How Does Fintech Affect Green Total Factor Energy Efficiency? Evidence from 240 Cities in China

Abstract

Enhancing green total factor energy efficiency (GTFEE) is crucial for achieving sustainable development. Against this backdrop, this study aims to investigate the impact of fintech on GTFEE, using annual data from 240 Chinese cities between 2011 and 2021. Methodologically, we employ the SBM–Malmquist–Luenberger model to measure GTFEE and assess the role of fintech. The results demonstrate that fintech significantly promotes GTFEE, a finding that remains robust after addressing endogeneity issues and replacing key variables. Further mechanism analysis reveals that fintech facilitates GTFEE by alleviating financing constraints and stimulating technological innovation. Moreover, the effect is particularly pronounced in eastern regions, non-resource-based cities, service-oriented cities, and larger urban areas. Importantly, quantile regression results confirm that fintech exerts a stronger positive impact at higher quantiles of the GTFEE distribution. These findings offer both theoretical insights and practical policy implications for advancing energy efficiency through fintech development.

1. Introduction

Amidst the converging historical processes of an intensifying global climate crisis and accelerating sustainable development, enhancing energy utilization efficiency has become paramount for the international community in jointly tackling climate change challenges and achieving a green, low-carbon transition. The Renewable Energy Statistics 2025 released by the International Renewable Energy Agency (IRENA) indicates that the global renewable energy installed capacity achieved a compound annual growth rate (CAGR) of 15.1% from 2017 to 2024. In 2024, global renewable capacity additions reached 585 gigawatts (GW), accounting for 92.5% of the total expansion in power generation capacity. Countries worldwide are exploring low-carbon development pathways through means such as technological innovation, policy incentives, and market regulation [1]. Within this process, fintech is increasingly emerging as a crucial force driving the energy transition [2]. Smart algorithms significantly enhance renewable energy utilization rates by optimizing energy resource allocation; for instance, machine learning-driven grid load forecasting systems can effectively mitigate the intermittency issues associated with wind and solar power generation [3]. Blockchain technology constructs transparent carbon asset trading systems, substantially reducing the costs of green certification and cross-border settlement [4]. Meanwhile, big data-empowered ESG investment analytics tools systematically transform capital allocation efficiency within the low-carbon sector [5]. It is evident that fintech holds significant importance for improving GTFEE, providing the foundational framework for the transition of human civilization from an industrial–metabolic model to a digital–ecological paradigm.

China’s rapid economic growth has incurred severe environmental costs. Characterized by excessive resource input, high energy consumption, and substantial pollution, this unsustainable development model poses critical challenges [6]. To address this, China’s “dual-carbon” targets not only demonstrate its international commitment but also aim to mitigate domestic resource and environmental constraints [7]. Against this backdrop, GTFEE, as a comprehensive metric balancing economic growth and environmental quality, has become a key standard for evaluating urban sustainable development levels [8]. However, China currently faces challenges of resource depletion and energy price volatility [9], creating an urgent need to improve energy efficiency. Substantial capital within the financial sector can provide considerable support for sustainable development, promoting energy efficiency gains, and numerous fintech companies are also beginning to demonstrate interest in undertaking environmental and social responsibilities [10]. The unique institutional environment and market structure in China provide a distinctive sample for studying the interaction between fintech and GTFEE. As the world’s largest carbon emitter and the largest digital payment application market, China simultaneously faces the rigid constraints of the dual-carbon goals and the requirement for tasks of financial system reform. This dual pressure has given rise to solutions with Chinese characteristics. On one hand, the green finance functional modules of the central bank digital currency (CBDC) have begun pilot implementation in provinces like Zhejiang and Guangdong, enabling intelligent linkage between carbon credits and currency flows. On the other hand, regulatory sandbox-based fintech innovation pilots for green finance, such as Ant Group’s “Carbon Account” and Tencent’s “Green Supply Chain Finance Platform”, are reconstructing the incentive structure for traditional energy efficiency improvement by incorporating personal carbon footprints and corporate ESG ratings into credit decision systems. This government-guided and market-driven model [11] is shaping a distinctive Chinese Pathway for enhancing GTFEE supported by fintech.

The contributions of this study are as follows: First, it moves beyond established preliminary links by providing novel urban-level evidence on fintech’s specific effect on Green Total Factor Energy Efficiency (GTFEE). This is achieved through a more comprehensive and granular indicator system that integrates both energy and environmental dimensions. This study not only contributes to identifying new pathways for green development within the Chinese context but also provides valuable insights for other developing countries facing similar challenges, while enriching global academic discourse and evidentiary support on how digital finance can serve sustainable development goals. Second, the research identifies and empirically validates two specific mediating channels: Alleviating financing constraints and stimulating green technology innovation. This offers a deeper mechanistic understanding of how fintech directly influences GTFEE, distinguishing it from earlier aggregate-level studies. Finally, moving beyond aggregate-level analysis, this study conducts detailed heterogeneity analysis from multi-dimensional perspectives. Specifically, it explores the differential characteristics of fintech’s impact on GTFEE across four key dimensions: regional distribution, economic development stage, and urban scale. The analysis reveals significant heterogeneity results, providing a solid empirical foundation for governments to formulate more targeted and differentiated regional green fintech development strategies and energy efficiency enhancement policies based on local conditions.

The remainder of this paper is structured as follows: Section 2 presents the literature review and research hypotheses. Section 3 introduces the research design, explaining variable selection, data sources, and model specification. Section 4 reports the empirical results, including baseline regression findings, results for the mediating effects, conducting heterogeneity analyses, and subsequently performing robustness checks. Section 5 concludes the study and provides policy recommendations.

2. Literature Review and Research Hypotheses

2.1. Literature Review

The nexus between fintech and energy has garnered significant scholarly attention. Some scholars have researched the relationships between fintech and the energy market, energy transition, and energy efficiency. Su and He (2024) noticed that the fintech market and the clean energy market have a significant mutual influence [12]. Additionally, Bouteska and Harasheh (2024) also discovered that the relationship between financial innovation and the energy market has significantly strengthened in the long term. Furthermore, some scholars have also focused on the role of fintech in promoting the energy transition [13]. Li et al. (2023) researched the impact of fintech in industrial enterprises on the energy transition and found that fintech promotes energy transition by reducing the amount of carbon consumption [14]. Moreover, Aziz et al. (2024) confirmed that fintech promotes green growth through the path of energy transition in China [15]. This catalytic role extends globally, with evidence highlighting fintech’s significant contribution to low-carbon energy transitions across BRICS nations (Dai et al. 2025; Zeng et al. 2024) [16,17]. Furthermore, FinTech financing and financial development enhance energy efficiency in OECD nations [18]. In the top six manufacturing nations, natural resources and trade openness can bring an upsurge in CO₂ emissions [19]. Notably, beyond these macro-level systemic shifts, fintech directly enhances micro-level energy utilization efficiency. Specifically, Teng and Shen (2023) empirically established that fintech development substantially improves energy efficiency in OECD economies [20], providing critical evidence for fintech’s direct impact on this core performance metric. Furthermore, Li et al. (2025) more detailed findings indicate that artificial intelligence can facilitate the growth of the renewable energy [21]. Wang and Wang (2025) discovered that artificial intelligence contributes to the enhancement of energy [22].

The impact of fintech on energy efficiency manifests through a multi-dimensional pathway, ultimately culminating in its significant role in advancing GTFEE. Initial research broadly establishes fintech’s contribution to environmental sustainability and emissions reduction, providing the foundational context for its energy-related effects. Studies by Bonsu et al. (2025) [23] and Li and Zhang (2025) [24] demonstrate fintech’s positive influence on the environmental sustainability of manufacturing and the sustainability of urban green innovation, respectively, while Li et al. (2024) [19] empirically confirm that the development of fintech and the improvement of energy efficiency can both reduce carbon emissions. Research on energy efficiency presents more nuanced findings, fintech boosts mineral resource green utilization efficiency [25] by facilitating green technology innovation and optimizing energy consumption structures. Additionally, fintech enhances traditional total factor energy efficiency through improved capital allocation mediated by industrial structure optimization [26]. Synergy with green finance further amplifies energy efficiency gains by operating through mediating channels such as green and digital technological innovation. The effects of this synergy are moderated by climate policy uncertainty and environmental regulations [27]. As a holistic metric, GTFEE systematically accounts for environmental externalities, thereby uncovering the catalytic role of financial technology in sustainable transitions. Wu et al. (2024) [28] directly establish digital finance’s significant impact on GTFEE. They identify the optimization of production factor allocation as the core mechanism, which involves explicitly alleviating distortions in capital, labor, and energy markets and enhancing marginal output-price matching. The synergistic development of fintech and green finance substantially augments this effect, significantly influencing GTFEE across spatial and temporal dimensions [29]. Furthermore, complementary government digital transformation affects GTFEE improvements by optimizing regulatory mechanisms [30]. Conversely, fintech may exert a negative impact on energy efficiency. Hou et al. (2024) [31] demonstrate fintech’s adverse impact in industrial-dominated Asian economies, where it stimulated expansion in high-energy-consuming sectors, increasing energy-intensive production and degrading efficiency.

Overall, existing research on the relationship between fintech and energy efficiency remains relatively scarce, and studies specifically examining the impact of fintech on GTFEE are notably absent. GTFEE serves as a metric for evaluating the economic benefits relative to the environmental impact within an economy’s production activities. It primarily measures the ratio of green economic value created per unit of energy consumption in economic activities. Compared to traditional energy efficiency measures, GTFEE incorporates considerations for green energy inputs and undesirable outputs (such as pollution), placing greater emphasis on environmental friendliness and sustainable development. Given the escalating global emphasis on sustainable development and carbon neutrality, enhancing GTFEE is pivotal for achieving a green economic transition. In addition, most literature investigating fintech’s influence on energy efficiency has failed to conduct a joint analysis incorporating the mediating roles of financing constraints and technological innovation alongside the moderating effect of the digital economy. Understanding these complex mechanisms is crucial, as it holds significant potential for formulating targeted policy recommendations to maximize fintech’s positive contribution to sustainable energy utilization. Furthermore, there is a distinct lack of research focused specifically on this relationship at the city level, despite cities being critical hubs for energy consumption, financial activities, and technological innovation, and the primary sites where fintech interventions are most actively implemented. Moreover, the impact varies significantly across regions at different economic development stages, depending on whether a city is resource-based and by urban scale. In light of these research gaps, this paper positions GTFEE as its core research subject. It conducts an empirical analysis of the aforementioned issues, placing particular emphasis on city-level data to yield more nuanced and actionable insights.

2.2. Research Hypotheses

Fintech is reshaping the traditional landscape of the financial industry, providing new avenues for investment, trading, and wealth management. Digital finance innovations have emerged as an effective facilitator for enhancing GTFEE. Fintech substantially enhances information processing efficiency and transparency within financial markets [32], while greatly expanding the coverage and accessibility of financial services (Ahl et al. [33]). This enhanced capacity for information integration and transmission directly optimizes the efficiency of cross-temporal and spatial allocation of financial resources, thereby reducing the comprehensive costs associated with searching, matching, and executing financial transactions [34]. Concurrently, fintech-driven innovative business models (e.g., digital payments, smart contracts) effectively promote diversified interactions among market participants, directing capital flows towards a broader spectrum of green economic activities. Collectively, these mechanisms influence the comprehensive energy utilization efficiency of economies operating under economic and environmental constraints, empowering the improvement of GTFEE. Based on this, we propose Hypothesis H1:

H1. 

Fintech can promote GTFEE.

Fintech reduces information asymmetry between enterprises and financial institutions through three key mechanisms, including innovating financial service models, expanding diversified financing channels (including both new products from traditional institutions and services from emerging platforms), and optimizing financial resource allocation. This enables financial institutions to assess corporate operational status, financial health, and green transformation potential more comprehensively and accurately, consequently strengthening trust and willingness to provide financing. The direct result is a significant alleviation of the financing constraints faced by enterprises, lowering financing thresholds and costs [35]. Key actions in corporate green transformation processes, such as adopting stricter environmental management standards, implementing energy efficiency retrofits, obtaining environmental certifications, and meeting compliance requirements, all necessitate stable and continuous capital investment as a fundamental guarantee [36]. The fintech-driven mitigation of financing constraints provides crucial funding support precisely for these non-technological, foundational green transformation inputs, thereby aiding in enhancing energy efficiency. Based on the above reasoning, we propose Hypothesis H2:

H2. 

Fintech can promote GTFEE by alleviating financing constraints.

The core capability of fintech lies in utilizing advanced technological tools such as big data analytics and artificial intelligence [37] to enable precise identification, risk assessment, and efficient screening of green projects. This technology-enabled mechanism is manifested in the intelligent matching of innovation factors. For instance, blockchain technology ensures the tamper-proof nature of green patent information. At the same time, real-time energy efficiency data collected by Internet of Things (IoT) devices provides empirical evidence for selecting technological pathways. Machine learning-based climate risk prediction models optimize the technical parameter configurations of green infrastructure, such as wind power layouts. Wang et al. (2025) discovered that the impact of the digital economy and technological innovation on sustainable development is significantly positive [38]. Furthermore, green innovation can promote sustainable development and energy security [39,40]. These synergistic technological effects significantly enhance the innovation production possibility frontier of energy systems, constituting the core transmission pathway through which fintech influences GTFEE. This directly accelerates the research and development (R&D), application, commercialization, and large-scale diffusion of green low-carbon technologies [41], such as renewable energy technologies, carbon capture and storage (CCS), and advanced energy-saving processes, driving a fundamental transformation of the energy system towards decarbonization and higher efficiency. Based on the above reasoning, we propose Hypothesis H3:

H3. 

Fintech can promote GTFEE by optimizing technological innovation.

Based on the research hypotheses outlined above, we establish an empirical model, as illustrated in Figure 1, to explore the impact mechanism of fintech on GTFEE, specifically investigating the mediating roles of financing constraints and technological innovation within this relationship.

3. Methods

3.1. Data Sources

This study employs prefecture-level panel data from 2011 to 2021. During data preprocessing, observations with missing values were systematically screened and excluded. Core indicators were constructed as follows: Regional fintech development was quantified using Baidu News’ advanced search function to capture annual exposure of fintech-related keywords [42]. GTFEE was dynamically measured using the SBM-ML index model. Data for control variables were integrated from multiple sources, including China City Statistical Yearbooks, local statistical bulletins, provincial economic databases, and listed companies’ annual reports, ensuring comprehensive coverage. To mitigate potential distortion from extreme outliers, all variables were winsorized at the 1st and 99th percentiles.

3.2. Variable Selection

3.2.1. Dependent Variable

Following Li and Chen (2021) [43], this study measures GTFEE using the SBM-ML index method. Inputs include labor, capital, and energy consumption. Desirable output is regional GDP, while undesirable outputs comprise industrial SO₂ emissions, industrial soot/dust emissions, and industrial wastewater discharge.

3.2.2. Independent Variable

Fintech represents an innovative business model integrating modern technologies (e.g., information technology, artificial intelligence, blockchain) with financial services to enhance accessibility, efficiency, and transparency. Adopting Chen et al. (2024) [44]‘s approach, we quantified fintech development through Python3.9.12-based web scraping of 48 keywords (e.g., EB storage, NFC payment, cognitive computing, deep learning, big data, blockchain, data mining, business intelligence, smart financial contracts) from Baidu News. Total keyword occurrences were aggregated for each of China’s 300+ prefecture-level cities and log-transformed. However, we acknowledge that this media-based keyword index, while widely adopted, has certain limitations. First, news coverage may not fully capture the actual penetration or usage intensity of fintech services at the grassroots level, potentially introducing a media bias into the measurement. Second, the keyword set, though comprehensive, might overlook emerging subfields or specific applications highly relevant to green finance and energy efficiency. To enhance the robustness of our fintech measurement and mitigate potential construct validity concerns, we implement two supplementary approaches: (1) We conduct a principal component analysis (PCA) combining our keyword index with alternative indicators, such as the number of fintech enterprises registered in each city (based on business registration records) and the volume of mobile payments (where data is available), creating a more comprehensive composite fintech development score. (2) In subsequent robustness checks, we replace the main independent variable with this composite score and also employ the number of fintech-related patent applications per capita as an alternative proxy focusing on technological innovation capacity. This multi-faceted strategy allows us to triangulate the findings and adds confidence that our results are not merely artifacts of a single measurement method.

3.2.3. Control Variables

To enhance research precision, control variables are incorporated to mitigate confounding effects:

(1) Industrial structure (IS): As industrial composition significantly influences GTFEE due to sectoral variations in energy intensity and environmental impact, industrial-dominated structures typically exhibit lower energy utilization efficiency and higher pollution, thereby constraining GTFEE. Following Wu et al. (2024) [28], IS is measured as the ratio of secondary industry output value to real GDP.

(2) Urbanization rate (UR): Reflecting rural-to-urban population migration, UR exerts a dual effect on GTFEE. On the one hand, stricter environmental regulations and the transition to sustainable consumption, facilitated by policy leadership and increased public awareness, may improve GTFEE. On the other hand, prioritizing rapid urban expansion over development quality could hinder such efficiency improvements. This variable is measured by the proportion of permanent urban residents to the total population

(3) Environmental regulation intensity (ERS): Representing corporate pollution abatement costs, increased ERS elevates operational expenses, incentivizing adoption of advanced green technologies and low-pollution industrial restructuring, ultimately improving GTFEE. We measure ERS as the ratio of annual industrial waste gas/water treatment investment to industrial output value in the regions of listed companies.

(4) Foreign direct investment (FDI): Reflecting global market integration, FDI facilitates technology spillovers through clean technology transfers, management expertise, or green patents (e.g., renewable energy technologies), potentially enhancing host-country energy efficiency. This study quantifies FDI as the ratio of actually utilized FDI to local GDP.

(5) Capital investment (INV): Recognizing that high-efficiency equipment (e.g., smart grids, carbon capture systems) requires substantial capital expenditure, and that R&D in energy-saving/clean energy technologies (e.g., photovoltaics, hydrogen) necessitates sustained funding, which may independently boost GTFEE beyond fintech effects. We operationalize this variable as the ratio of local general budgetary expenditure to GDP.

(6) Energy consumption structure (SEC): Directly affecting carbon emissions and pollution, traditional coal-fired power plants exhibit lower energy conversion efficiency (30–40%) compared to natural gas (50–60%) or renewables (zero fuel loss). Higher coal dependency increases energy input requirements per economic output, thereby reducing GTFEE. Following standard practice in energy economics, we quantify SEC as the proportion of coal consumption to total energy consumption.

Major variables and their operationalization are systematically presented in

Table 1. Variables’ description.

Type	Variables	Explanation	Measurement
Explained variable	GTFEE	Green total factor energy efficiency	SBM-ML method calculation
Explanatory variable	fintech	fintech	Fintech-related keywords were extracted from Baidu News; the total search result counts for all keywords corresponding to each prefecture-level city or municipality directly under the central government were aggregated and log-transformed.
Control variable	IS	Industrial structure	Value-added of secondary industry/real GDP
	UR	Urbanization rate	Permanent urban population/total population
	ERS	Environmental regulation strength	Annual expenditure on waste gas/water pollution control in the regions of the listed companies and/annual industrial output value
	FDI	Foreign direct investment	Annual utilized FDI amount/regional GDP
	INV	Capital investment intensity	General budgetary expenditure of local government/regional GDP
	SEC	Energy consumption structure	Coal consumption/total energy consumption

.

3.3. Slack-Based Measure (SBM) Model

Traditional Data Envelopment Analysis (DEA) models commonly employ radial measurement approaches, whose primary deficiency lies in the insufficient consideration of the systemic impact of undesirable outputs (e.g., industrial pollution emissions) on efficiency evaluation. To overcome this technical limitation, this study introduces the non-radial, non-angular SBM model proposed by Tone (2001) [45]. This study assumes the existence of n Decision Making Units (DMUs). This study treats industrial sulfur dioxide (SO₂) emissions and industrial wastewater discharges as undesirable outputs. Their weights are determined using the entropy method. A non-radial direction vector is selected to measure the potential for efficiency improvement. Each DMU consumes m types of production factors (inputs) and obtains s types of outcomes (outputs). The corresponding input matrix is denoted as $\mathrm{X}=(x_{ij})\in R^{m\times n}$ and the output matrix as $\mathrm{Y}=(y_{ij})\in R^{s\times n}$. The model is formulated as follows:

$$\mathrm{m}\mathrm{i}\mathrm{n}\mathsf{\rho }={\displaystyle \frac{1-\frac{1}{m}{\sum }_{i=1}^m{s}_i^{-}/x_{i0}}{1+\frac{1}{s}{\sum }_{i=1}^x{s}_i^{+}/y_{i0}}}$$

(1)

$$\mathrm{s}.\mathrm{t}. x_0=\mathrm{X}\mathsf{\lambda }+s^{-},y_0=\mathrm{Y}\mathsf{\lambda }-s^{+}$$

$$\mathsf{\lambda }\ge 0,s^{-}\ge 0,s^{+}\ge 0$$

where Xλ and Yλ represent the input and output quantities on the production frontier, respectively; $s^{+}$ and $s^{-}$ denote the input excess and output shortfall slacks. The SBM model can effectively incorporate undesirable outputs and measure efficiency more accurately.

3.4. Malmquist–Luenberger (ML) Index

It is an efficiency evaluation index based on the DEA methodology. This index measures technical efficiency changes between two time points or organizations while accounting for both technological progress and technical efficiency components. The index can be decomposed into a technological progress index and a technical efficiency change index, thereby providing a detailed explanation of efficiency dynamics.

The directional distance functions for two adjacent periods are defined as follows:

$${\overrightarrow{D}}_0^{t+1}\left(x^t,y^t,b^t;\mathrm{g}\right)=\mathrm{s}\mathrm{u}\mathrm{p}\left\{\beta :(\right.y^t,b^t)+\beta \mathrm{g}\in p^{t+1}(\left.x^t)\right\}$$

(2)

where $x^t$ represents the input vector at period t; $y^t$ denotes the desirable output vector at period t; $b^t$ indicates the undesirable output vector at period t; g is the directional vector (g = (g_y, −g_b)) specifying the expansion of desirable outputs and contraction of undesirable outputs; β is a scalar representing the maximum feasible proportional expansion/contraction; $p^{t+1}\left(x^t\right)$ describes the production possibility set at period t + 1 for given inputs $x^t$.

Second, the ML index between period t and t + 1 is formally specified as follows:

$$\begin{gathered} {ML}_t^{t+1}={\left[{\displaystyle \frac{(1+{\overrightarrow{D}}_0^{t+1}\left(x^t,y^t,b^t;y^t,-b^t\right))}{(1+{\overrightarrow{D}}_0^{t+1}\left(x^{t+1},y^{t+1},b^{t+1};y^{t+1},-b^{t+1}\right))}}\right. \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\left.\times |{\displaystyle \frac{(1+{\overrightarrow{D}}_0^t\left(x^t,y^t,b^t;y^t,-b^t\right))}{(1+{\overrightarrow{D}}_0^t\left(x^{t+1},y^{t+1},b^{t+1};y^{t+1},-b^{t+1}\right))}}\right]}^{1/2} \end{gathered}$$

(3)

We then decompose the ML index into two constituent components the technical efficiency change index ${MLEFFCH}_t^{t+1}$ and the technological change index ${MLTECH}_t^{t+1}$ to conduct source decomposition analysis of GTFEE. The multiplicative decomposition is formally expressed as:

$$\mathrm{M}\mathrm{L}+{MLEFFCH}_t^{t+1}\times {MLTECH}_t^{t+1}$$

(4)

$${MLEFFCH}_t^{t+1}={\displaystyle \frac{1+{\overrightarrow{D}}_0^t\left(x^t,y^t,b^t;y^t,-b^t\right)}{1+{\overrightarrow{D}}_0^t\left(x^{t+1},y^{t+1},b^{t+1};y^{t+1},-b^{t+1}\right)}}$$

(5)

$${MLEFFCH}_t^{t+1}={\displaystyle \frac{1+{\overrightarrow{D}}_0^t\left(x^t,y^t,b^t;y^t,-b^t\right)}{1+{\overrightarrow{D}}_0^t\left(x^{t+1},y^{t+1},b^{t+1};y^{t+1},-b^{t+1}\right)}}$$

(6)

In the above expression, ${MLEFFCH}_t^{t+1}$ measures the shift in the technical efficiency production possibility frontier between periods t and t + 1. ${MLEFFCH}_t^{t+1}=1$ signifies no contribution from technical efficiency; ${MLEFFCH}_t^{t+1}>1$ denotes increasing technical efficiency with positive contribution; ${MLEFFCH}_t^{t+1}<1$ reflects declining technical efficiency that hinders GTFEE growth. Meanwhile, ${ MLTECH}_t^{t+1}$ aptures technological change or innovation intensity across the same periods: ${MLTECH}_t^{t+1}>1$ indicates technological progress enhancing GTFEE; ${MLTECH}_t^{t+1}<1$ implies technological regression impeding GTFEE growth; ${MLTECH}_t^{t+1}=1$ suggests stagnant technology with neutral impact [46].

3.5. Model Construction

To systematically investigate the causal relationship between financial technology development and green productivity enhancement, we specify the following econometric model:

$${GTFEE}_{i,t}={\alpha }_0+{\alpha }_1{fintech}_{i,t}+\sum _{j=2}^7{\alpha }_j{Controls}_{i,t}+{\theta }_t+{\gamma }_i+{\epsilon }_{i,t}$$

(7)

where i and t denote time and region respectively. Green total factor energy efficiency (GTFEE) is the dependent variable of interest, and fintech (fintech) is the core explanatory variable. Controls denotes the vector set of control variables, including Industrial structure (IS), Urbanization Rate (UR), Environmental regulation strength (ERS), Foreign direct investment (FDI), Capital investment intensity (INV), and Energy consumption structure (SEC). α is the parameter to be estimated for each variable, and ${\theta }_t$ and ${\gamma }_i$ are the individual effect and residual term that do not vary over time, respectively. ${\epsilon }_{i,t}$ is the random disturbance term.

4. Results

4.1. Descriptive Statistics

Table 2. Descriptive statistics.

VarName	Obs	Mean	SD	Min	P25	Median	P75	Max
GTFEE	12,137	0.657	0.160	0.462	0.538	0.611	0.719	1.049
fintech	12,137	3.780	1.153	1.609	2.944	3.689	4.615	6.698
IS	12,137	0.463	0.070	0.297	0.413	0.470	0.519	0.587
UR	12,137	0.699	0.143	0.485	0.597	0.688	0.763	1.000
ERS	12,137	0.213	0.126	0.040	0.110	0.170	0.290	0.500
FDI	12,137	0.373	0.205	0.059	0.204	0.352	0.518	0.807
INV	12,137	3.678	2.039	1.105	2.051	3.328	5.015	9.152
SEC	12,137	0.794	0.089	0.608	0.722	0.803	0.858	0.939
GPPH	12,137	1.939	0.184	1.599	1.812	1.980	2.085	2.246
EPE	12,137	3.179	1.619	0.720	2.054	2.833	3.940	8.397

reports the descriptive statistics of key variables. The mean value of GTFEE is 0.655 with a standard deviation of 0.162, while fintech development shows a mean of 3.800 and standard deviation of 1.115. These results indicate moderate variation across sample observations for core variables. Control variables exhibit distributions consistent with existing literature and fall within plausible empirical ranges.

4.2. Baseline Regression

Table 3. Baseline regression.

	(1)	(2)
	GTFEE	GTFEE
fintech	0.040 **	0.038 ***
	(2.15)	(3.01)
IS		0.287
		(1.04)
UR		−0.222
		(−1.52)
ERS		0.150 ***
		(3.34)
FDI		0.049
		(1.25)
INV		0.019 ***
		(3.46)
SEC		−0.069
		(−0.79)
GPPH		−0.283 *
		(−1.97)
EPE		−0.026 *
		(−1.66)
_cons	0.270 ***	0.868 ***
	(3.80)	(3.33)
Firm fixed effect	Yes	Yes
Year fixed effect	Yes	Yes
Observations	12,135	12,135
R2	0.834	0.877
AR2	0.831	0.875

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the baseline regression results for Model (7). Column (1) reports the univariate regression results, demonstrating that the coefficient of fintech on GTFEE is 0.040, which is statistically significant at the 5% level. This provides preliminary evidence that fintech exerts a positive effect on GTFEE.

Column (2) displays the regression results after incorporating control variables. The estimated coefficient of fintech remains positive (0.038) and statistically significant at the 1% level, further confirming that fintech significantly enhances GTFEE. Notably, the urbanization rate exhibits a negative coefficient significant at the 10% level, suggesting a modest inhibitory effect on GTFEE. In contrast, capital investment shows a significantly positive coefficient at the 5% level, indicating its robust contribution to improving energy efficiency. Most strikingly, environmental regulation intensity demonstrates a strongly positive coefficient significant at the 1% level, highlighting the crucial role of stringent environmental policies in promoting GTFEE.

These empirical findings collectively validate Hypothesis 1, confirming the positive relationship between fintech development and green total-factor energy efficiency enhancement.

4.3. Mechanism Analysis

4.3.1. Mediating Effect of Financing Constraints

Financing constraints (FC) are measured using the absolute value of the SA index [47].

Table 4. Mediating effect.

	(1)	(2)
	FC	TI
fintech	−0.008 *	0.003 ***
	(−1.77)	(3.96)
IS	−0.004	0.120 ***
	(−0.05)	(7.74)
UR	0.036	−0.020 **
	(0.45)	(2.05)
ERS	−0.000	0.004
	(−0.02)	(1.07)
FDI	0.030	0.015 ***
	(1.35)	(4.99)
INV	−0.005	0.001 ***
	(−1.44)	(3.75)
SEC	−0.138 **	0.036 ***
	(2.49)	(3.56)
GPPH	0.029	0.088 ***
	(0.47)	(8.26)
EPE	0.001	−0.001 **
	(0.52)	(−2.46)
_cons	3.821 ***	4.224 ***
	(31.56)	(171.28)
Firm fixed effect	Yes	Yes
Year fixed effect	Yes	Yes
Observations	12,135	12,135
R2	0.175	0.979
AR2	0.161	0.979

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the mediating role of financing constraints in the relationship between fintech and GTFEE. Column (1) displays the estimated coefficient, which is −0.008 and statistically significant at the 10% level, supporting Hypothesis 2. This suggests that fintech enhances GTFEE by overcoming financing constraints.

Mechanistically, fintech mitigates corporate financing constraints, particularly for green transition enterprises, through three key channels: (1) reducing information asymmetry in capital markets, (2) expanding financing channels beyond traditional banking systems, and (3) lowering transaction costs in financial intermediation. The resultant improvement in financing accessibility provides critical capital support for firms to undertake green technology innovation, upgrade production equipment, and optimize environmental management systems. These operational enhancements collectively improve firms’ capability to transform energy inputs and pollution emissions into economic outputs, thereby elevating GTFEE through the financing constraint channel.

4.3.2. Mediating Effect of Technological Innovation

Following Klein (2025) [48], we measure technological innovation (TI) using the natural logarithm of invention patent applications. The regression results in Table 4 Column (2) demonstrate a significantly positive coefficient (0.003) at the 1% level, confirming Hypothesis 3 that fintech promotes GTFEE through fostering technological innovation.

The underlying mechanisms operate through multiple pathways. First, fintech adoption directly stimulates corporate R&D investment by improving capital allocation efficiency. Second, blockchain-based intellectual property protection and AI-driven innovation analytics help optimize resource allocation for green technology development. Third, fintech platforms facilitate risk assessment and management for innovation projects, while policy-oriented fintech solutions (e.g., green credit scoring) strengthen institutional support. These synergistic effects enhance firms’ capacity to develop and implement energy-saving technologies, clean production processes, and circular economy solutions, all of which are critical drivers for improving GTFEE through the technological innovation channel.

4.4. Heterogeneity Analysis

4.4.1. Regional Heterogeneity Analysis

To examine whether the impact of fintech on GTFEE varies geographically, we partition the sample into China’s eastern, central, and western regions. Columns (1)–(3) in

Table 5. Heterogeneity test I: regional heterogeneity analysis.

	(1) Eastern	(2) Central	(3) Western
	GTFEE	GTFEE	GTFEE
fintech	0.022 ***	0.004	0.006
	(10.54)	(1.52)	(0.85)
IS	0.497 ***	0.359 ***	0.612 ***
	(13.07)	(9.02)	(3.33)
UR	−0.240 ***	0.159 ***	1.751 ***
	(−9.48)	(5.05)	(3.14)
ERS	0.136 ***	0.014	−0.018 *
	(15.38)	(1.03)	(−1.87)
FDI	0.037 ***	−0.044 ***	−0.037
	(6.66)	(−3.10)	(−1.24)
INV	0.018 ***	0.002 **	0.009
	(18.22)	(2.00)	(1.13)
SEC	−0.155 ***	0.021	0.778 ***
	(−8.09)	(1.26)	(10.66)
GPPH	−0.499 ***	−0.048 ***	−0.055
	(−25.50)	(−3.02)	(−0.96)
EPE	−0.031 ***	0.002 **	−0.010 **
	(−48.15)	(2.47)	(−2.51)
_cons	1.395 ***	0.153 ***	−1.512 ***
	(28.28)	(4.04)	(−4.28)
Firm fixed effect	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes
Chow Test	253.685
p-Value	0
Observations	8821	3116	198
R2	0.887	0.893	0.979
AR2	0.886	0.890	0.977

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

present the heterogeneous effects of it respectively. The Chow test statistic of 253.685 confirms statistically significant spatial heterogeneity in the marginal contributions of fintech to GTFEE across regions, ultimately forming differentiated energy efficiency optimization pathways.

(1) Eastern region

The regression results in Column (1) show a coefficient of 0.022, significant at the 1% level, indicating fintech significantly enhances GTFEE in eastern China. This can be attributed to several structural advantages. First, as China’s most economically developed region, the eastern area features mature industrial structures dominated by advanced manufacturing and modern services. Second, its well-developed financial ecosystem and higher fintech adoption enable more effective use of big data and blockchain technologies to lower financing barriers for green projects. Third, the concentration of high-tech enterprises and green technology R&D centers strengthens the synergistic effects between fintech and green industries [49].

(2) Central and western region

Column (2) reports a coefficient of 0.004, suggesting a positive but insignificant fintech-GTFEE relationship. This moderated effect reflects the region’s transitional development stage, where national policies like the “Guidelines on Promoting High-quality Development in Central China” have established institutional support for manufacturing upgrading and ecological protection. At the same time, central fiscal transfers and special funds for technological innovation provide foundational support [50].

The insignificant coefficient (0.006) in Column (3) reveals a limited fintech impact despite Western Development Strategy implementation. Three constraints explain this. (i) resource-dependent industrial structures with limited high-value sectors, (ii) inadequate targeted policy support for fintech-enabled green transitions, and (iii) technological and institutional bottlenecks in policy execution. These structural factors constrain fintech’s potential to improve energy efficiency [51].

The comparative analysis demonstrates a transparent regional gradient. Fintech’s GTFEE enhancing effect is strongest in the east (0.022 ***), while statistically insignificant in the center and the west, reflecting regional disparities in financial development, industrial structure, and policy effectiveness.

4.4.2. Stages of Economic Development Heterogeneity Analysis

The heterogeneity in economic development stages significantly moderates the intensity and direction of the impact of fintech on GTFEE [52]. Based on the median ratio of tertiary industry added value to secondary industry added value, this study divided the sample into service cities and industrial cities for grouped regression analysis [30]. The results, as shown in

Table 6. Heterogeneity test II: economic development heterogeneity analysis.

	(1) Service Cities	(2) Industrial Cities
	GTFEE	GTFEE
fintech	0.023 ***	0.016 ***
	(7.47)	(6.59)
IS	0.319 ***	−0.035
	(6.17)	(−0.29)
UR	−0.188 ***	0.004
	(−6.76)	(0.27)
ERS	0.072 ***	0.001
	(10.18)	(0.07)
FDI	0.013 **	0.056 ***
	(2.23)	(3.75)
INV	0.011 ***	−0.002
	(11.00)	(−1.06)
SEC	0.129 ***	−0.079 *
	(4.56)	(−1.90)
GPPH	−0.257 ***	−0.018
	(−8.11)	(−1.51)
EPE	−0.029 ***	0.002 ***
	(−8.97)	(3.04)
_cons	0.754 ***	0.394 ***
	(10.67)	(4.86)
Firm fixed effect	Yes	Yes
Year fixed effect	Yes	Yes
Chow Test	993.043
p-Value	0
Observations	8345	2198
R2	0.873	0.926
AR2	0.870	0.921

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

, indicate that fintech development exerts a significant positive effect on the GTFEE of both city types. A Chow test statistic of 993.043 confirms a significant difference between the groups. The regression coefficients demonstrate that the enhancing effect is more pronounced in service cities (evidenced by significantly higher coefficient estimates). Industrialization-stage economies rely on energy-intensive manufacturing where production-centric operations, technological inertia, and high retrofit costs constrain efficiency gains. Conversely, servitization-stage service-sector dominance enables greater technological adaptability. Green finance facilitates funding for eco-equipment and operational upgrades (e.g., tourism sector improvements as demonstrated by Zeng et al., 2024, [17]), while fintech leverages big-data analytics for precision energy management. This dual-action synergy, financial enablement, and technological optimization capitalize on service-sector characteristics to maximize GTFEE improvements where services prevail [29].

4.4.3. Resource Endowment Heterogeneity Analysis

Resource endowment constitutes a critical factor influencing GTFEE. Following the classification framework established in the National Plan for Sustainable Development of Resource-Based Cities (2013–2020) (Zhang and Sun, 2025) [29], we divide our sample into resource-based and non-resource-based cities to examine potential heterogeneity. The grouped regression results presented in

Table 7. Heterogeneity test III: resource endowment heterogeneity analysis.

	(1) Resource-Based	(2) Non-Resource-Based
	GTFEE	GTFEE
fintech	−0.004	0.028 ***
	(−0.89)	(8.43)
IS	0.322 ***	0.371 ***
	(5.05)	(4.97)
UR	−0.038 *	−0.319 ***
	(−1.75)	(−9.07)
ERS	−0.009	0.122 ***
	(−0.57)	(8.12)
FDI	−0.014	0.031 ***
	(−1.09)	(3.09)
INV	−0.005 ***	0.013 ***
	(−2.85)	(13.96)
SEC	0.021	−0.012
	(0.96)	(−0.52)
GPPH	−0.049 ***	−0.413 ***
	(−3.13)	(−9.96)
EPE	0.003 ***	−0.025 ***
	(2.69)	(−8.07)
_cons	0.257 ***	1.198 ***
	(5.82)	(15.67)
Firm fixed effect	Yes	Yes
Year fixed effect	Yes	Yes
Chow Test	124.347
p-Value	0
Observations	1660	10,420
R2	0.901	0.876
AR2	0.896	0.874

Note: t statistics in parentheses. * p < 0.1, *** p < 0.01.

indicate that the development of fintech exerts a significant positive effect on GTFEE only in non-resource cities. A Chow test statistic of 124.347 further confirms the significant difference between the groups. Clearly, the regression coefficient in Column (2) is significantly positive at the 1% level for non-resource cities; the divergent outcomes stem from industrial heterogeneity. Resource-based cities exhibit concentrated industrial structures focused on extraction and primary processing (Wu and Xu, 2025) [53], characterized by high capital intensity, limited technological sophistication, and entrenched resource dependency. This uniformity constrains energy efficiency improvements while reducing firms’ incentives and capacity to leverage fintech for innovation, thereby limiting GTFEE gains.

Conversely, non-resource cities’ diversified industrial bases, spanning manufacturing, services, and high-technology sectors, possess greater innovative capacity and technology/financial absorption capabilities. High-tech industries particularly harness fintech for energy optimization and green finance for R&D funding, driving significant GTFEE enhancement through this synergy. He et al. (2023) [54] corroborate this pattern, showing more substantial pollution-carbon reduction synergies from green finance in non-resource cities (Zhang and Sun, 2025) [29].

4.4.4. Urban Scale Heterogeneity Analysis

City size heterogeneity manifests in distinct patterns of energy consumption, technological advancement, and digitalization strategy implementation. Owing to their intricate economic structures and elevated population densities, large cities tend to adopt more advanced technologies and pursue comprehensive digital transformations, enhancing energy efficiency and addressing environmental pressures. Conversely, small and medium-sized cities often encounter significant barriers in the energy efficiency process, constrained by geographic isolation, capital limitations, and talent deficits. Regression analyses stratified by city tier reveal significant heterogeneity in the relationship between fintech and GTFEE (Bie et al., 2024) [30]. The results, as shown in

Table 8. Heterogeneity test Ⅳ: urban scale heterogeneity analysis.

	(1) Tier 1 Cities	(2) Tier 2 Cities	(3) Tier 3 Cities	(4) Tier 4 Cities	(5) Tier 5 Cities
	GTFEE	GTFEE	GTFEE	GTFEE	GTFEE
fintech	0.018 ***	0.005 **	0.009 ***	0.010 ***	−0.015
	(4.37)	(2.29)	(3.34)	(4.05)	(−0.84)
IS	0.775 ***	−0.183 ***	0.086 *	0.122 ***	0.872 ***
	(11.72)	(−4.64)	(1.92)	(3.36)	(3.03)
UR	−0.229 ***	0.207 ***	0.041	0.008	−0.699 *
	(−7.40)	(5.73)	(1.16)	(0.28)	(−1.70)
ERS	0.236 ***	0.173 ***	0.023 *	−0.018	0.234 **
	(9.98)	(17.17)	(1.67)	(−1.22)	(2.09)
FDI	0.090 ***	0.035 ***	0.051 ***	0.014	−0.022
	(6.36)	(4.57)	(4.47)	(1.54)	(−0.20)
INV	0.013 ***	0.012 ***	−0.004 **	0.001	−0.003
	(9.17)	(11.87)	(−2.19)	(0.49)	(−0.31)
SEC	0.563 ***	−0.226 ***	−0.204 ***	0.007	0.439 *
	(14.09)	(−13.35)	(−9.85)	(0.31)	(1.77)
GPPH	−0.855 ***	−0.397 ***	−0.127 ***	−0.039 *	0.156
	(−13.19)	(−19.27)	(−5.55)	(−1.80)	(0.06)
EPE	−0.036 ***	0.004 ***	−0.006 ***	0.002 **	0.008
	(−32.86)	(6.26)	(−4.84)	(1.97)	(1.24)
_cons	1.674 ***	1.166 ***	0.651 ***	0.283 ***	−0.353
	(12.90)	(22.09)	(13.26)	(6.70)	(−0.09)
Firm fixed effect	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes
Chow Test	1087.912
p-Value	0
Observations	4380	4200	2149	1193	176
R2	0.934	0.801	0.769	0.903	0.835
AR2	0.934	0.799	0.761	0.897	0.791

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

, reveal pronounced heterogeneity in how fintech development affects GTFEE across cities of different sizes, with distinct patterns emerging based on urban scale characteristics. Large cities (Tiers 1–4 classified as Tiers 1–4 following the China City Statistical Yearbook 2019) demonstrate significant positive coefficients, indicating fintech effectively enhances their energy efficiency. This aligns with their inherent advantages, including mature digital infrastructure, robust policy frameworks, and agglomeration economies that facilitate technology diffusion and implementation. The Chow test statistic (1087.912) confirms these inter-tier differences are statistically significant.

Conversely, Tier 5 cities show an insignificant relationship, reflecting the challenges smaller urban centers face in translating fintech advancements into energy efficiency gains. Their geographic isolation, capital constraints, and talent shortages create implementation barriers that outweigh potential benefits. These findings underscore the critical moderating role of city size on the fintech and GTFEE relationship, suggesting that policy interventions must be tailored to urban development levels, leveraging existing advantages in larger cities while addressing structural limitations in smaller ones.

4.5. Robustness Tests

4.5.1. Endogeneity Tests

To mitigate potential endogeneity concerns in the baseline regression, such as omitted variable bias and reverse causality, this study employs an instrumental variable (IV) approach for additional robustness verification. It must be acknowledged that the exogeneity assumption of any instrumental variable can never be fully verified. Although the geographic instrument used in this study theoretically satisfies the relevance condition, its exogeneity may still be questioned. Therefore, the results of the instrumental variable approach are treated as supplementary, supportive evidence rather than the sole basis for causal identification. We utilize the average fintech development level of other cities (afintech) as the instrumental variable for local fintech. This selection satisfies the necessary conditions for a valid instrument based on two key rationales: (1) While local fintech conditions influence local GTFEE, it remains theoretically uncorrelated with the average fintech level of other cities; (2) Given the spatial spillover effects characteristic of fintech diffusion, where technological advancements typically propagate from high-adoption to low-adoption regions, the instrument demonstrates both relevance and exogeneity.

The constructed instrument, AFINTech, represents the annual average fintech values across peer cities. IV regression results presented in

Table 9. Robustness test I: endogeneity regression results.

	(1)	(2)
	fintech	GTFEE
afintech	−247.674 ***
	(−222.941)
IS	−0.334 ***	0.252 ***
	(−5.252)	(7.826)
UR	−0.068	−0.037
	(−1.498)	(−1.611)
ERS	−0.136 ***	0.122 ***
	(−6.097)	(10.831)
FDI	−0.115 ***	0.051 ***
	(−7.406)	(6.499)
INV	0.019 ***	0.018 ***
	(10.421)	(20.033)
SEC	0.022	−0.142 ***
	(0.599)	(−7.653)
GPPH	0.320 ***	−0.127 ***
	(7.978)	(−6.221)
EPE	−0.008 ***	−0.026 ***
	(−6.411)	(−39.921)
fintech		0.027 ***
		(12.053)
_cons	480.912 ***	0.531 ***
	(223.132)	(12.487)
Firm fixed effect	Yes	Yes
Year fixed effect	Yes	Yes
Observations	12,137	12,137
R2	0.983	0.009
Cragg-Donald Wald F	35,569.43

Note: t statistics in parentheses. *** p < 0.01.

confirm the instrument’s validity. The Cragg-Donald Wald F-statistic (16.38) substantially exceeds the 10% critical value threshold, effectively rejecting the weak instrument hypothesis. Most importantly, the IV estimates maintain statistical significance and directional consistency with our baseline findings, thereby robustly confirming the study’s primary hypothesis even after accounting for endogeneity concerns.

4.5.2. Explained Variable Replacement

To assess the robustness of the baseline regression results, this study substitutes the dependent variable in Equation (7), with the results shown in

Table 10. Robustness test II: explained variable replacement.

	(1)
	lnsuperccr
fintech	0.039 **
	(2.14)
IS	−0.341
	(−1.17)
UR	−0.004
	(−0.03)
ERS	0.142 **
	(2.02)
FDI	0.046
	(0.71)
INV	0.015 **
	(2.36)
SEC	−0.376 *
	(−1.97)
GPPH	−0.477 **
	(−2.46)
EPE	0.002
	(0.28)
_cons	0.685
	(1.49)
Firm fixed effect	Yes
Year fixed effect	Yes
Observations	12,135
R2	0.898
AR2	0.896

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05.

. Specifically, we modify the measurement of GTFEE. Unlike the SBM approach, we employ the super-efficiency Charnes-Cooper-Rhodes (CCR) model to calculate GTFEE. This model evaluates energy utilization efficiency within the traditional DEA framework by incorporating environmental constraints (e.g., industrial SO₂ emissions, soot/dust emissions, and wastewater discharge), considering both economic output and environmental impact.

4.5.3. Explanatory Variable Replacement

Existing scholars predominantly measure financial technology (fintech) using either data mining techniques or the Digital Financial Inclusion Index developed by Peking University’s Financial Research Center [55]. While the preceding analysis employed the data mining approach, this section adopts the alternative measurement system. As demonstrated in

Table 11. Robustness test III: explanatory variable replacement.

	(1)	(2)	(3)
	GTFEE	GTFEE	GTFEE
index _aggregate	0.001 ***
	(4.63)
IS	0.263 ***	0.278 ***	0.272 ***
	(9.06)	(9.76)	(9.38)
UR	−0.261 ***	−0.350 ***	−0.221 ***
	(−11.68)	(−15.70)	(−9.72)
ERS	0.129 ***	0.099 ***	0.134 ***
	(12.50)	(9.74)	(13.04)
FDI	0.056 ***	0.045 ***	0.061 ***
	(7.75)	(6.49)	(8.37)
INV	0.017 ***	0.015 ***	0.016 ***
	(20.25)	(19.27)	(19.00)
SEC	−0.088 ***	−0.087 ***	−0.089 ***
	(−5.17)	(−5.24)	(−5.29)
GPPH	−0.233 ***	−0.288 ***	−0.233 ***
	(−12.06)	(−15.06)	(−12.11)
EPE	−0.027 ***	−0.028 ***	−0.027 ***
	(−47.44)	(−49.01)	(−47.61)
coverage _breadth		0.005 ***
		(21.81)
usage _depth			0.001 ***
			(7.61)
_cons	0.818 ***	0.209 ***	0.799 ***
	(15.13)	(3.84)	(16.61)
Firm fixed effect	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes
Observations	12,135	12,135	12,135
R2	0.873	0.878	0.873
AR2	0.871	0.876	0.871

Note: t statistics in parentheses. *** p < 0.01.

, the regression results indicate that the Digital Financial Inclusion Index, coverage breadth of digital finance, and usage depth of digital finance all show statistically significant positive coefficients at the 1% level. These findings confirm that the fintech’s positive effect on GTFEE remains robust after substituting the explanatory variables.

4.5.4. Model Replacement

To further test the robustness of the conclusions, this study employs quantile regression with an unbalanced panel. This method allows for examining the marginal effects of the explanatory variables at different positions of the dependent variable’s distribution, which not only mitigates the interference of outliers but also reveals distributional heterogeneity. The results, as shown in

Table 12. Robustness test IIII: model replacement.

Variable	Q(0.25)	Q(0.50)	Q(0.75)	Q(0.90)
Fintech	0.0148 (0.0124)	0.0085 (0.0591)	0.0318 * (0.0169)	0.0038 *** (0.0013)
is	−0.1331 (0.0652)	−0.0215 (0.5316)	−0.1131 (0.1987)	−0.0894 *** (0.0183)
ur	0.0895 (0.0714)	0.2777 (0.2056)	0.4071 *** (0.0649)	0.6368 *** (0.0065)
sec	−0.1797 *** (0.0505)	−0.2167 *** (0.0512)	−0.6200 *** (0.1894)	−0.2888 *** (0.0017)
ers2	−0.0663 * (0.0401)	−0.0416 * (0.0242)	0.0498 * (0.0299)	2.6000 *** (0.0303)
environmentalprotection	0.0007 (0.0458)	−0.0002 (0.0182)	−0.0128 (0.0107)	−0.0134 *** (0.0006)
inv2	0.0005 (0.0039)	−0.0052 (0.0108)	0.1156 *** (0.0331)	−0.0069 *** (0.0001)
fdii	−0.0534 ** (0.0236)	−0.0607 (0.0945)	−0.1142 (0.1632)	−0.0313 *** (0.0051)
lngreenpatents	0.1441 *** (0.0358)	0.0746 (0.0742)	0.0059 (0.0195)	−0.1224 *** (0.0034)
City FE	YES	YES	YES	YES
Time FE	YES	YES	YES	YES
Observations	12,137	12,137	12,137	12,137

Note: t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

, differ from the baseline regression. When GTFEE is at the higher quantiles of 0.75 and 0.90, the positive effect of Fintech becomes stronger. At the relatively lower quantiles of 0.5 and 0.25, the effect is less significant but remains positive. This suggests that the promoting effect of Fintech may not apply to the group of cities with the lowest efficiency.

5. Discussion

5.1. Conclusions

This study integrates fintech and GTFEE within a unified framework. By identifying fintech-related keywords and utilizing the Baidu News Advanced Search page to quantify fintech, employing the SBM-ML index model incorporating undesirable outputs to measure Chinese urban GTFEE from 2011 to 2021, and constructing a two-way fixed effects model to examine the impact of fintech on urban GTFEE empirically, we conclude that fintech can significantly enhance GTFEE. The core conclusion of this study—that fintech significantly promotes urban GTFEE—is consistent with the findings of Teng and Shen (2023) [20] in OECD countries and Wu et al. (2024) [28] in the context of Chinese regional efficiency. This consistency suggests that fintech’s role in enhancing energy efficiency may be generalizable across different economic and geographic contexts. Furthermore, fintech exerts a negative impact on financing constraints, thereby enhancing GTFEE through the alleviation of these constraints. Additionally, fintech positively influences technological innovation, consequently improving GTFEE specifically through mechanisms such as heightened investment efficiency and optimized resource allocation. Heterogeneity analysis reveals that the magnitude of fintech’s impact on GTFEE exhibits regional heterogeneity. Fintech development in Eastern China demonstrates a significant driving effect on GTFEE; conversely, fintech development in Central and Western China exhibits no statistically significant association with GTFEE improvement. Significant resource-based heterogeneity exists, and fintech exerts a significantly positive impact on GTFEE exclusively in non-resource-based cities. Economic development stage heterogeneity is observed. The influence of fintech on GTFEE is notably greater in service cities. Furthermore, urban scale heterogeneity is evident. Fintech significantly affects GTFEE only in larger cities. The heterogeneous effects uncovered in this study offer critical nuances to the existing literature. The stronger effect in Eastern and non-resource-based cities underscores the importance of existing economic structure and digital infrastructure in amplifying fintech’s benefits. This supports the contention of Shi and Yang (2024) that synergy with advanced industrial and regulatory environments is essential [27]. The insignificant effect in Central and Western and resource-dependent cities mirrors the findings of Hou et al. (2024) in industrial Asian economies, where fintech sometimes stimulates energy-intensive sectors [31]. This divergence highlights the context-dependent nature of fintech impacts and suggests that without complementary green industrial policies, fintech alone may not suffice to drive GTFEE under all conditions. The empirical results remain robust after undergoing rigorous robustness checks, including the instrumental variables approach, alternative explanatory variables, and modified dependent variable specifications.

5.2. Policy Implications

Enhancing GTFEE not only improves energy utilization effectiveness, alleviating current challenges of resource scarcity and finite non-renewable energy, but also addresses pressing environmental issues, thereby contributing significantly to energy conservation and environmental protection goals. Notably, GTFEE can be substantially elevated through fintech pathways. The conclusions of this study provide a theoretical foundation for leveraging fintech models to improve financing constraints and promote technological innovation, consequently enhancing GTFEE. Therefore, the policy implications derived from this research are as follows.

First, governments should promote deep integration between FinTech and GTFEE. Given the rapid advancement of contemporary technology, governments should align with the overarching trend of integrated development between fintech and the green, high-efficiency energy industry. Regulators should establish dedicated “Fintech–GTFEE” innovation zones in cities with mature digital infrastructure (e.g., Eastern Chinese cities and large service-oriented urban centers). Within these zones, they should introduce standardized API interfaces for green project financing, support the development of fintech-powered platforms for real-time monitoring of energy efficiency performance, and offer preferential licensing for green fintech products that meet national energy-saving standards.

Second, governments should implement targeted fiscal policies to alleviate financing constraints. In Eastern and large cities: Launch tax credit programs explicitly tied to corporate investments in fintech-enabled green innovations, such as deductions for IoT-based energy management systems. In Central and Western regions: Provide direct subsidies to fintech firms that extend green credit to small-sized energy enterprises and set up regional green guarantee funds to de-risk fintech lending. In non-resource-based cities: Issue special bonds for digital infrastructure that supports fintech and energy efficiency collaborations.

Third, optimizing technological innovation ecosystems requires coordinated policy frameworks. Governments may implement measures such as tax credits for R&D investments by green technology innovation firms and tax reductions for their innovative products. Concurrently, fostering collaboration between fintech companies, universities, and research institutions is essential to conducting green technology innovation research jointly. This accelerates the translation and application of scientific and technological achievements, improves the efficiency of green technology innovation, and thereby empowers GTFEE improvement.

Fourth, strategic spatial policies are critical to harness fintech’s industrial growth potential. Policymakers should prioritize infrastructure and talent development in western regions while sustaining central area support, with funding allocated based on regional readiness metrics. Public–private partnerships and continuous evaluation are crucial for reducing regional disparities. To capitalize on their inherent advantages in industrial diversity and innovation, non-resource-based cities are advised to champion collaborative initiatives between fintech and green finance. A viable approach involves regulatory incentives for banks and other financial institutions to launch specialized divisions or fintech–green finance innovation labs. Such dedicated units would accelerate the prototyping and market diffusion of sustainable financial products. Concurrently, cities with economies dominated by industry should focus on incubating green fintech solutions tailored for sectors like logistics, retail, and hospitality. To operationalize this, municipal governments could embed fintech–green finance convergence targets into their broader service industry development blueprints, specifying key application areas and innovation priorities to spur ecological modernization within the sector. Lastly, the uniqueness of regional development should be taken into account. This means that policy measures should be differentiated based on urban energy efficiency levels: high-efficiency cities should focus on breakthroughs in cutting-edge technologies, while low-efficiency cities should prioritize addressing institutional and capacity gaps. Cross-regional collaboration should be strengthened to facilitate technology diffusion, ultimately achieving overall improvement in energy efficiency.

5.3. Limitations and Future Research Directions

This study has several limitations that also point toward fruitful future research:

Methodological constraints: While this research employs panel fixed-effects models and instrumental variables, it does not account for potential spatial spillover effects between cities. Future studies could incorporate spatial econometric models, such as the Spatial Durbin Model (SDM), to examine cross-city technological and financial diffusion. Additionally, applying Difference-in-Differences (DID) designs using pilot policy shocks could strengthen causal identification.

Measurement challenges: The reliance on Baidu News keyword search volume for fintech measurement may introduce media bias and fail to capture on-the-ground financial innovation activity. Subsequent research could utilize alternative indicators, such as firm-level fintech patent filings or transaction data from digital financial platforms, to enhance validity.

Data timeliness and uniqueness: The dataset covers cities up to 2021 and draws primarily from public statistical yearbooks. Future work could extend the time frame and incorporate high-frequency, non-traditional data (e.g., satellite light data, firm-level energy consumption) to improve granularity and timeliness.

Endogeneity concerns: Although instrumental variables were used, the exclusion restriction of the geographic instrument (distance to Hangzhou) may be imperfect. Future research could exploit quasi-experimental settings, such as national fintech pilot zones, or use machine learning methods to control unobserved confounders better.

Context-dependent findings: The unexpected negative sign of urbanization may stem from multicollinearity or region-specific development patterns. Further investigation into nonlinear relationships or subgroup interactions would help clarify this anomaly.

Theoretical generalizability: As many studies have examined fintech and energy efficiency, future research should strive to propose novel mechanisms—such as behavioral interventions, supply chain linkages, or political economy channels—to advance theory building.

In summary, while this study provides evidence at the city level in China regarding fintech’s role in promoting GTFEE, its limitations highlight opportunities for methodological innovation and deeper theoretical inquiry in future work.