” that encourages companies to innovate in ways that both meet environmental requirements and achieve economic gains, resulting in a win–win for economic and environmental benefits [30,35]. EPT increases compliance costs, which for businesses translates into higher production costs. This compels firms to reevaluate and adjust their production processes, energy usage, and waste management practices to reduce tax burdens. Technological innovation, which involves advancements in products, processes, or management practices to improve productivity, reduce costs, or enhance competitiveness, plays a critical role in boosting energy efficiency. Studies show that technological innovation can significantly reduce pollution emissions and energy consumption during production [28,44,47]. By introducing advanced production equipment, companies can lower energy use; by improving production processes, they can utilize raw materials more effectively, thereby reducing resource waste. Clean technology, a form of technology that minimizes pollutant generation or promotes waste recycling, and renewable energy investments, which include solar and wind energy (as opposed to traditional, high-pollution energy sources), also contribute to this shift. EPT incentivizes investment in clean technology and renewable energy through its tax-based leverage, reducing reliance on traditional fossil fuels [49]. This incentive mechanism accelerates the transition to green technologies as companies adopt low-pollution, energy-efficient production methods, thereby improving overall energy efficiency. The existence of the tax encourages companies to consider energy-saving and emission-reduction measures in their production cost calculations, progressively enhancing energy efficiency by promoting investments in clean technology, optimizing production processes, and improving resource management strategies. Policy pressure spurs companies to increase innovation efforts by introducing new equipment, refining processes, and enhancing management practices [48]. Innovation gradually becomes integral to corporate strategy, making energy conservation and efficiency an essential component of long-term competitiveness [42]. The impact of technological innovation extends beyond the short term and has lasting effects on a company’s future development. As companies’ environmental awareness and social responsibility increase, technological innovation gradually becomes a core approach to enhancing energy efficiency, supporting the low-carbon transformation of urban areas [23]. Thus, through technological innovation, companies can achieve the production goals of “high efficiency, low consumption”, accomplishing both economic and environmental benefits. Therefore, the following hypothesis is proposed:

Hypothesis 2 (H2). 

The implementation of the EPT can enhance urban energy efficiency by strengthening environmental enforcement.

Hypothesis 3 (H3). 

The implementation of the EPT can improve urban energy efficiency by promoting technological innovation.

3. Research Design

3.1. Sample Selection and Data Sources

The Pollution Charge System in China was officially implemented on 1 July 2003. There are significant differences between cities in terms of industrial structure, regulatory enforcement, and economic development. It also provides a rich panel data set that allows for more precise assessment of policy impacts at different regional and city scales. To explore the comparative impact of the EPT system versus the pollution charge system, and based on the availability of data, this study selects 282 prefecture-level cities from 2006 to 2021 as the research subjects, after excluding years and cities with significant data gaps. Using the EPT, implemented in 2018, as a quasi-natural experiment, the study applies the difference-in-differences (DID) method to investigate the effect of the transition from pollution charges to taxes on urban energy efficiency. The data for environmental penalties at the prefecture level, used for mechanism analysis, were obtained from the PKULaw (Chinalawinfo) database. Other data were sourced from the China City Statistical Yearbook and the EPS database. Due to data availability issues, cities with significant missing data were excluded, resulting in a final dataset of 4512 observations from 282 cities for the period 2006–2021. The original data table of representative prefecture-level cities is shown in

Table A1. Original data table of representative prefecture-level cities.

City	Year	Treat_Time	SBM	Pop_Size	gdp	Openness	Urban_Rate	Industry	Market_Level
Shanghai	2006	0	0.267993336	7.221163576	10.96292579	0.054652406	0.887	0.5059	0.787742091
Shanghai	2007	0	0.301919206	7.22901235	11.10295522	0.049405959	0.887	0.5258	0.72580733
Shanghai	2008	0	0.328435243	7.237806948	11.19991191	0.051128264	0.886	0.5366	0.794443561
Shanghai	2009	0	0.368200081	7.244727391	11.27706388	0.047843491	0.886	0.5936	0.835396072
Shanghai	2010	0	0.390631919	7.252989021	11.23946183	0.04385798	0.8927	0.5728	0.873433451
Shanghai	2011	0	0.376833311	7.257989526	11.32128058	0.042397242	0.893	0.5805	0.738465575
Shanghai	2012	0	0.382038192	7.263259538	11.35478517	0.047494637	0.8931	0.6045	0.708782257
Shanghai	2013	0	0.384973197	7.267036823	11.92406113	0.048106063	0.896	0.6224	0.691192569
Shanghai	2014	0	0.394740978	7.271495207	11.48627343	0.047348564	0.8957	0.6482	0.690946116
Shanghai	2015	0	0.415786588	7.274458769	11.55018271	0.045762611	0.8761	0.6776	0.811869816
Shanghai	2016	0	0.462992385	7.279318835	11.6661786	0.04364087	0.8789	0.6978	1.0243453
Shanghai	2017	0	0.495637572	7.28276118	11.74222602	0.037738732	0.8771	0.6918	1.129827347
Shanghai	2018	0	0.536457658	7.28756064	11.81289672	0.035031233	0.88	0.699	1.118615215
Shanghai	2019	0	0.52099918	7.292337176	11.96577658	0.034438098	0.881	0.7273	1.116430029
Shanghai	2020	0	0.592949915	7.300472814	11.95632841	0.037063804	0.893	0.7315	1.114661158
Shanghai	2021	0	0.736158157	7.308542798	12.06450908	0.033654914	0.905	0.7328	1.113199979
Peking	2006	0	0.243654667	7.088074833	10.82907494	0.046106238	0.8433	0.7091	0.428849335
Peking	2007	0	0.250849973	7.101066231	10.97170936	0.041182954	0.845	0.7209	0.593465961
Peking	2008	0	0.254173572	7.170004152	11.05135022	0.040272647	0.849	0.7325	0.631466173
Peking	2009	0	0.277938388	7.127557253	11.16268691	0.03440479	0.8501	0.7553	0.635057722
Peking	2010	0	0.297198359	7.137119442	11.23773834	0.030522557	0.8596	0.7511	0.665538716
Peking	2011	0	0.307439443	7.152973385	11.31029507	0.028035692	0.8623	0.7607	0.393506488
Peking	2012	0	0.305602705	7.168194615	11.37910832	0.02839668	0.862	0.7646	0.665987594
Peking	2013	0	0.310117051	7.182580049	11.90618978	0.027072018	0.863	0.7685	0.862559343
Peking	2014	0	0.324618297	7.19548735	11.51287546	0.026035618	0.864	0.7795	1.096224156
Peking	2015	0	0.326800095	7.20429798	11.57587209	0.035171805	0.8651	0.7965	1.224335584
Peking	2016	0	0.339726071	7.217443432	11.68011646	0.033713545	0.865	0.8023	0.479836446
Peking	2017	0	0.367477817	7.214504414	11.7691517	0.058679732	0.8645	0.8056	0.434638779
Peking	2018	1	0.398961029	7.226936018	11.85090371	0.037781386	0.865	0.8098	0.915779641
Peking	2019	1	0.609955401	7.242082359	12.00896227	0.027719972	0.866	0.8352	0.696877712
Peking	2020	1	0.658804755	7.24814839	12.0130278	0.021235755	0.8755	0.8387	0.461913416
Peking	2021	1	1.023323144	7.254177846	12.12258234	0.012844028	0.875	0.8167	0.209051508
Tianjin	2006	0	0.215970486	6.85529288	10.62529507	0.075541498	0.7572	0.4021	0.597794872
Tianjin	2007	0	0.238684125	6.865995345	10.73904534	0.079463185	0.7632	0.4054	0.687543702
Tianjin	2008	0	0.268081574	6.876130444	10.92365169	0.081095424	0.7721	0.3794	0.757240417
Tianjin	2009	0	0.31939494	6.887389293	11.04410514	0.081914151	0.7802	0.4527	0.887974213
Tianjin	2010	0	0.347643261	6.892489345	11.19813252	0.079614861	0.7957	0.4595	0.997179674
Tianjin	2011	0	0.351920529	6.904148783	11.35290928	0.074576929	0.805	0.4616	0.863815986
Tianjin	2012	0	0.37053991	6.900932054	11.44221326	0.07351595	0.8156	0.4699	0.858664822
Tianjin	2013	0	0.380694491	6.9117473	11.8715016	0.072528891	0.82	0.4805	0.58723545
Tianjin	2014	0	0.392761605	6.924317367	11.56391321	0.073691899	0.8227	0.4934	0.423801173
Tianjin	2015	0	0.419379273	6.934299834	11.58951607	0.079593804	0.8264	0.5215	0.440521256
Tianjin	2016	0	0.463933591	6.950814768	11.65314817	0.114480636	0.8291	0.5644	1.139821853
Tianjin	2017	0	0.471185479	6.956545443	11.6774394	0.039036357	0.8292	0.5815	1.248599144
Tianjin	2018	0	0.477399932	6.986566459	11.70115454	0.017066394	0.8315	0.5862	0.791602577
Tianjin	2019	0	0.486266845	7.010311867	11.4116787	0.023143088	0.8348	0.6346	0.741674813
Tianjin	2020	0	0.505819465	7.029972912	11.5289366	0.022588024	0.847	0.6439	0.695103036
Tianjin	2021	0	0.573386077	7.049254841	11.64160008	0.01846765	0.8488	0.6126	0.651559882
Guangzhou	2006	0	0.372443737	6.634265353	11.05247605	0.038369018	0.8204	0.576	0.731419193
Guangzhou	2007	0	0.397926474	6.650899813	11.18175117	0.0351449	0.8217	0.5841	0.783920604
Guangzhou	2008	0	0.463606416	6.664587576	11.30507685	0.030624472	0.8223	0.5902	0.862129112
Guangzhou	2009	0	0.45421596	6.677864013	11.39731257	0.028206856	0.8253	0.6085	0.854253881
Guangzhou	2010	0	0.481717456	6.692257425	11.54853389	0.025058205	0.8378	0.6101	0.916021431
Guangzhou	2011	0	0.516625693	6.702697195	11.48850981	0.022199228	0.8413	0.6151	0.824531653
Guangzhou	2012	0	0.496745495	6.712105292	11.57033551	0.02209467	0.8502	0.6359	1.298412053
Guangzhou	2013	0	0.540653014	6.724192953	12.12993646	0.019293728	0.8527	0.6462	0.852771719
Guangzhou	2014	0	0.565911848	6.736254961	11.76351296	0.018777728	0.8543	0.6523	0.995249669
Guangzhou	2015	0	0.626279119	6.750153651	11.82179156	0.018637768	0.8553	0.6711	1.164128319
Guangzhou	2016	0	0.649412676	6.768493212	11.86311039	0.019372909	0.8606	0.6935	1.189749227
Guangzhou	2017	0	0.751198662	6.800170068	11.90976683	0.020027569	0.7693	0.696589397	1.24144514
Guangzhou	2018	0	1.000984247	6.833031733	11.95434313	0.019137975	0.8638	0.7175	1.234334656
Guangzhou	2019	0	0.672038972	6.860663671	11.96034473	0.02085546	0.8646	0.7162	1.129494925
Guangzhou	2020	0	0.726326242	6.89060912	11.81337814	0.021162032	0.8619	0.7251	1.048590431
Guangzhou	2021	0	1.025808695	6.91968385	11.9208276	0.018757407	0.8646	0.7156	0.984264245
Chengdu	2006	0	0.161842633	7.006151601	10.13344782	0.053511343	0.615	0.4885	0.691705566
Chengdu	2007	0	0.174915263	7.014167242	10.18584296	0.026025687	0.626	0.4768	0.959060664
Chengdu	2008	0	0.182731636	7.025502758	10.33705409	0.039972482	0.6358	0.4651	1.07597416
Chengdu	2009	0	0.209908286	7.038458927	10.46922741	0.042433912	0.6485	0.4959	1.238563878
Chengdu	2010	0	0.251824795	7.046708199	10.78952524	0.059212798	0.6551	0.5017	1.252136007
Chengdu	2011	0	0.283283429	7.059016073	10.79456304	0.061746249	0.67	0.4936	0.789595315
Chengdu	2012	0	0.326954367	7.06757557	10.96169443	0.066623356	0.6844	0.4946	1.100722562
Chengdu	2013	0	0.296662792	7.0800265	11.24731795	0.076258322	0.694	0.5022	0.40511468
Chengdu	2014	0	0.375690449	7.098953984	11.15652191	0.05350811	0.7037	0.5162	0.83391656
Chengdu	2015	0	0.341414495	7.113182824	11.21550277	0.043363449	0.7147	0.5281	0.44565509
Chengdu	2016	0	0.35168347	7.243512975	11.25104109	0.03277167	0.7062	0.5311	0.50989657
Chengdu	2017	0	0.320330857	7.268920128	11.36060145	0.049282019	0.7185	0.531597104	0.46316288
Chengdu	2018	1	0.323436319	7.297091005	11.4593348	0.052942575	0.7312	0.5412	0.497917251
Chengdu	2019	1	0.310166554	7.313220387	11.54622484	0.053398193	0.7441	0.6557	0.532880911
Chengdu	2020	1	0.267483061	7.33171497	11.35836303	0.054750211	0.788	0.6572	0.564191262
Chengdu	2021	1	0.293759658	7.349873705	11.45764529	0.048448604	0.795	0.6638	0.592392468
Shenzhen	2006	0	0.416761947	5.282340412	11.14836235	0.044825896	1	0.4742	1.436001826
Shenzhen	2007	0	0.435421819	5.358377123	11.28533454	0.040942556	1	0.4984	1.454480213
Shenzhen	2008	0	0.467160154	5.429652599	11.40549614	0.035854563	1	0.5104	1.478797904
Shenzhen	2009	0	0.485515697	5.505168921	11.43790015	0.034649345	1	0.5325	1.40354009
Shenzhen	2010	0	0.506409921	5.560181506	11.57946199	0.030360732	1	0.5272	1.377598609
Shenzhen	2011	0	0.512792073	5.590613777	11.61205561	0.025818348	1	0.535	1.548489629
Shenzhen	2012	0	0.569043155	5.661570626	11.72194575	0.025490878	1	0.5565	1.538257857
Shenzhen	2013	0	0.539063464	5.738183901	13.05568711	0.02335393	1	0.5654	0.965955837
Shenzhen	2014	0	0.618026949	5.805737197	11.91501823	0.02228312	1	0.5739	0.956270548
Shenzhen	2015	0	1.00376102	5.87208962	11.97025537	0.023120772	1	0.5878	0.970119894
Shenzhen	2016	0	0.671506787	5.953243334	12.02820715	0.022940888	1	0.6005	1.030495416
Shenzhen	2017	0	1.011730859	6.075346031	12.09228505	0.022862775	0.9974	0.576765584	1.029231937
Shenzhen	2018	0	1.009798397	6.120297419	12.15250308	0.022410474	0.9975	0.5878	1.15490672
Shenzhen	2019	0	1.008251744	6.311734809	12.22336723	0.020007213	0.9952	0.6093	1.577281137
Shenzhen	2020	0	1.034003564	6.381816017	11.97860099	0.018486341	0.9932	0.6213	1.984391051
Shenzhen	2021	0	1.006573305	6.447305863	12.06487192	0.014773967	1	0.6294	2.37704926

.

3.2. Variable Definitions

3.2.1. Dependent Variable

Energy efficiency (EE) is the dependent variable. Energy efficiency reflects the relationship between energy consumption and the economic or social benefits generated in each production or economic activity. It is a comprehensive indicator of the level of energy consumption and output performance [5,7]. Previous studies on measuring energy efficiency at the prefecture level typically used methods such as night-time lighting intensity or the ratio of energy inputs to outputs. However, these methods fail to fully capture the effectiveness of energy use and the quality of economic development. Given that improvements in energy efficiency are often due to capital support or technological advancements from investments in advanced equipment and processes, this study, referring to the research of Shi et al. [43], selects labor, capital, and energy as inputs, and regional GDP as the desirable output. Industrial sulfur dioxide, industrial smoke and dust, and industrial wastewater emissions are considered undesirable outputs. The SBM (Slacks-based measure) index method is used to calculate the energy efficiency of prefecture-level cities. The SBM index is a type of data envelopment analysis (DEA) method used to measure the efficiency of production or decision-making units. Compared to traditional DEA models, the SBM index has greater precision as it not only evaluates the ratio of inputs to outputs but also considers slack variables representing input surpluses and output shortfalls.

3.2.2. Independent Variable

The core explanatory variable in this study is the environmental protection tax (EPT). After the implementation of the EPT, 30 provincial-level administrative units in China divided their environmental tax standards into provinces where the tax rate was significantly higher than the pollution discharge fees and provinces where it remained about the same. Provinces with significantly higher tax rates include Hebei, Jiangsu, Shandong, Henan, Hunan, Sichuan, Chongqing, Guizhou, Hainan, Guangxi, Shanxi, and Beijing. Provinces with tax rates similar to the discharge fees include Hubei, Zhejiang, Fujian, Jilin, Anhui, Jiangxi, and Shaanxi. Thus, the “tax rate increase” group is defined as the treatment group, and the “unchanged tax rate” group as the control group. A city-level dummy variable is constructed, where treat = 1 for the treatment group and treat = 0 for the control group. Since the law was implemented on 1 January 2018, the period 2018 and after is defined as time = 1, and before 2018 as time = 0. The interaction term of treat and time (treat_time) is used as the independent variable in this study. The EPT rates for air and water pollutants across various provinces and cities are shown in

Table 1. EPT rates in each province.

Province	Atmospheric Pollutants (Requirements 1.2–12)	Water Contamination (Defined Range 1.4–14)
Peking	12	14
Shanghai	Nitrogen oxide 8.5, SO2 7.5, others 1.2	COD 5, ammonia nitrogen 4.8, others 1.4
Tianjin	Nitrogen oxide 8, SO2, dust 6, others 1.2	COD, ammonia 7.5, others 1.4
Hebei	Tiered: 9.6, 6, 4.8	Tiered: 11.2, 7, 5.6
Shandong	NOx, SO2 6, others 1.2	COD, ammonia, heavy metals 3, others 1.4
Jiangsu and Henan	4.8 (Nanjing 8.4)	5.6 (Nanjing 8.4)
Zhejiang	Heavy metals 1.8, others 1.2	Heavy metals 1.8, others 1.4
Sichuan	3.9	2.8
Chongqing	3.5	3
Hunan	2.4	3
Guizhou and Hainan	2.4	2.8
Hubei	NOx, SO2 2.4, others 1.2	COD, ammonia, heavy metals 2.8, others 1.4
Guangdong and Guangxi	1.8	2.8
Heilongjiang and Shanxi	1.8	2.1
Fujian	1.2	COD, ammonia, heavy metals 1.5, others 1.4
Anhui, Jiangxi, Yunnan, Liaoning, Jilin, Gansu, Ningxia, Qinghai, Shaanxi, Xinjiang	1.2	1.4

.

3.2.3. Other Variables

• Strengthening Environmental Enforcement.

Strengthening environmental enforcement refers to the rigorous implementation of environmental laws and regulations to enhance control over polluting behaviors, effectively reducing environmental pollution and subsequently promoting efficient energy use. Following the approach of Wang [50], this study measures the strengthening of environmental enforcement by using the number of environmental penalty cases in each prefecture-level city, as reported in the PKULaw database. Specifically, the natural logarithm of each city’s number of environmental penalty cases plus one is used to reflect the intensity of enforcement. This method captures how changes in enforcement affect urban energy efficiency and explores the potential impact of environmental policies on sustainable urban development.

• Technological Innovation.

Technological innovation refers to the development and application of new technologies and processes to enhance production efficiency and resource utilization, thereby promoting sustainable economic and social development. Referring to previous studies, the level of technological innovation in cities is measured by taking the natural logarithm of the number of granted invention patents and utility model patents, plus one, at the prefecture level.

3.2.4. Control Variables

In line with previous research, the following control variables are selected: population size, level of economic development, level of openness, level of urbanization, industrial structure, and level of marketization in the prefecture-level cities. The detailed definitions of these variables are presented in

Table 2. Variable definition table.

Variable Name	Symbol	Definition
Energy Efficiency	EE	Measured using SBM index method
EPT	treat_time	Interaction term between treat and time
Strengthening Environmental Enforcement	Law	Natural logarithm of the number of environmental penalty cases plus one
Technological Innovation	RD	Natural logarithm of patent grants plus one
Population Size	pop_size	Natural logarithm of year-end total population plus one
Economic Development Level	gdp	Natural logarithm of per capita GDP plus one
Openness Level	openness	Actual utilized foreign capital/regional GDP
Urbanization Level	urban_rate	Urban permanent residents/total permanent residents
Industrial Structure	industry	Tertiary industry value added/regional GDP
Marketization Level	market_level	Urban private and individual employees/total urban employees

below:

3.3. Model Design

To test Hypothesis 1, the following Model 1 is constructed:

$${EE}_{i,t}=\alpha +\beta {treat}_i\_{time}_t+\gamma {Control}_{i,t}+{City}_i+{Year}_t+{\epsilon }_{i,t}$$

(1)

In model (1), t and i represent the year and city, respectively. ${EE}_{i,t}$ is the energy efficiency of city i in year t. treat is the city dummy variable, and time is the time dummy variable. The interaction term treat_time is the core explanatory variable, indicating whether the environmental tax rate was increased in the city after the implementation of the EPT. Control represents the related control variables. City and Year are the city and year fixed effects, and ε denotes the random error term.

3.4. Technology Roadmap

This study benefits from the technical roadmap outlined in Figure 1, which encompasses the research content, research approach, and methodology, systematically elucidating the overall framework and analytical steps of the research. Figure 1 provides a detailed structure of the research design, illustrating how the study progresses from theoretical analysis to empirical testing, ensuring a logical and rigorous research process.

4. Empirical Analysis

4.1. Descriptive Statistics

The descriptive statistics are presented in

Table 3. Descriptive statistics.

Variables	N	Mean	SD	Min	Max
treat_time	4512	0.106	0.308	0	1
EE	4512	0.318	0.128	0.0214	1.177
pop_size	4506	5.877	0.697	2.868	8.136
gdp	4506	10.49	0.724	4.595	13.06
openness	4506	0.0182	0.0196	0	0.229
urban_rate	4506	0.525	0.168	0	1.001
industry	4506	0.400	0.108	0	0.839
market_level	4506	1.238	1.525	0	76.70
law	2018	3.990	1.667	0.693	8.763
RD	4506	7.674	1.855	0	12.927

. The maximum value of urban energy efficiency is 1.177, while the minimum value is 0.0214, indicating significant variation in energy efficiency across different cities.

4.2. Baseline Regression Results

The baseline regression results, shown in

Table 4. Baseline regression results.

Variables	EE	EE
treat_time	0.025 ***	0.027 ***
	(4.68)	(5.12)
pop_size		0.069 ***
		(3.49)
gdp		0.015 **
		(2.20)
openness		−0.350 ***
		(−3.73)
urban_rate		−0.050 **
		(−2.05)
industry		−0.016
		(−0.68)
market_level		0.004 ***
		(4.59)
Constant	0.221 ***	−0.292 **
	(48.85)	(−2.14)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	4512	4506
R-squared	0.267	0.275

Note: Values in parentheses represent T-values. *** and ** indicate significance at the 99% and 95% confidence levels, respectively.

, indicate that the coefficient of the core explanatory variable “treat_time” remains statistically significant at the 99% level, regardless of the inclusion of control variables and after controlling for city and year fixed effects. This finding confirms that the implementation of the EPT has significantly improved energy efficiency in “high tax” cities, thus validating hypothesis H1. As an economic tool for environmental protection, EPT increases environmental compliance costs, prompting firms to reassess their energy usage efficiency and adopt more effective energy management and environmental protection measures under policy pressure. Compared to the previous pollution charge system, this tax mechanism not only raises compliance costs but also imposes stricter demands on energy efficiency, incentivizing firms to adopt cleaner technologies and efficient equipment in production to mitigate the tax burden. Moreover, through market-based incentives, the implementation of EPT effectively encourages firms to reduce reliance on high-pollution energy sources and shift investments toward cleaner energy and more efficient production technologies, thereby reducing both energy consumption and emissions. This incentive structure stimulates green innovation and technological advancement within firms, fostering a beneficial “policy-driven firm response” cycle that aligns economic goals with improved energy efficiency. In addition to its direct impact on firms, EPT also strengthens the role of local governments in energy management and environmental enforcement. Local authorities, by applying stricter policy enforcement and oversight, have facilitated the comprehensive implementation of EPT, creating favorable conditions for citywide improvements in energy efficiency. Consequently, the effectiveness of EPT in high-tax-rate cities, as compared to the pollution charge system, provides evidence of the feasibility of using economic incentives to drive sustainable urban development and offers valuable insights into the environmental policy frameworks in other regions.

4.3. Robustness Test

4.3.1. Parallel Trend Test

A key prerequisite for applying the DID model is the parallel trend assumption, which requires that the explanatory variables in both the treatment and control groups follow the same time trend before the implementation. Otherwise, the conclusions drawn from the DID model will be biased. Following the research of Zhao et al. [51], to mitigate endogeneity issues, a de-moaned difference-in-differences approach was used to conduct the parallel trend test, with the period immediately before the policy implementation as the benchmark. The results are shown in Figure 2. Prior to the implementation of the EPT, fluctuations in energy efficiency were minimal, and before the policy took effect, the coefficient of treat_time did not differ significantly from zero at the 95% confidence level. The estimated coefficients for all pre-policy years failed to pass significance tests, indicating that before the implementation of the EPT, there were no significant differences in the trends of environmental pollution between the treatment and control groups. After the implementation of the EPT, several years passed the significance test, indicating that the model meets the parallel trend assumption.

4.3.2. Placebo Test

To test whether the impact of the EPT on energy efficiency is influenced by other unobservable factors, a placebo test was conducted using random sampling. The sampling and regressions were repeated 500 times, generating 500 placebo regression coefficients and their corresponding p-values. The results, shown in Figure 3, reveal that these coefficients are concentrated around 0, and most p-values exceed 0.1, following a normal distribution. This indicates that the regression results are unlikely to have occurred by chance. Furthermore, the true estimated coefficient from the baseline regression (0.027) significantly differs from the coefficients obtained in the placebo test. This confirms that the effect of the EPT on energy efficiency is not influenced by other unobservable factors, passing the placebo test.

4.3.3. Excluding the Influence of Other Policies

During the sample period, China implemented several other environmental policies aimed at reducing emissions, which may have affected the impact of the EPT on urban energy efficiency. For instance, the Low-Carbon City Pilot Policy, implemented in 2010, and the Carbon Emission Trading Policy, initiated in 2011, could influence energy efficiency. Some scholars argue that energy efficiency could be affected by the Low-Carbon City Pilot Policy, which promotes clean energy technologies and infrastructure development, encouraging cities to adopt more efficient solutions in energy production and consumption, thereby significantly improving energy efficiency. However, the unequal implementation of these policies across cities, due to resource and technological limitations, might lead to inefficiencies in some cases. To address this, cities that implemented the Low-Carbon City Pilot Policy and the Carbon Emission Trading Policy were removed from the baseline regression model to eliminate the potential influence of these policies. The regression results, shown in

Table 5. Exclusion of other policies: regression results.

Variables	EE	EE
treat_time	0.025 ***	0.025 ***
	(4.02)	(4.02)
pop_size		−0.002
		(−0.07)
gdp		0.028 ***
		(3.58)
openness		−0.288 **
		(−2.54)
urban_rate		−0.038
		(−1.28)
industry		−0.020
		(−0.80)
market_level		0.003 ***
		(3.86)
Constant	0.217 ***	−0.018
	(40.99)	(−0.11)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	3328	3328
R-squared	0.271	0.280

Note: Regardless of whether control variables are included, the coefficient of the core explanatory variable is significantly positive at the 99% level, passing the robustness test against the influence of other policies. *** and ** indicate significance at the 99% and 95% confidence levels, respectively.

, demonstrate that the coefficients remain significant even after excluding the cities impacted by these other policies, further confirming the robustness of the findings.

4.3.4. Replacing the Dependent Variable

Referring to previous scholarly research, energy efficiency per unit output was calculated based on total electricity consumption, the supply of natural gas, and liquefied petroleum gas, all converted into tons of standard coal. The data for energy efficiency per unit output were derived by dividing actual GDP by the total tons of standard coal, which serves as a proxy variable for urban energy efficiency per unit output. The regression results are shown in

Table 6. Replacing the dependent variable: regression results.

Variables	EE	EE
treat_time	0.128 ***	0.149 ***
	(4.67)	(5.44)
pop_size		0.575 ***
		(5.61)
gdp		0.079 **
		(2.28)
openness		−0.807 *
		(−1.67)
urban_rate		−1.037 ***
		(−8.30)
industry		−0.097
		(−0.81)
market_level		0.008 *
		(1.90)
Constant	2.487 ***	−1.139
	(105.85)	(−1.62)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	4511	4505
R-squared	0.436	0.451

Note: Regardless of whether control variables are included, the coefficient of the core explanatory variable is significantly positive at the 99% level, passing the robustness test using alternative dependent variables. ***, **, and * indicate significance at the 99%, 95%, and 90% confidence levels, respectively.

, and the coefficient of the core explanatory variable, treat_time, remains significantly positive, passing the robustness check.

5. Mechanism Analysis

Based on the previous analysis, the implementation of the EPT significantly improves urban energy efficiency. This section analyzes the specific pathways through which energy efficiency is enhanced, focusing on two aspects.

5.1. Strengthening Environmental Enforcement

After the implementation of the EPT Law, in cities with increased tax rates, enterprises face higher environmental costs. These firms may attempt to reduce tax liabilities through improper means. However, the EPT Law emphasizes transparency and compliance, creating an opportunity to reduce collusion between businesses and government officials. The government needs to strengthen environmental enforcement by improving oversight mechanisms and information disclosure, thereby enhancing tax management transparency and reducing opportunities for enterprises to gain advantages through connections. Following Jiang [52], the following Model 2 is built based on Model 1 to test this mechanism:

$${Law}_{i,t}=\alpha +\beta {treat}_i\_{time}_t+\gamma {Control}_{i,t}+{City}_i+{Year}_t+{\epsilon }_{i,t}$$

(2)

The regression results of Model 2 are shown in

Table 7. Mechanism test for strengthening environmental enforcement.

Variables	EE	Law
treat_time	0.027 ***	0.423 ***
	(5.12)	(5.26)
pop_size	0.069 ***	1.812 ***
	(3.49)	(3.06)
gdp	0.015 **	−0.069
	(2.20)	(−0.42)
openness	−0.350 ***	−0.367
	(−3.73)	(−0.21)
urban_rate	−0.050 **	0.099
	(−2.05)	(0.14)
industry	−0.016	0.142
	(−0.68)	(0.43)
market_level	0.004 ***	0.010
	(4.59)	(0.93)
Constant	−0.292 **	−8.716 **
	(−2.14)	(−2.15)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	4506	2018
R-squared	0.275	0.568

Note: *** and ** indicate significance at the 99% and 95% confidence levels, respectively.

. The coefficient of treat_time for Law is significantly positive at the 99% confidence level, indicating that the EPT Law’s implementation has significantly strengthened environmental enforcement. According to the Porter hypothesis, appropriate environmental regulations can incentivize firms to innovate and improve production efficiency [23]. By imposing stricter environmental standards, enhanced enforcement encourages enterprises to adopt cleaner and more efficient production processes, which reduces emissions and improves energy efficiency simultaneously [53,54]. Thus, Hypothesis 2 is confirmed: the implementation of the EPT improves urban energy efficiency by strengthening environmental enforcement.

5.2. Technological Innovation

The implementation of the EPT provides a strong incentive for technological innovation. The imposition of environmental taxes forces enterprises to re-evaluate their production processes and technical equipment. In response to the increased environmental costs, firms are driven to seek innovative technologies to enhance energy efficiency and reduce pollution emissions. Following the approach of Jiang [52], the following model is constructed:

$${RD}_{i,t}=\alpha +\beta {treat}_i\_{time}_t+\gamma {Control}_{i,t}+{City}_i+{Year}_t+{\epsilon }_{i,t}$$

(3)

The regression results for Model 3, shown in

Table 8. Mechanism test for technological innovation.

Variables	EE	RD
treat_time	0.027 ***	0.166 ***
	(5.12)	(3.06)
pop_size	0.069 ***	−0.383 *
	(3.49)	(−1.89)
gdp	0.015 **	−0.067
	(2.20)	(−0.98)
openness	−0.350 ***	0.234
	(−3.73)	(0.25)
urban_rate	−0.050 **	−0.404
	(−2.05)	(−1.64)
industry	−0.016	−0.095
	(−0.68)	(−0.40)
market_level	0.004 ***	0.037 ***
	(4.59)	(4.19)
Constant	−0.292 **	11.581 ***
	(−2.14)	(8.31)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	4506	4506
R-squared	0.275	0.437

Note: ***, **, and * indicate significance at the 99%, 95%, and 90% confidence levels, respectively.

, indicate that the coefficient of treat_time on RD is significantly positive at the 99% level. This suggests that the implementation of the EPT Law has significantly improved the level of technological innovation in cities. By combining the results of Model 1 with a wealth of literature that confirms how technological innovation improves energy efficiency through the application of new technologies, enhancement of management practices, and promotion of industrial upgrading [37,39,48,55], it can be concluded that the EPT Law improves urban energy efficiency by fostering technological innovation. Thus, Hypothesis 3 is validated.

6. Further Research

6.1. Heterogeneity in City Size

City size has a significant impact on energy efficiency. On one hand, larger cities tend to have more advanced infrastructure and resource allocation, allowing for more efficient energy use. Their concentrated economic activities, along with more efficient public transportation and energy supply systems, reduce energy consumption per unit of GDP. On the other hand, smaller cities, despite facing relatively limited resources and technology, tend to have more flexible development models and can quickly adapt to policy changes. This flexibility may make them more capable of implementing innovative energy-saving measures, especially in the context of environmental tax reform. To further investigate the effect of city size on the outcomes of environmental tax reform, this paper divides the sample cities into “large cities” and “small cities”. The classification follows the guidelines outlined in the 2014 State Council’s notification, On Adjusting the Standards for City Size. Cities with a permanent population exceeding 3 million are classified as large cities, while those with a population below 3 million are considered small cities. The research findings are presented in

Table 9. Results of city size heterogeneity.

Variables	Larger City	Smaller City
treat_time	−0.012	0.040 ***
	(−1.19)	(5.65)
pop_size	−0.117 ***	0.256 ***
	(−3.44)	(7.09)
gdp	0.033 ***	0.000
	(3.29)	(0.01)
openness	−0.209	−0.240 *
	(−1.26)	(−1.70)
urban_rate	−0.046	−0.010
	(−1.00)	(−0.26)
industry	−0.066 *	0.069 **
	(−1.86)	(2.00)
market_level	0.002 **	0.012 ***
	(2.16)	(5.85)
Constant	0.572 ***	−1.348 ***
	(2.82)	(−4.96)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	1603	2341
R-squared	0.120	0.258

Note: ***, **, and * indicate significance at the 99%, 95%, and 90% confidence levels, respectively.

. For small cities, the coefficient of the core explanatory variable treat_time is not statistically significant. In contrast, for large cities, the coefficient of treat_time is significantly positive at the 99% level. This indicates that the implementation of the EPT has a more pronounced effect on improving energy efficiency in large cities compared to small cities. The stronger policy enforcement capabilities and better resource integration in large cities likely drive this difference. Enterprises in large cities are more inclined to adopt new technologies and innovative measures to comply with environmental regulations, further promoting improvements in energy efficiency. On the other hand, small cities, with limited resources and smaller market scales, may lack the motivation and capacity to make large-scale technological upgrades and investments in response to the environmental tax policy, thus yielding less significant effects.

6.2. City Location Heterogeneity

Cities in different regions of China have significant disparities in terms of natural conditions and industrial structure. In northern China, coal-based heating systems are used during the winter, leading to high coal consumption and potentially lower energy efficiency, particularly during the heating season. Additionally, the industrial system in northern China is heavily reliant on polluting heavy chemical industries, which means northern cities may face greater challenges when adjusting to the implementation of the EPT. In contrast, southern cities tend to rely more on clean energy and the service sector, which might provide more room for improvement in energy efficiency following the policy implementation. Therefore, the impact of the EPT on energy efficiency may differ between regions, with northern cities potentially requiring stronger policy support and technological innovation to achieve desired environmental outcomes, while southern cities may more easily adapt to the policy shift. This study distinguishes between northern and southern cities based on the Qinling-Huaihe Line, analyzing the differential impacts of the EPT implementation on energy efficiency. The results, shown in

Table 10. Results of city location heterogeneity.

Variables	North	South
treat_time	0.016 *	0.022 ***
	(1.84)	(3.24)
pop_size	0.138 ***	0.102 ***
	(3.11)	(4.33)
gdp	0.054 ***	0.018 *
	(4.80)	(1.85)
openness	−0.497 ***	−0.359 ***
	(−3.89)	(−2.64)
urban_rate	−0.083 **	−0.043
	(−2.53)	(−1.20)
industry	−0.084 ***	0.074 **
	(−2.58)	(2.33)
market_level	0.002 **	0.011 ***
	(2.51)	(5.56)
Constant	−1.031 ***	−0.549 ***
	(−3.91)	(−2.92)
Year FE	Yes	Yes
City FE	Yes	Yes
Observations	2048	2426
R-squared	0.325	0.267

Note: ***, **, and * indicate significance at the 99%, 95%, and 90% confidence levels, respectively.

, indicate that the regression coefficient for northern cities is significant at the 90% level, while for southern cities, the coefficient is significant at the 99% level. This suggests that the implementation of the EPT has a more pronounced positive effect on energy efficiency in southern cities. This outcome may reflect the relatively lower baseline energy consumption and more flexible economic structures in southern cities, which enable them to quickly transition to more efficient energy use patterns under the policy incentives.

7. Discussion

As a major reform in China’s environmental tax system, the EPT has significantly promoted urban energy efficiency. Based on panel data from 282 prefecture-level cities in China from 2006 to 2021, this study treats the 2018 implementation of the EPT as a quasi-natural experiment and employs a DID approach to assess the impact of EPT on urban energy efficiency. The findings indicate that the EPT has significantly enhanced energy efficiency in cities, primarily by strengthening environmental law enforcement and fostering technological innovation. The EPT’s impact on energy efficiency is particularly notable in southern and larger cities. Baseline tests show that EPT implementation raised urban energy efficiency by approximately 2.7 percentage points more in treated cities compared to control cities, confirming Hypothesis 1. This result aligns with most studies, though some research uses the 2016 policy announcement date as the quasi-natural experiment point [24,25]. While the EPT was announced in 2016, its actual implementation began on 1 January 2018, and it is the enactment—rather than the announcement—that affects firm behavior or urban energy efficiency. In economic policy research, policy effects often have a lag and typically become apparent only after implementation. Although the 2016 announcement may have triggered initial market responses, it is unlikely to have led to significant changes in energy efficiency, so using the policy’s official implementation as the time point better ensures that the model captures the genuine effects of policy enforcement rather than early anticipatory responses to the announcement.

In the parallel trend test, while the interaction term coefficients do not significantly differ from zero at the 5% confidence interval, we observe notable changes in the coefficient in the year prior to implementation. This result could differ from zero to 10% confidence level, possibly due to preliminary market reactions to the EPT announcement in 2016. However, significant effects became more apparent following its full implementation in 2018. Notably, the coefficient in the second year after the policy shock is not significantly different from zero, unlike in other post-treatment years. This may be due to a period of adjustment or temporary relaxation in EPT enforcement. These findings suggest that EPT may improve energy efficiency immediately after post-implementation but face disturbances or variations in enforcement in the second year, resuming its impact in later years. Additionally, a lack of complete model specification or improper selection of control variables may have contributed to this anomaly. Future research should carefully review control variables and model settings to accurately capture policy effects.

Mechanism tests reveal that EPT enhances urban energy efficiency by strengthening environmental enforcement and promoting technological innovation. Previous studies have found that EPT also promotes green technological innovation, industrial structure upgrading, and foreign direct investment (FDI), all of which can enhance urban energy efficiency [24,25,33]. Policy pressures have driven traditional high-energy-consumption industries to transition toward low-energy, cleaner sectors, accelerating the optimization and upgrading of industrial structures. EPT implementation has attracted more foreign investments focused on sustainable development, introducing advanced technologies and management practices that further improve urban energy efficiency.

In heterogeneity tests, the EPT had a more pronounced effect on energy efficiency in southern and larger cities. This result aligns with findings in most scholarly research. Some studies, however, have observed that EPT has a greater impact on environmental performance in northern China, where heavy industry prevails, and where higher energy consumption and pollution emissions offer greater potential for reductions under policy intervention [37]. Limited research has examined regional heterogeneity differences in energy efficiency, marking this study as a valuable contribution to the literature on regional policy effects.

8. Conclusions

8.1. Conclusions and Recommendations

Based on panel data from 282 prefecture-level cities in China from 2006 to 2021, this study utilizes the 2018 implementation of the EPT as a quasi-natural experiment to empirically investigate the impact of the reform of EPT on urban energy efficiency. The findings demonstrate that the law significantly enhances urban energy efficiency, a conclusion that remains robust across multiple tests, including parallel trend tests and placebo checks. Mechanism analysis reveals that this improvement is driven by strengthened environmental law enforcement and technological innovation. Moreover, heterogeneity analysis indicates that the policy’s effects are more pronounced in southern cities compared to northern cities, and that large cities benefit more significantly from the law’s implementation than small cities.

From these results, several policy recommendations are proposed.

Policy makers should support corporate investment in energy-saving technology by implementing targeted incentive policies. For example, they could establish a “Innovation Fund” to finance the research and application of efficient, clean technologies, thereby enhancing overall energy efficiency. Additionally, policymakers need to reinforce environmental law enforcement to ensure effective policy implementation. This includes cross-departmental collaboration to strengthen environmental regulations on high-pollution, high-energy industries, creating regular compliance audits, and strictly penalizing violations as a deterrent. Recognizing the economic structure and energy-use differences between northern and southern cities, policymakers should also implement differentiated environmental policies, such as encouraging northern cities to increase clean energy use and supporting green, low-carbon transitions in the service sector of southern cities.

Local governments can encourage energy-saving efforts by developing reward mechanisms that offer tax reductions or financial subsidies to companies adopting energy-efficient technologies, motivating them to make real progress in energy efficiency and emissions reduction. Additionally, local governments should rigorously enforce the EPT, especially in high-pollution industries, to ensure that the policy is fully executed at the local level. Local governments should also tailor policy recommendations to the specific needs of their regions; for example, they can enforce high-intensity environmental measures in larger cities and adopt more cost-effective energy-saving solutions in smaller ones to optimize resource allocation and regional development.

As key stakeholders, businesses should proactively respond to policies by increasing investment in energy-saving technologies and equipment upgrades to mitigate environmental costs. Companies can take advantage of the market-driven incentives provided by policy, exploring clean technologies and efficient production methods that achieve energy efficiency and enhance competitiveness. Additionally, businesses should strengthen environmental compliance management by conducting regular internal reviews to ensure alignment with environmental policy requirements, reducing the burden of environmental taxes, and advancing their sustainability goals. This policy-driven incentive mechanism encourages businesses to pursue green innovation, contributing to the improvement of urban energy efficiency and the achievement of sustainable development goals.

8.2. Limitations and Future Research Directions

Despite these contributions, the study has some limitations. First, while this research offers an insightful analysis based on available panel data, potential data constraints in capturing finer-grained details of environmental law enforcement intensity and specific technology adoption at the city level may limit the comprehensiveness of the mechanism analysis. Future studies could seek to incorporate detailed datasets on enforcement actions and technology use at the corporate level to provide a more nuanced understanding of EPT’s impact pathways.

Second, this study primarily focuses on the EPT’s impact within China; thus, the generalizability of findings to other countries may be limited. Comparative studies could explore the effects of environmental taxation across different institutional and economic contexts, enabling a broader understanding of how environmental taxes influence energy efficiency globally.

Finally, the study does not fully explore the long-term adaptation behaviors of firms in response to environmental taxes. Future research could examine the dynamic effects of the EPT over an extended period to better capture shifts in corporate innovation strategies and resource allocation decisions, thus contributing further to sustainable policy development insights.