Economic Analysis and Policy 82 (2024) 1113–1130

Contents lists available at ScienceDirect

Economic Analysis and Policy

journal homepage: www.elsevier.com/locate/eap

Analyses of Topical Policy Issues

Can digital financial inclusion promote the coupling coordination

between pollution reduction and low-carbon development?
Evidence from China
Yuxi Zhang a, b, Adrian (Wai Kong) Cheung a, *, Xiaodong Qu a
a
Faculty of Finance, City University of Macau, Taipa, Macao
b
School of Business, Shanghai Normal University Tianhua College, Shanghai, China

A R T I C L E I N F O A B S T R A C T

JEL classification: Pollution reduction and low-carbon development are two specific but intertwined goals of
G28 addressing climate change and sustainable development of the United Nations in 2030, and the
Q50 effective use of digital financial inclusion is imperative for driving technical innovation,
Q58
upgrading industrial structure, and addressing climate change. Therefore, this paper investigates
Keywords: the impact of digital financial inclusion on the coupling coordination between pollution reduction
Digital financial inclusion
and low-carbon development using a panel dataset of 30 China’s provinces from 2011 to 2020.
Pollution reduction and low-carbon develop­
The results show that the coupling coordination between pollution reduction and low-carbon
ment
Coupling coordination degree model development in China improves significantly during the sample period. The synergistic effect
Mediating effect of different economic zones and provinces varies greatly in space and time. In addition, digital
financial inclusion has a promoting effect on the coupling coordination of pollution and carbon
emission reduction. A mechanism analysis also reveals that technological innovation, industrial
structure and energy consumption are the core transmission channels through which digital
financial inclusion influences the synergistic effect of pollution and carbon emission reduction.
This suggests that the need of coordinated governance by policymakers is warranted, and digital
financial inclusion is an important way to this coordinated governance.

1. Introduction

Along the course of industrialization and urbanization, global climate change and air pollution have become serious challenges in
the world. In 2022, the World Health Organization has suggested that hazardous air has still been breathed by a billion individuals
worldwide, particularly in developing nations. Long-term exposure to low levels of ambient air pollution will have a negative impact
on people’s health and quality of life in addition to impeding social and economic growth (Costello et al., 2009). China, the largest
developing nation, has taken a number of actions to enhance the quality of the environment. (Wang et al., 2021b). Since greenhouse
gas emissions have the same origin, source, and manner, it is viable to regulate them with the same frequency, impact, and pathway. As
a blueprint, China’s Ministry of Ecology and Environment launched the Action Plan for Pollution Reduction and Carbon Reduction in
June 2022. Lessons from developed countries show that co-benefits between air pollution and climate change can amplify the effects of
policies and lower operational expenditures. While there has been a significant amount of research exploring the impact of various

* Corresponding author.
E-mail address: adriancheung@cityu.mo (A.(W.K. Cheung).

https://doi.org/10.1016/j.eap.2024.05.007
Received 28 July 2023; Received in revised form 17 April 2024; Accepted 8 May 2024
Available online 9 May 2024
0313-5926/© 2024 Economic Society of Australia, Queensland. Published by Elsevier B.V. All rights are reserved, including those for text and data
mining, AI training, and similar technologies.
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

financial activities on environmental quality (see, e.g. Tao et al. 2023), there has been a lack of research on the relationship between
digital financial inclusion and the cooperation between pollution control and low-carbon growth. To address this gap in the literature,
our study investigates the effects of digital financial inclusion on the degree of cooperation between pollution reduction and carbon
reduction.
Economists have expressed different views on the impact of digital financial inclusion on the environment. On the one hand, the
fast growth of internet infrastructure may increase the use of non-renewable energy sources in cities, which could accelerate CO2
emissions. For example, improved access to financial services may raise the amount of industrial and manufacturing activities at firm
level. Increasing these activities may also result in energy poverty, hastening the rise of CO2 emissions (Zhao et al., 2021). Similarly,
digital financial inclusion also results in the greater availability of consumer goods and thus consumers are more likely to acquire
energy-hungry electrical items, including cars and air conditioners; this increases energy consumption and endangers the environment
(Le et al., 2020). In addition, the environmental effect increases with decreasing urbanization levels (Shafiei and Salim, 2014). On the
other hand, digital financial inclusion enables those in undeveloped, impoverished areas to get access to money to buy renewable
energy, including solar energy. Compared to fossil fuels, renewable energy will be more eco-friendly (Baulch et al., 2018). At the
business level, digital financial inclusion supports the development of sustainable energy and aids in product enhancement, all of
which contribute to lower air pollution (Milliman and Prince, 1989). At the city level, digital financial inclusion can ease corporate
financing restrictions, assist small and medium-sized businesses that previously found it difficult to secure financing (Li et al., 2024),
and give financial assistance for enhancing pollution control and energy-saving production equipment, therefore supporting the
optimization and upgrading of industrial structure and improving the environment. The link between air pollution and digital finance
is reversed at the city level in China; in particular digital finance can promote environmental optimization.
Pollution reduction and carbon reduction are both distinct and interdependent subsystems. Fossil fuels are widely used in human
production and daily life, and their combustion and utilization process will affect both air quality and low-carbon development. On the
one hand, improving air quality necessitates pollutant reduction, but mitigating climate warming requires carbon reduction.
Numerous pollutant emission such as SO2, NOx, and particulate matter, affect the ambient air quality. The simultaneous greenhouse
gases emission of carbon dioxide, on the other hand, accelerates climate warming. In addition, the two subsystems are interconnected.
The emissions of greenhouse gas and air pollutant share the same origin, source, and method in this process. We need to examine from
both pollution reduction and carbon reduction perspectives to better assess the green development of enterprises. Prior work often
focuses on the effects of digital financial inclusion on carbon emissions or other pollutant emissions, separately, without paying
attention to the spillover effect from one emission to others and other possible interactions between carbon and pollution emissions.
Yet, these effects cannot easily be identified separately due to the subsystems’ complexity. Thus, it is important to determine how much
of a synergy exists between the two.
"Coupling" is a physical process, and the degree of coupling coordination refers to the combined value of the interaction of two or
more systems. To measure coupling, we propose the coupling coordination degree model (CCDM) to examine the interaction between
pollution reduction and low-carbon development because it can effectively support the coordinated evolution of regional population,
resources, environment, and development (Liu et al., 2021b) and shows good performance in terms of reducing uncertainty caused by
arbitrary weight assignment. CCDM has been used to model the interaction between the environment and climate (Reichler and Kim,
2008) and between the environment and urbanization (Song et al., 2018).
The paper investigates the impact of digital financial inclusion on the degree of cooperation between pollution reduction and low-
carbon growth (CCDPL), with the following main contributions. Firstly, the linkage between digital financial inclusion and the degree
of cooperation between pollution reduction and low-carbon growth (CCDPL) is important but little is known about this linkage in the
literature. The existing research on co-emission reduction uses SO2 and CO2 as indicators of pollution and carbon emissions,
respectively (Pacca et al., 2020; Shahbaz et al., 2022). Yet little thought has been given to how closely pollution control and low-carbon
growth should be coordinated. This paper is the first to examine this issue using the CCDM for a pollution reduction-low carbon
emission system (PCS) that accounts for their connections and/or interactions. The result shows that one standard deviation (0.67)
increase in digital financial inclusion is associated with 2 % (=0.67×0.031×100 %) increase in the coupling coordination. It indicates
that the influence of digital inclusive finance on the coupling coordination between the reduction of pollution and low-carbon
development is significant in both statistical and economic sense. This empirical evidence suggests that the need of coordinated
governance by policymakers is warranted.
Secondly, we contribute to the literature of the relationship between digital financial inclusion and coupling coordination by
examining several factors: innovation effect, structural effect, and consumer preference. We provide strong evidence that digital
financial inclusion has a complete effect on the coupling coordination for these mechanisms. This finding diverges from Bui (2020).
They contend that when the financial system develops, there is a corresponding increase in energy demand and pollution emissions,
both of which may worsen environmental harm.
Finally, we present new evidence that the association between digital financial inclusion and CCDPL can be moderated by unequal
regional economic development, regional natural conditions, and digital financial inclusion policies. The result shows that there are
regional heterogeneity, development heterogeneity, and digital financial inclusion policy heterogeneity, and their effects are more
pronounced in the central and western regions, high coordination degree and the area that promotes digital financial inclusion,
respectively.
The remainder of the paper is structured as follows: Section 2 provides a systematic assessment of relevant research and evaluates
the influence of digital financial inclusion on the coordinated development of pollution and carbon emission reduction from a
theoretical standpoint and proposes research hypotheses.
The choice of variables/indicators, empirical model, and the data sources are presented in Section 3. Section 4 presents the

1114
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

empirical test results, including baseline regression, mechanism analyses, robustness test, endogeneity test, and heterogeneity ana­
lyses. The study’s main results are summarized, discussed, and policy recommendations are put forth in Section 5.

2. Literature review and hypothesis development

A new emblem in China’s policy agenda is the integrated control of air pollution and climate change, which is also a very significant
worldwide issue (Ramanathan and Feng, 2009). Digital financial inclusion may offer the society more convenient and secure banking
services (Ozili, 2018), which could be a crucial economic support for the improvement of the ecological environment for the following
reasons. First, digital financial inclusion can alleviate corporate financing constraints and financial mismatches (Lu et al., 2022), which
in turn promotes green innovation (Xue and Zhang, 2022), reduces industrial structure distortion, and increases the proportion of clean
energy consumption (Li et al., 2023). In other words, the implications of digital financial inclusion are not confined to technology
sector, industrial structure, and consumer preference are also affected (Huang et al., 2023b; Zhou et al., 2024). Among them, tech­
nological effects can enable European countries to produce renewable energy through trade opening, which is conducive to envi­
ronmental sustainability. By lowering final demand across economic sectors, structural impacts can meet emission reduction
objectives. Through factors such as household wealth (Barrington-Leigh et al., 2019) and the coordination and capability of gov­
ernment agencies, the consumer choice effect creates co-benefits between greenhouse gas emissions and air pollution. This view
suggests that the changes brought about by the development of digital inclusive finance are precisely in line with the requirements of
technological innovation, industrial structure upgrading, and energy consumption substitution required in the process of coordinated
governance of air pollution and climate change. It is expected to have a significant impact on the coupling and coordination of
pollution reduction and carbon reduction. Therefore, we set Hypothesis 1a as follows:
Hypothesis 1a. Digital financial inclusion positively affects the coupling coordination degree between pollution reduction and low-
carbon development.
However, to make efficient use of digital financial inclusion technology, enterprises need to invest in equipment in the early stage.
The intelligent transformation of traditional office infrastructure may lead to increase carbon emissions. Simultaneously, intelligent
transformation expands firms’ access to financial services. Improving access to financial services may raise the amount of industrial
and manufacturing activities. Increases in these activities may also result in energy poverty, hastening the rise of CO2 emissions (Zhao
et al., 2021). Government may also expand early-stage investment in internet infrastructure to develop digital inclusive finance and
make financial resources more accessible and helpful to the general population. The rapid growth of internet infrastructure may in­
crease the use of non-renewable energy sources in cities, which could accelerate CO2 emissions. As a result of the greater availability of
consumer goods via digital financial inclusion, consumers are more likely to acquire energy-hungry electrical items, including cars and
air conditioners, and this increases energy consumption and endangers the environment (Le et al., 2020). Therefore, we set Hypothesis
1b as follows:1
Hypothesis 1b. Digital financial inclusion negatively affects the coupling coordination degree between pollution reduction and low-
carbon development.

3. Research design

3.1. Key variables

3.1.1. Coupling coordination degree measure (CCDM) construction

We specify a coupling coordination degree measure (CCDM) for a pollution reduction-low carbon emission system (PCS) that
captures both pollution reduction and low-carbon development as follows.
Firstly, we use the index of CO2 emissions per capita to describe the carbon emission and Ei to describe the amount of pollutant
discharge emission equivalent, defined as the pollutant discharge divided by the pollutant quantity.
Qi QSO2 QNOx QPM
Ei = = + + (1)
Wi α β γ

where Ei represents the amount of pollutant emission equivalent. Qi indicates the pollutant emission equivalent, and Wi means the
coefficients of the pollutant emission. The pollutants emitted (i) are composed of sulfur dioxide (SO2), nitrogen oxide (NOx) and
particulate matter (PM). The coefficients α, β and γ, respectively, are 0.95, 0.95 and 4, according to Measures on the Administration of
the Collection Standards for Pollutant Discharge Fees issued by State Development Planning Commission of the People’s Republic of
China.
The emission of pollutant and CO2 are standardized using formulas (2) and (3) are attempting to counteract the influence of
dimension, magnitude, and negative orientation.

1
We would like to thank an anonymous reviewer for suggesting this hypothesis to us.

1115
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

maxE − E
u1 = (2)
maxE − minE

maxCO2 − CO2
u2 = (3)
maxCO2 − minCO2

where u1 and u2 represent standardized pollution reduction and low-carbon development (unit: 104 t) respectively.
Tang et al. (2022) show that the reliability and validity of the coupling coordination degree model (CCDM) can be further improved
with the following modifications.
√̅̅̅̅̅̅̅̅̅̅ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
2 u1 u2 u1
c= = [1 − (u2 − u1 )] (4)
u1 + u2 u2

T = αu1 + βu2 = αF(x, t) + βG(x, t) (5)

√̅̅̅̅̅̅̅̅̅̅̅
D= c×T (6)

where c presents the coupling degree of the two systems; T stands for integrated coordination index; α and β represent undetermined
coefficients. Carbon emission reduction and air pollutant control are equally important so that α = β = 0.5, which draws on the
research of (Liu et al., 2021a). D represents the coupling coordination degree of the two systems the range of which is [0,1]. The closer
the D value is to 1 (0), the better (poorer) the coordination between pollution reduction-low carbon emission system (PCS), and so is
the synergistic effect.

3.1.2. Measures of digital financial inclusion

The emergence of digital financial inclusion has impacted global financial inclusion (Song et al., 2021). Junjun (2023) shows that
between 2012 and 2021, China’s digital economy grew from 11 trillion yuan to more than 45 trillion yuan in size, and its share of the
country’s GDP rose from 21.6 % to 39.8 %. Digital or electronic technology, such as mobile payments, online banking, online in­
surance, etc., are used for digital financial inclusion transactions. This technology compensates for traditional finance’s flaws (Tay
et al., 2022). In particular, faster information sharing has the potential to reduce market information asymmetry, financial service
obstacles and transaction costs, resulting in improved access to financial services for vulnerable and financially excluded sectors. Thus,
digital inclusive finance is regarded as a fresh approach to financial development because it is both a digital financial service
development using the most recent information technology (Dai, 2021) and a financial infrastructure that may help underprivileged
groups to meet their funding challenges (Lyons and Kass-Hanna, 2021).
Digital Financial Inclusion (lnfin) is measured as the log of the Digital Inclusive Finance Development Index published by Institute
of Digital Finance of Peking University. The composite index can be further divided into three dimensions: usage depth, coverage
breadth, and digitalization level, has been commonly used as a proxy of financial inclusion in the literature (see, e.g., Song et al. 2021,
Junjun 2023).

3.2. Data source

The data are gathered from 2011 to 2020 at the level of 30 provinces (municipalities), totaling 300 observations; Tibet, Taiwan,
Hong Kong, and Macao were left out due to data availability. Data for each province are mostly gathered from the China Statistical
Yearbook, China Energy Statistical Yearbook, and Peking University’s Digital Financial Inclusion Index (2011–2020).
Table 1 presents description statistics of key variables (see Appendix for descriptions of these variables). The mean value of the
coupling coordination degree of pollution reduction and low-carbon development (D) is 0.816, and the standard deviation is 0.153,
suggesting that the overall distribution is relatively flat. The digital financial inclusion index (lnfin) has a mean value of 5.220, a
standard deviation of 0.665, and the mean and variance of the three sub-indexes are not much different.

Table 1
Descriptive statistics of key variables.
Variable N Mean SD Min Max P25 P50 P75

D 300 0.816 0.153 0.297 0.997 0.774 0.868 0.912

lnfin 300 5.220 0.665 2.993 6.011 5.007 5.412 5.684
lndf 300 5.207 0.622 3.125 6.085 4.935 5.316 5.656
lnwf 300 5.082 0.784 1.910 5.947 4.803 5.291 5.622
lndig 300 5.517 0.672 3.259 6.094 5.454 5.778 5.953
lnenergy 300 9.437 0.647 7.492 10.60 9.048 9.391 9.898
stru 300 0.409 0.0800 0.169 0.548 0.376 0.419 0.464
lninnov 300 2.507 1.008 0.440 4.484 1.762 2.444 3.180
lntrade 300 16.09 2.539 8.709 20.14 14.47 16.39 17.93
lnhuman 300 7.838 0.283 7.094 8.601 7.692 7.799 8.008
infra 300 15.89 4.787 4.110 25.78 13.22 15.47 18.48
policy 300 0.760 0.428 0 1 1 1 1

1116
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

The correlation matrix of the key variables is shown in Table 2 and Tolerance and VIF are shown in Table 3. The correlation co­
efficient between the explanatory variable, the dependent variable, and control variables are low after the correlation test. The average
VIF is 1.510, which is less than 10, indicating that there are no severe collinearity issues with the model.

3.3. Empirical models

In this section, we describe the empirical models used to estimate the effects of digital financial inclusion on the coupling coor­
dination degree between pollution reduction and low-carbon development (CCDPL).

3.3.1. Contemporaneous effects of digital financial inclusion: fixed effects (FE) estimation
To estimate the direct effect of digital financial inclusion on coupling coordination degree, we specify a panel fixed effects
regression model as follows. To eradicate the contemporaneous effect of heteroskedasticity, a portion of the model’s variables are
logarithmically treated.
∑
Dit = β0 + β1 lnfinit + γ it Controlsit + ui + ε1it (7)

where Dit is the coupling coordination degree between pollution reduction and low-carbon development in province’s i in year t, and
the (log of) digital financial inclusion index is denoted by lnfinit in province i in year t. ui denotes provincial fixed effects; and the
random error term is denoted by ε1it . It is obvious that β1 is the key parameter of this paper, and a positive value of β1 is anticipated. It
indicates the decoupling of lnfinit . The key explanatory variable is the digital financial inclusion (lnfin), as proxied by the China Digital
Financial Inclusive Index (Guo et al., 2020). The index was established by the Center for Digital Financial Inclusion of Peking Uni­
versity, which contains one primary index, “Aggregate Digital Financial Inclusion Index” (fin), and three sub-indexes, “Coverage
Breadth” (wf), “Usage Depth” (df), and “Digitization Level” (dig), covering 31 provinces, 337 prefecture-level cities and roughly 2800
counties in China.
Controlsit represent a set of control variable in province i in year t, including total volume of import and export trade (trade) (Tang
et al., 2022), level of human capital (human), level of transport infrastructure (infra) (Zeng and He, 2023), policy of carbon (policy)
(Wang et al., 2022a). Some characteristics of economic development are correlated with digital financial inclusion and may directly
influence the synergistic effect of pollution and carbon emission reduction. For instance, the development of digital financial inclusion
is affected by the level of urban economic development, and high economic growth areas tend to have higher innovation capacity and
production efficiency may directly enhance environmental governance level. Therefore, failing to control for these economic char­
acteristics may result in omitted variable bias. Trade refers to the total import and export of goods (ten thousand Yuan), reflecting the
foreign economic trade condition in the region; human is the average number of college students per 100,000 population, reflecting the
quality of the social labor force; infra is per capita urban road area (square meters), reflecting the industrial development; and policy is a
dummy variable that indicates whether the city is a carbon pilot city or not (1= yes, 0= no), reflecting the value that the government
places on the growth of greener economies in a particular city.

4. Empirical results

4.1. Coupling coordination degree between pollution reduction and low-carbon development

Fig. 1 shows the averages of pollution reduction system and low-carbon development system from 2011 to 2020 and Fig. 2 depicts
CCDPL in China.
As shown in Fig. 1, carbon emissions increase slowly year by year, and pollutant emission equivalent changes little during the 12th
Five-Year Plan period (i.e., 2011–2015) and decreases significantly since 2014. However, in the 13th Five-Year Plan period (i.e.,
2016–2020), the pollutant emission reduction effect is obvious. It shows that since the 12th Five-Year Plan, China’s air pollution
prevention and control has achieved remarkable results, while the reduction of carbon emission has not achieved synchronous effect in
general due to the impact of the carbon reduction policy of "carbon peak", which is similar to the research findings of Jiang et al.,
(2019).

Table 2
Correlation matrix.
D lnfin lntrade lnhuman infra policy

D 1
lnfin 0.171*** 1
lntrade -0.259*** 0.180*** 1
lnhuman 0.159*** 0.378*** 0.610*** 1
infra -0.407*** 0.227*** -0.064 -0.139** 1
policy 0.002 0.479*** 0.222*** 0.282*** 0.125** 1
***
p < 0.01.
**
p < 0.05
*p < 0.1.

1117
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Table 3
Tolerance and variance inflation factor (VIF).
Variable VIF 1/VIF

lnfin 1.540 0.651

lntrade 1.620 0.617
lnhuman 1.910 0.523
infra 1.130 0.885
policy 1.340 0.749
Mean VIF 1.510

Fig. 1. Trend of pollution reduction system and low-carbon development system, 2011–2020.

Fig. 2. Coupling coordination degree (D) between pollution reduction and low-carbon development.

Fig. 2 shows that the average values of coupling coordination degree of carbon emission reduction and air pollutant control in 30
provinces (autonomous regions and municipalities directly under the Central Government) show a stepwise upward trend. The
coupling coordination degree increases from 0.76 to 0.85 from 2011 to 2020, indicating that the synergistic effect of carbon pollution
reduction in general is significantly improved. The interaction between carbon emission reduction and air pollutant control systems is
increasing. Feng et al. (2019) show that the Action Plan for Air Pollution Prevention and Control implemented by China from 2013 to
2017 has played a good role in the governance of air pollutants in various provinces, which also promoted the rise of coupling co­
ordination level to a certain extent.
Figs. 3–5 show the spatial distribution of pollutant emissions, carbon emissions, pollution reduction and carbon reduction coupling
coordination degree and digital financial inclusion sub-index, respectively. It is clear from these figures that there are significant
variations between regions. For example, Figs. 3 and 4 demonstrate that provinces with high pollution and carbon emissions are mostly

1118
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Fig. 3. Spatial distribution of pollutant emissions.

Fig. 4. Spatial distribution of carbon emissions.

1119
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Fig. 5. Spatial distribution of pollution reduction and carbon reduction coupling coordination degree and digital financial inclusion sub-index.

concentrated in the north and northeast, and that emissions are decreasing year by year. South-central provinces have lower pollution
and carbon emissions. Fig. 5 depicts the rising degree of coupling and coordination of pollution reduction and carbon reduction, with
the central and southern regions standing out as the most visible. Simultaneously, all three sub-indices of digital financial inclusion
have risen over time.

4.2. Fixed effects (FE) regression analysis results

Table 4 reports the regression results of the contemporaneous effects of digital financial inclusion. Models (1)–(4) are FE regression
models where control variables are not included. The regression results reveal that digital financial inclusion and its sub-indexes such
as usage depth, coverage breadth and digitization level positively affect coupling coordination between pollution reduction and low-
carbon development and are statistically significant at the 1 % level, indicating that digital financial inclusion has a catalytic effect on
coupling coordination. The results of models (5)–(8) show that there is still a significant positive influence of digital financial inclusion
on coupling coordination even when the control variables are included. In particular, a 1 % increase in digital financial inclusion is
positively associated with 3.1 % of coupling coordination. And one standard deviation (0.67) increase in digital financial inclusion is
associated with 2 % (=0.67×0.031×100 %) increase in coupling coordination. It indicates that the influence of digital inclusive
finance on the coupling coordination between reduce pollution and low-carbon development is significant in both statistical and
economic senses. This empirical evidence suggests that coordinated governance among policymakers is needed. In summary, hy­
pothesis 1a is supported.
The results are consistent with the view that the improvement of the development level of digital inclusive finance can achieve
economic development by reducing financial transaction costs, easing financing constraints, promoting consumption, scientific and
technological innovation, entrepreneurship, etc. (Yao and Yang, 2022). While paying attention to economic development, countries
and enterprises are also paying more and more attention to social responsibility (Mu et al., 2023) to cope with global climate change.
Attention to the connection/interaction between pollution reduction and emission reduction put forward higher requirements for the
current stage of environmental development, to achieve the goal of coordinated development of pollution reduction and emission
reduction.

1120
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Table 4
Contemporaneous effects of digital financial inclusion on coupling coordination degree.
VARIABLES (1) (2) (3) (4) (5) (6) (7) (8)
D D D D D D D D

lnfin 0.039*** 0.031***

(12.58) (6.41)
lnwf 0.033*** 0.025***
(12.31) (5.93)
Df 0.045*** 0.038***
(13.28) (7.34)
Dig 0.033*** 0.019***
(9.95) (4.63)
lntrade 0.014** 0.015*** 0.013** 0.013**
(2.51) (2.65) (2.35) (2.29)
lnhuman 0.051* 0.049 0.051* 0.092***
(1.66) (1.55) (1.73) (3.03)
infra 0.003* 0.003* 0.002 0.004**
(1.85) (1.86) (1.60) (2.55)
policy -0.011 -0.008 -0.015** -0.005
(-1.47) (-1.05) (-1.99) (-0.61)
Controls NO NO NO NO YES YES YES YES
Province FE YES YES YES YES YES YES YES YES
Constant 0.610*** 0.646*** 0.582*** 0.636*** -0.003 0.028 -0.012 -0.274
(37.07) (46.44) (32.85) (34.92) (-0.01) (0.12) (-0.05) (-1.23)
Observations 300 300 300 300 300 300 300 300
Number of prov. 30 30 30 30 30 30 30 30
Adjusted R2 0.300 0.289 0.329 0.188 0.345 0.331 0.371 0.299

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

4.3. Mechanism analyses

To explore the path of overall environmental impacts of digital financial inclusion, this paper identifies some channels through
which digital financial inclusion can affect the coupling coordination degree. They are innovation effect, structural effect, and con­
sumer preference (see Fig. 6).
We expect that the digital financial inclusion promotes technology innovation, which likewise promotes pollution reduction and
carbon reduction. Firstly, through enhancing the allocation of credit resources, consumption, and industrial upgrading, digital
financial inclusion may improve innovation. In fact, the expansion of finance has encouraged the modernization and improvement of
technology innovation. For example, the number of urban patents grew by 5.3 % after the adoption of pertinent laws on funding digital
inclusion, suggesting that any effort to enhance digital inclusion can have a favorable effect on urban innovation (Li and Li, 2022).
Furthermore, upgrading technology innovation increases a synergistic impact of pollution and carbon reduction through the
innovation-driven effect, and enhances the ascent of the manufacturing value chain which in turn has a substantial positive impact on
energy conservation and emission reduction. This is because natural resource efficiency may be improved by technological innovation,
and result in less energy and resource use, lowering both carbon emissions and pollution. Secondly, through technical innovation,
digital inclusive finance promotes the delivery of energy services and improves the operational efficiency of energy institutions,
resulting in energy poverty reduction (Dong et al., 2022). Energy poverty alleviation enhances household welfare by increasing
thermal efficiency and electricity and natural gas consumption, lowering fuel costs and enhancing the cleanliness of energy con­
sumption, and eventually lowering CO2 emissions (Zhao et al., 2021) and air pollution such as PM2.5 (Liu et al., 2019).
We expect that the digital financial inclusion promotes industrial structure, which likewise promotes pollution reduction and
carbon reduction. First, finance encourages the modernization and improvement of industrial structure because the primary driving
force behind industrial change is financial development. Big data and finance have altered how customers, producers, and markets are
organized. The consumer demand (or market) driven model has expedited the trend of industrial transformation while promoting the
rate of social advancement (Bruhn and Love, 2014). In agriculture, inclusive finance can enable farmers to engage in credit despite
income disparities. This may enhance the investment structure, achieve the adjustment of the industrial structure in rural regions, and
direct the flow of financial resources to industries and sectors with high economic returns (Liu et al., 2021a). Yin and Xu (2022) argue
that through raising citizens’ income level, digital inclusive finance promotes consumption. Thus, diversification of consumer demand
and an increase in consumption level will support industrial structure change and provide essential capital/financial resources for
industrial transformation. Second, upgrading industrial structure may increase a synergistic impact of pollution and carbon reduction.
In the process of reducing backward production capacity and introducing high-tech industries, the level of industrial structure has been
optimized. This upgrading can transfer labor, capital, and energy resources from agricultural sector to industrial and services sectors.
For example, green development efficiency has begun to rise to promote healthy economic growth and enhance environmental
governance (Chen et al., 2023; Huang et al., 2023a). Therefore, Zhu et al. (2019) suggest that modifying the industrial structure is one

1121
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Fig. 6. Mechanism analyses.

of the effective ways to achieve green development. Targeted treatment of industrial pollution and improved management of industrial
structure are efficient strategies to enhance pollution reduction and emission reduction in the manufacturing industry.
There are several reasons as to why energy consumption positively affects the relationship between digital financial inclusion and
the coupling coordination degree of pollution reduction and carbon reduction. First, finance encourages the modernization and
improvement of energy consumption (Liu et al., 2019). Based on its inclusive qualities, Wang et al. (2022b) contend that digital finance
can reduce enterprise financing limitations in its early stages of development, enabling the expansion of enterprise investment scale
and thus an increase in overall energy consumption. Xue et al. (2022) argued that by influencing economic growth, R&D spending,
human capital, financial development, energy efficiency, and industrial structure, the development of the digital economy can
encourage the growth of energy consumption scale. Second, upgrading energy consumption can increase a synergistic impact of
pollution and carbon reduction. From the perspective of low-carbon development, renewable energy expansion is a major driver of
CO2 emissions in most Belt and Road countries because the carbon emission reduction effect of growing renewable energy is sub­
stantially greater than that of improved renewable energy technologies (Wang et al., 2021a). For example, Bekhet and Othman (2018)
find that GDP growth from renewable energy has a large detrimental effect on carbon dioxide through their examination of Malaysia’s
economic activity and environmental quality. From the perspective of air quality control, the primary cause of SO2 enhancement is a
rise in total energy consumption, and the most effective way to reduce SO2 emissions is to manage total energy consumption and
production structure (Yang et al., 2016). By altering the structure of energy production, clean energy substitution can provide green
synergies. Energy consumption is expected to positively the relationship between digital financial inclusion and the coupling coor­
dination degree of pollution and carbon reduction.
Following Wen et al. (2022), we specify the mechanism analyses model as follows.
∑
lninnovit = α0 + α1 lnfinit + γit Controlsit + ui + ε2it (8)
∑
struit = θ0 + θ1 lnfinit + γit Controlsit + ui + ε3it (9)
∑
lneneygyit = δ0 + δ1 lnfinit + γit Controlsit + ui + ε4it (10)
∑
Dit = φ0 + φ1 lnfinit + φ2 lninnovit + φ3 struit + φ4 lneneygyit + γ it Controlsit + ui + ε5it (11)

In Eqs. (8)–(11), lninnovit , struit and lneneygyit represent technological innovation, industrial structure, and energy consumption,
respectively (Sun et al., 2023; Zhang et al., 2023). We use energy consumption (104t in standard coal), industrial structure (share of
value-added of secondary industry in GDP) and technological innovation (number of patents granted per capita) as the mechanism
variables in this paper, respectively. The error terms are ε2it , ε3it , ε4it and ε5it , respectively. It is assumed that the error terms follow a
normal distribution and are independent of each other. Eqs. (8)–(10) capture the innovation effect, structural effect, and consumer
preference of digital financial inclusion, and the model that considers both the direct and indirect (i.e., mediating) impacts of digital
financial inclusion is Eq. (11). The mechanism variables’ response to digital financial inclusion is captured by the regression co­
efficients α1 , θ1 , and δ1 . The regression coefficient φ1 shows the direct effect of digital financial inclusion on the degree of coupling
coordination between pollution reduction and low-carbon growth, while φ2 , φ3 , and φ4 show that of the technological innovation,
industrial structure, and energy consumption, respectively. The mechanism analysis of technological innovation, industrial structure,
and energy consumption are partially affected if φ1 is significant.
Table 5 shows the results of our model. Model (1) of Table 5 shows the total effect of digital financial inclusion development on
coupling coordination. The coefficient of lnfin passes the significance test of 1 % and the sign is positive, indicating that the higher the
development degree of digital financial inclusion in each province, the higher the coupling coordination. Models (2)–(4) respectively

1122
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Table 5
Mechanism analyses of the model.
(1) (2) (3) (4) (5)
Variables D lninnov stru lnenergy D

lnfin 0.031*** 0.433*** -0.035*** 0.031*** 0.004

(6.41) (13.25) (-10.06) (2.98) (0.61)
lninnov 0.030***
(3.55)
stru -0.500***
(-6.47)
lnenergy -0.107***
(-4.04)
lntrade 0.014** -0.190*** -0.005 -0.032*** 0.013**
(2.51) (-5.07) (-1.36) (-2.74) (2.59)
lnhuman 0.051* 1.480*** -0.024 0.168** 0.012
(1.66) (7.08) (-1.07) (2.56) (0.42)
infra 0.003* 0.013 -0.001 0.012*** 0.003**
(1.85) (1.27) (-1.29) (3.59) (2.14)
policy -0.011 0.088* -0.013** 0.008 -0.019***
(-1.47) (1.73) (-2.48) (0.53) (-2.89)
Province FE -0.003 -8.577*** 0.897*** 8.286*** 1.595***
(-0.01) (-5.62) (5.54) (17.27) (4.93)
Observations 298 300 300 300 298
Number of prov. 30 30 30 30 30
Adjusted R2 0.345 0.775 0.569 0.319 0.484

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

show that the effects on the three mediating variables (i.e., lninnov, stru and lnenergy) are statistically significant. In Model (5), the
regression coefficient of lnfin is not significant, indicating that after controlling for its indirect effects via lninnov, stru and lnenergy,
digital financial inclusion has no direct influence on the coupling coordination degree.
Taking a closer look at models (1), (2) and (5) reveals that digital financial inclusion is conducive to promoting technological
innovation and has the impact on innovation. The innovation effect size was 0.77 (0.433×0.030/(0.004+0.433×0.030)). Therefore,
digital financial inclusion improves coupling coordination through innovation effect.
Model (3) suggests that digital financial inclusion significantly reduce the proportion of the added value of the secondary industry
in the gross regional product and improve indirectly the coupling coordination by optimizing the structural effect path of the industrial
structure. The industrial structure effect size was 0.81*(-0.035*(-0.500)/(0.004+(-0.035)*(-0.500)) and statistically significant.
Model (4) shows that the development of digital financial inclusion significantly increases energy consumption, which in turn
reduces CCDPL. The signs of the regression coefficient in model (1) are positive (0.031) and the regression coefficient product in
models (4) and (5) are negative (0.031*(-0.107)). However, the decrease in the CCDPL caused by increased energy consumption in
models (4) and (5) is offset by the increase in the CCDPL caused by the development of digital inclusive finance in model (1), resulting
in an overall (net) effect being a positive one. This is consistent with the view that the development of digital financial inclusion can
help firms reduce unnecessary economic activities and increase the energy consumption selectivity of economic entities.
To sum up, the results (1)–(5) of Table 5 show that digital financial inclusion has a complete effect on the coupling coordination
degree: namely, the development of digital financial inclusion mainly affects the coupling coordination through three paths: tech­
nological progress, industrial structure and energy consumption.2

4.4. Robustness test

4.4.1. Alternative measures of digital financial inclusion

We employ alternative measure of the key explanatory variable as robustness check. Following Li et al. (2022), we replace lnfin by
its sub index called coverage breadth (lnwf). Table 6 reports that coverage breadth positively affects coupling coordination degree at
the 1 % significance level, indicating that the sub index has a direct synergistic effect via the innovation effect, structural effect, and
consumer preference. Therefore, the test results confirm that the main findings are robust.

4.4.2. Alternative measures of provincial level data – city level data

To check whether the main result is sensitive to the use of data at provincial level, we replace the provincial data with the city level
data and re-do the main regression. The results are shown in Table 7. Table 7 shows that the results are similar, suggesting that the
results remain robust.

2
The result still stands when lags in mechanism and independent variable are included.

1123
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Table 6
Robustness test of the model.
(1) (2) (3) (4) (5)
Variables D lnenergy stru lninnov D

lnwf 0.025*** 0.027*** -0.029*** 0.368*** 0.002

(5.93) (2.98) (-9.36) (12.69) (0.35)
lnenergy -0.108***
(-4.05)
stru -0.507***
(-6.65)
lninnov 0.031***
(3.72)
lntrade 0.015*** -0.031*** -0.006 -0.176*** 0.014***
(2.65) (-2.66) (-1.60) (-4.62) (2.64)
lnhuman 0.049 0.160** -0.020 1.399*** 0.012
(1.55) (2.39) (-0.86) (6.46) (0.41)
infra 0.003* 0.012*** -0.001 0.013 0.003**
(1.86) (3.56) (-1.29) (1.20) (2.15)
policy -0.008 0.010 -0.017*** 0.124** -0.019***
(-1.05) (0.67) (-3.10) (2.44) (-2.85)
Province FE YES YES YES YES YES
Constant 0.028 8.357*** 0.851*** -7.791*** 1.605***
(0.12) (17.08) (5.06) (-4.92) (4.96)
Observations 300 300 300 300 300
Number of prov. 30 30 30 30 30
Adjusted R2 0.331 0.319 0.552 0.767 0.484

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

Table 7
Robustness test using city level data.
(1) (2) (3) (4) (5) (6) (7) (8)
Variables D D D D D D D D

lnfin 0.009*** 0.009***

(11.19) (10.07)
lnwf 0.008*** 0.008***
(10.22) (8.99)
lndf 0.009*** 0.010***
(11.56) (10.57)
lndig 0.007*** 0.007***
(11.25) (10.11)
lntrade 0.001 0.001 0.001 0.001*
(1.12) (1.38) (1.16) (1.81)
lnhuman -0.000 -0.000 -0.000 -0.000
(-0.51) (-0.10) (-0.90) (-0.99)
infra 0.000* 0.000 0.000** 0.000
(1.72) (1.56) (2.31) (0.98)
policy -0.002 -0.001 -0.002* -0.002
(-1.44) (-0.59) (-1.70) (-1.24)
Controls NO NO NO NO YES YES YES YES
City FE YES YES YES YES YES YES YES YES
Constant 0.908*** 0.914*** 0.906*** 0.916*** 0.894*** 0.897*** 0.891*** 0.896***
(223.11) (234.92) (219.83) (273.76) (75.19) (75.24) (74.79) (75.64)
Observations 2,850 2,850 2,850 2,850 2,810 2,810 2,810 2,810
Adjusted R2 0.943 0.942 0.944 0.942 0.943 0.942 0.945 0.942

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

4.4.3. Addressing endogeneity concerns

4.4.3.1. Instrumental variable (IV) estimation. Endogeneity exists owing to the impact of reverse causality and/or omitted factors on
the outcomes. For example, by increasing demand for green credit, the rise of digital financial inclusion may strengthen the coupling

1124
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

and coordination of pollution reduction and carbon reduction initiatives. In turn, the state’s encouragement and support for integrated
development policies that aim at reducing pollution and carbon emissions may encourage consumers to utilize digital financial in­
clusion. As a result, there may be some reverse causality issue. The instrumental variable technique is used to mitigate the endogeneity
problem.
We utilize the Geographic Information System (GIS) to calculate the geographic distance from each province capital city to
Hangzhou. Following Zhang et al. (2020), we create an interaction variable between the geographical distance from each provincial
capital city to Hangzhou and the national average value of the digital financial development index (except for the province) and use it
as an instrumental variable with temporal effects. As we know, the development of digital finance represented by Alipay originated in
Hangzhou. The closer it is geographically to Hangzhou, the better the growth of digital inclusive finance should be. As the GIS interacts
with the averaged digital financial development index, it cannot only anticipate the degree of growth of digital financial inclusion but
also be exogenous in CCDPL. Thus, the IV influences economic development, but it remains constant as the economy grows. In
addition, the IV is not only directly related to the development of digital financial inclusion, but also does not directly affect the level of
CCDPL in various places. Therefore, using this distance-based instrumental variable can better isolate the impact only caused by digital
financial inclusion.
∑
lnfinit = α + βGeom + γ it Controlsit + ui + εit (12)

∑
Dit = α + βln fin
̂it + γit Controlsit + ui + εit (13)

In the first stage, the impact of GIS interacted with the averaged digital financial development index (Geom) on digital financial
inclusion is first evaluated (see Eq. (12)). In the second stage, CCDPL is regressed with the estimated value of digital inclusive finance
(ln fin
̂it ). Following these regressions, a LIML weak instrument test is run to assess the instrumental variable’s quality.
To facilitate comparison, we report the results of ordinary least squares (OLS), fixed-effects (FE) regression and IV regression in
Table 8, respectively. The instrumental variable is the interaction between the geographical distance from each provincial capital city
to Hangzhou and the national average digital financial development index (except for the province). Columns (3) and (4) of Table 8
give the results of the reduced form and the first stage regression, indicating that IV has a significant effect on D through lnfin. The first
stage F statistic is larger than 10, which suggests that the instrumental variable does not suffer from any weak instrument problem.
Column (5) contains the results of the second stage of the IV regression, and the coefficient of lnfin is positive and statistically sig­
nificant. Comparing the coefficients of the explanatory variables in columns (2) and (5), we find that the estimated coefficients of
digital financial inclusion are not significantly different from each other, indicating that the endogeneity issue is not severe. In short,
the main findings of the baseline regression still hold after controlling for endogeneity.

4.4.3.2. Two-stage least squares (2sls) – Lewbel (2012) approach. Since the development level of digital financial inclusion is taken into
consideration when calculating the average of the National Digital Finance Development Index, one could argue that the IV (GIS
interacted with the averaged digital financial development index) utilized in the previous section may not be appropriate. Thus, we

Table 8
Endogeneity concern: IV regression results.
(1) (2) (3) (4) (5) (6)
OLS FE Reduced Form First Stage Second Stage Liml
VARIABLES D D D lnfin D D

lnfin 0.061*** 0.031*** 0.033*** 0.033***

(4.35) (4.67) (4.90) (6.30)
lnGeom 0.032*** 0.991***
(4.76) (25.25)
lntrade -0.032*** 0.014 0.012 -0.048 0.014 0.014***
(-11.06) (1.46) (1.29) (-1.54) (1.52) (2.76)
lnhuman 0.175*** 0.048 0.056 0.408** 0.043 0.043
(6.84) (0.76) (0.88) (2.06) (0.69) (1.32)
infra -0.015*** 0.003 0.003 0.008 0.003 0.003**
(-8.41) (1.20) (1.25) (1.19) (1.18) (2.05)
policy -0.015 -0.009 -0.011 -0.020 -0.010 -0.010
(-0.75) (-0.90) (-1.06) (-0.59) (-1.03) (-1.62)
Constant -0.119 0.018 -0.242 -8.956*** 0.130 0.130
(-0.67) (0.04) (-0.48) (-6.75) (0.24) (0.47)
Observations 300 300 300 300 300 300
R2 0.415 0.412 0.413 0.961 0.955 0.955
No. of provinces 30 30 30 30 30 30
F value 37.4

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05
*p < 0.1.

1125
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

also add an alternative 2SLS analysis using the Lewbel (2012) method, which can address the endogeneity issue when conventional
identifying information (like external instruments) is absent or unavailable, in order to further guarantee that potential endogeneity is
controlled for. The Lewbel (2012) approach has been used in a number of recent investigations which is developed to address the
endogeneity concern in the absence of traditional identifying information such as external instruments or when no other such in­
formation is available. This method exploits model heteroskedasticity to construct instruments using the available regressors and
identification can be achieved by having regressors that are uncorrelated with the product of heteroskedastic errors—a feature of many
models where error correlations are due to an unobserved common factor.
Table 9 shows the estimation results. First, the Pagan-Hall general test statistic is statistically significant, indicating that hetero­
scedasticity exists in the error term. Therefore, the Lewbel (2012) approach is appropriate to mitigate the endogeneity concern.
Second, we generate four IVs based on Lewbel (2012) method and they are just the product term of four control variables and the
(estimated) error term. They are iv_lnfin_lntrade_g, iv_lnfin_lnhuman_g, iv_lnfin_infra_g and iv_lnfin_policy_g. To reduce the over­
identification problem, we select only three of them (i.e., iv_lnfin_lntrade_g, iv_lnfin_infra_g and iv_lnfin_policy_g) and apply the standard IV
estimation method with these three selected instruments. The estimated results of the 2SLS regression in Table 2 suggest that digital
financial inclusion and its sub-indices (lnfin and lndig) has a positive correlation with coupling coordination (i.e., CCDPL), which is
consistent with our baseline results. The statistic of Kleibergen-Paap rk LM is significant, therefore, the selected IVs are relevant. The
weak identification test (Kleibergen-Paap rk Wald F statistic) exceeds the cut-off values of Stock-Yogo 10 % maximal IV relative bias,
therefore, the selected IVs are associated with our endogenous regression factor. In addition, the p-value of Hansen’s J
over-identification test statistic is large, indicating that the selected IVs do not suffer any over-identification issues. In summary, we
confirm that the positive correlation between digital financial inclusion and CCDPL is still strong after controlling for the endogeneity
problem.

4.5. Heterogeneity analysis

4.5.1. Regional difference: Eastern, Central, Western and Northeast Region

As there are differences in the development of different regions of China, their impact on how digital financial inclusion on the level
of CCDPL may also be different. To explore this possibility, the sample cities were further classified into four regions, namely the
eastern, central, western, and northeastern regions, according to the classification standards of the National Bureau of Statistics, and
the differences in the influence of digital inclusion finance on the synergistic effect of pollution reduction and carbon reduction in each
region are explored by sub-sample regression.3
In Table 10, the results of models (1)–(4) show that the development of digital financial inclusion has a significant impact on the
synergistic effect of pollution reduction and carbon reduction only in the central and western regions but not in other regions. This
indicates that the impact of the development of digital financial inclusion on coupling coordination is most obvious in the central and
western regions. The evidence that the impact of inclusive finance development on CCDPL in the eastern and northeastern regions
cannot pass the significance test demonstrates that as pollution and carbon reduction technology advances and industrial structures
improve, the synergistic impacts of pollution and carbon reduction in the eastern and northeastern regions will be rarely altered by
digital inclusive financial policies. The overall results are consistent with the view that promoting the digitalization of inclusive finance
in cities of the central region will help improve CCDPL, narrowing the urban-rural gap and promoting common prosperity. Since the
economic development of the western region is lagging the central region, the western region can promote digital inclusive finance,
focus on optimizing the industrial structure, eliminate backward production capacity, and further improve the synergy of local
pollution reduction and carbon reduction.
The robustness of the above results is further examined. As before, we replace lnfin by its sub-index called coverage breadth (wf) for
sub-sample regression, and the results of models (5)–(8) of Table 10 show similar results. The coefficient estimates of key variables do
not change significantly, indicating that the estimation results are robust.

4.5.2. High vs low coupling coordination degree

We also examine whether the influence of digital financial inclusion depends on different degrees of coordination degree by
subgroup analysis. The mean value of the coupling coordination degree of pollution reduction and carbon reduction is 0.815. A
province is classified as provinces of high (low) coordination degree if its mean value is higher (lower) than the mean value. The results
are shown in Table 11. They reveal that digital inclusion finance has a significant positive impact on provinces with high coordination
degree and provinces with low coordination degree. However, the impact of digital inclusion finance on provinces with low coordi­
nation degree (0.044) is greater than that of provinces with high coordination degree (0.012). This demonstrates that for provinces
with low coordination, the higher the development of digital inclusive finance, the greater the influence on strengthening the prov­
ince’s coupling coordination of pollution reduction and carbon reduction.
To check if the above results is robust, we replace the core explanatory variable (lnfin) by the coverage span of the digital financial
inclusion sub-index (wf) for subsample regression. Table 11 indicates that the coefficient estimates of the key variables do not change
significantly and the results are consistent with that of the baseline model, indicating that the main estimation results are robust.

3
Specifically, the eastern part includes Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong and Hainan. The
central part covers Shanxi, Anhui, Jiangxi, Henan, Hubei and Hunan. The western part includes Inner Mongolia, Guangxi, Chongqing, Sichuan,
Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia and Xinjiang. The northeast covers Liaoning, Jilin and Heilongjiang.

1126
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Table 9
Endogeneity concern: two-stage least-squares estimation using Lewbel approach.
(1) (2)
VARIABLES D D

lnfin 0.932*
(1.79)
lndig 1.352***
(3.69)
lnhuman 4.962*** 4.963***
(3.50) (3.73)
infra -0.026 -0.031
(-0.34) (-0.40)
policy -0.203 -0.414
(-0.31) (-0.64)
Constant -22.238* -29.469***
(-1.90) (-2.74)
Controls YES YES
Province FE YES YES
Adjusted R2 0.244 0.154
IV heteroskedasticity test:
Pagan-Hall general test statistic 13.638 15.755
p-value 0.0340 0.0151
Underidentification test:
Kleibergen-Paap rk LM statistic 7.569 8.480
p-value 0.0558 0.0371
Weak identification test:
Kleibergen-Paap rk Wald F statistic 18.082 27.296
Stock-Yogo (2005) 10 % maximal IV relative bias 9.08 9.08
Overidentification test:
Hansen J statistic 4.371 3.327
p-value 0.1124 0.1895

Robust t-statistics in parentheses.

***
p < 0.01,
**p < 0.05,.
*
p < 0.1.

Table 10
Eastern, Central, Western and Northeast Region perspectives.
Dependent Variable: D Heterogeneity Analysis Robustness Tests

(1) (2) (3) (4) (5) (6) (7) (8)

Eastern Central Western Northeast Eastern Central Western Northeast

lnfin 0.035*** 0.001 0.019*** 0.008

(3.92) (0.09) (3.38) (0.72)
lnwf 0.037*** -0.003 0.014*** 0.009
(4.25) (-0.25) (3.15) (0.75)
lntrade -0.069*** 0.125*** 0.017*** -0.043 -0.060** 0.128*** 0.018*** -0.044
(-2.90) (4.57) (3.61) (-1.29) (-2.52) (4.67) (3.81) (-1.33)
lnhuman 0.076 0.157* -0.068* -0.108 0.059 0.166* -0.070* -0.108
(1.41) (1.91) (-1.90) (-1.03) (1.07) (1.98) (-1.91) (-1.02)
infra 0.002 0.009 0.007*** 0.021*** 0.002 0.009 0.007*** 0.020***
(0.82) (1.50) (3.89) (3.10) (0.89) (1.61) (4.00) (2.93)
policy -0.024* -0.063*** -0.007 0.010 -0.024* -0.064*** -0.005 0.010
(-1.95) (-3.09) (-0.78) (0.57) (-1.98) (-3.12) (-0.55) (0.56)
Provence YES YES YES YES YES YES YES YES
Constant 0.906 -5.208*** 0.119 0.731 0.858 -5.376*** 0.098 0.743
(0.64) (-5.13) (0.38) (0.52) (0.62) (-5.12) (0.30) (0.53)
Observations 100 60 110 30 100 60 110 30
Number of prov. 10 6 11 3 10 6 11 3
Adjusted R2 0.450 0.616 0.401 0.595 0.465 0.617 0.393 0.596

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

1127
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Table 11
High vs low coupling coordination degree.
Dependent Variable: D Heterogeneity Analysis Robustness Tests

(1) (2) (3) (4)

High Low High Low

lnfin 0.012*** 0.044***

(4.75) (3.12)
lnwf 0.010*** 0.044***
(4.50) (3.46)
lntrade 0.011*** 0.007 0.012*** 0.004
(4.01) (0.43) (4.12) (0.24)
lnhuman 0.045*** 0.098 0.043** 0.073
(2.64) (1.25) (2.46) (0.93)
infra 0.001 0.012*** 0.001 0.011**
(1.06) (2.86) (1.09) (2.58)
policy 0.006* -0.038* 0.008** -0.036*
(1.66) (-1.94) (2.06) (-1.89)
Provence YES YES YES YES
Constant 0.298** -0.642 0.317*** -0.374
(2.55) (-0.95) (2.65) (-0.54)
Observations 196 104 196 104
Number of prov. 23 14 23 14
Adjusted R2 0.480 0.438 0.474 0.451

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

4.5.3. The commencement of digital financial inclusion policy

“The development of inclusive finance" was officially proposed in the Decision of the Central Committee of the Communist Party of
China on Several Major Issues Concerning Comprehensively Deepening Reform and adopted by the Third Plenary Session of the 12th
Central Committee of the Communist Party of China in November 2013. This is the first time that the concept of "financial inclusion"
has been formally established as a national strategy. As the development of inclusive finance brings about new financial market and
product innovations, micro, small and medium-sized enterprises with large pollution and large emissions now have an opportunity to
participate in the governance of pollution reduction and carbon reduction. This section explores whether this policy commencement
can enhance synergies in pollution and carbon reduction. We create a grouping variable which is set to 1 after the policy was

Table 12
Heterogeneity analysis - The commencement of digital financial inclusion policy.
Dependent Variable: D Heterogeneity Analysis Robustness Tests

(1) (2) (3) (4)

After Before After Before

lnfin 0.113*** 0.005

(8.34) (1.09)
lnwf 0.112*** 0.003
(7.95) (0.72)
lntrade 0.008 0.012 0.007 0.012
(1.14) (1.35) (0.94) (1.43)
lnhuman -0.051 0.015 -0.068* 0.033
(-1.46) (0.20) (-1.85) (0.45)
infra 0.001 -0.000 0.001 0.000
(0.79) (-0.13) (0.39) (0.07)
policy -0.020** 0.008 -0.023** 0.008
(-2.29) (1.24) (-2.57) (1.34)
Constant 0.472* 0.456 0.659** 0.308
(1.88) (0.87) (2.49) (0.60)
Controls YES YES YES YES
Province FE YES YES YES YES
Observations 210 90 210 90
Number of ID 30 30 30 30
Adjusted R2 0.322 -0.277 0.303 -0.293

Robust t-statistics in parentheses.

***
p < 0.01.
**
p < 0.05.
*
p < 0.1.

1128
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

introduced (i.e., after 2013) and to zero if otherwise and estimate the baseline regression for the two subgroups. The results in Table 12
show that the effect of the policy on the coupling coordination degree is significant at the 1 % level only after (but not before) the policy
commencement.
The results are similar if the core explanatory variable is replaced by the coverage span of the digital financial inclusion sub-index
(wf) (see Columns (3) and (4), Table 12), indicating that the estimation results are robust.

5. Conclusion

This paper examines how digital financial inclusion affects the coupling coordination degree between pollution reduction and low-
carbon development. Using a panel data sample of 30 Chinese provinces from 2011 to 2020, we find that the CCDPL in China improves
significantly during the sample period. The synergistic effect of different economic zones and provinces varies greatly in space. Digital
financial inclusion has a promoting effect on the coupling coordination degree of pollution and carbon emission reduction. Techno­
logical innovation, industrial structure and energy consumption are the transmission channels through which digital financial in­
clusion influences the synergistic effect of pollution and carbon emission reduction. The findings of this paper shed light on the
importance of considering the coupling coordination in developing economic policies related to pollution reduction and low-carbon
development and the role of digital financial inclusion in dealing with the coupling coordination.

CRediT authorship contribution statement

Yuxi Zhang: Writing – review & editing, Writing – original draft, Software, Methodology, Investigation, Formal analysis, Data
curation, Conceptualization. Adrian (Wai Kong) Cheung: Writing – review & editing, Writing – original draft, Supervision, Meth­
odology, Conceptualization. Xiaodong Qu: Software, Methodology, Investigation, Formal analysis, Data curation.

Declaration of competing interest

None.

Appendix

Details of the variables constructed for analyses.

Types Symbols Variable Name Variable Definition

Dependent variable D Synergies of pollution and carbon the coupling coordination degree between pollution reduction and low-carbon
reduction development
Explanatory lnfin Digital financial inclusion level Log of aggregate digital financial inclusion index
variable df Usage depth sub index
lnwf Coverage breadth sub index
dig Digitization level sub index
Mediating variables lnenergy energy consumption Log of total regional energy consumption, 104 t (in standard coal)
stru industrial structure Share of value-added of secondary industry in GDP (%)
lninnov technological innovation Log of number of patents granted per capita (person/piece)
Control variables lntrade total volume of import and export trade Log of total import and export of goods (ten thousand Yuan)
lnhuman level of human capital Log of average number of college students per 100,000 population
infra level of transport infrastructure Log of per capita urban road area (square meters)
policy policy of carbon Carbon pilot city or not (1= yes, 0= no)

References

Barrington-Leigh, C., Baumgartner, J., Carter, E., Robinson, B.E., Tao, S., Zhang, Y., 2019. An evaluation of air quality, home heating and well-being under Beijing’s
programme to eliminate household coal use. Nat. Energy 4, 416–423.
Baulch, B., Do, T.D., Le, T.H., 2018. Constraints to the uptake of solar home systems in Ho Chi Minh City and some proposals for improvement. Renew. Energy 118,
245–256.
Bekhet, H.A., Othman, N.S., 2018. The role of renewable energy to validate dynamic interaction between CO2 emissions and GDP toward sustainable development in
Malaysia. Energy Econ. 72, 47–61.
Bruhn, M., Love, I., 2014. The real impact of improved access to finance: evidence from Mexico. J. Finance 69, 1347–1376.
Bui, D.T., 2020. Transmission channels between financial development and CO2 emissions: a global perspective. Heliyon 6, e05509.
Chen, Y., Wang, L., Ye, Y., Tao, Y., 2023. Digital M&A and firm productivity in China. Finance Res. Lett. 58, 104326.
Costello, A., Abbas, M., Allen, A., Ball, S., Bell, S., Bellamy, R., Friel, S., Groce, N., Johnson, A., Kett, M., 2009. Managing the health effects of climate change: lancet
and University College London Institute for Global Health Commission. Lancet 373, 1693–1733.
Dai, W., 2021. Development and supervision of robo-advisors under digital financial inclusion in complex systems. Complexity 2021, 1–12.

1129
Y. Zhang et al. Economic Analysis and Policy 82 (2024) 1113–1130

Dong, K., Taghizadeh-Hesary, F., Zhao, J., 2022. How inclusive financial development eradicates energy poverty in China? The role of technological innovation.
Energy Econ. 109, 106007.
Feng, Y., Ning, M., Lei, Y., Sun, Y., Liu, W., Wang, J., 2019. Defending blue sky in China: effectiveness of the “Air pollution prevention and control action plan” on air
quality improvements from 2013 to 2017. J. Environ. Manag. 252, 109603.
Guo, F., Wang, J., Wang, F., Kong, T., Zhang, X., Cheng, Z., 2020. Measuring China’s digital financial inclusion: index compilation and spatial characteristics. China
Econ. Q. 19, 1401–1418.
Huang, Z., Tao, Y., Luo, X., Ye, Y., Lei, T., 2023a. Regional digital finance and corporate investment efficiency in China. Appl. Econ. 55, 5115–5134.
Huang, Z., Tao, Y., Zhang, Q., Ye, Y., 2023b. The road to entrepreneurship: the effect of China’s broadband infrastructure construction. Econ. Anal. Policy 80,
1831–1847.
Jiang, J., Ye, B., Liu, J., 2019. Peak of CO2 emissions in various sectors and provinces of China: recent progress and avenues for further research. Renew. Sustain.
Energy Rev. 112, 813–833.
Junjun, H.X.L., 2023. Research on the carbon emission reduction effect of digital economy development under the background of "Double carbon". J. Lanzhou Univ.
Finance Econ. 1–15.
Le, T.H., Le, H.C., Taghizadeh-Hesary, F., 2020. Does financial inclusion impact CO2 emissions? Evidence from Asia. Finance Res. Lett. 34, 101451.
Lewbel, A., 2012. Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models. J. Bus. Econ. Stat. 30, 67–80.
Li, G., Wu, H., Jiang, J., Zong, Q., 2023. Digital finance and the low-carbon energy transition (LCET) from the perspective of capital-biased technical progress. Energy
Econ. 120, 106623.
Li, J., Dong, X., Dong, K., 2022. How much does financial inclusion contribute to renewable energy growth? Ways to realize green finance in China. Renew. Energy
198, 760–771.
Li, J., Li, B., 2022. Digital inclusive finance and urban innovation: evidence from China. Rev. Dev. Econ. 26, 1010–1034.
Li, X., Ye, Y., Liu, Z., Tao, Y., Jiang, J., 2024. FinTech and SME’performance: evidence from China. Econ. Anal. Policy 81, 670–682.
Liu, G., Fang, H., Gong, X., Wang, F., 2021a. Inclusive finance, industrial structure upgrading and farmers’ income: empirical analysis based on provincial panel data
in China. PLoS One 16, e0258860.
Liu, T.L., Song, Q.J., Lu, J., Qi, Y., 2021b. An integrated approach to evaluating the coupling coordination degree between low-carbon development and air quality in
Chinese cities. Adv. Clim. Change Res. 12, 710–722.
Liu, Y., Zhang, J., Zhu, Z., Zhao, G., 2019. Impacts of the 3E (economy, energy and environment) coordinated development on energy mix in China: the multi-
objective optimisation perspective. Struct. Change Econ. Dyn. 50, 56–64.
Lu, Z., Wu, J., Li, H., Nguyen, D.K., 2022. Local bank, digital financial inclusion and SME financing constraints: empirical evidence from China. Emerg. Mark. Finance
Trade 58, 1712–1725.
Lyons, A.C., Kass-Hanna, J., 2021. Financial inclusion, financial literacy and economically vulnerable populations in the Middle East and North Africa. Emerg. Mark.
Finance Trade 57, 2699–2738.
Milliman, S.R., Prince, R., 1989. Firm incentives to promote technological change in pollution control. J. Environ. Econ. Manag. 17, 247–265.
Mu, W., Liu, K., Tao, Y., Ye, Y., 2023. Digital finance and corporate ESG. Finance Res. Lett. 51, 103426.
Ozili, P.K., 2018. Impact of digital finance on financial inclusion and stability. Borsa Istanb. Rev. 18, 329–340.
Pacca, L., Antonarakis, A., Schröder, P., Antoniades, A., 2020. The effect of financial crises on air pollutant emissions: an assessment of the short vs. medium-term
effects. Sci. Total Environ. 698, 133614.
Ramanathan, V., Feng, Y., 2009. Air pollution, greenhouse gases and climate change: global and regional perspectives. Atmos. Environ. 43, 37–50.
Reichler, T., Kim, J., 2008. How well do coupled models simulate today’s climate? Bull. Am. Meteorol. Soc. 89, 303–312.
Shafiei, S., Salim, R.A., 2014. Non-renewable and renewable energy consumption and CO2 emissions in OECD countries: a comparative analysis. Energy Policy 66,
547–556.
Shahbaz, M., Li, J., Dong, X., Dong, K., 2022. How financial inclusion affects the collaborative reduction of pollutant and carbon emissions: the case of China. Energy
Econ. 107, 105847.
Song, H., Li, M., Yu, K., 2021. Big data analytics in digital platforms: how do financial service providers customise supply chain finance? Int. J. Oper. Prod. Manag. 41
(4), 410–435.
Song, Q., Zhou, N., Liu, T., Siehr, S.A., Qi, Y., 2018. Investigation of a “coupling model” of coordination between low-carbon development and urbanization in China.
Energy Policy 121, 346–354.
Sun, Z., Cao, C., He, Z., Feng, C., 2023. Examining the coupling coordination relationship between digital inclusive finance and technological innovation from a spatial
spillover perspective: evidence from China. Emerg. Mark. Finance and Trade 59, 1219–1231.
Tang, X., Zhang, Y., Cao, L., 2022. Spatio-temporal characteristics and influencing mechanism of synergistic effect of pollution and carbon emission reduction in
China. Res. Environ. Sci. 35, 2252–2263.
Tao, Y., Lu, H., Ye, Y., Wu, H., 2023. Does the firms’ digital transformation drive environmental innovation in China? Sustain. Dev. 1–14.
Tay, L.Y., Tai, H.T., Tan, G.S., 2022. Digital financial inclusion: a gateway to sustainable development. Heliyon 8, e09766.
Wang, H., Gu, K., Dong, F., Sun, H., 2022a. Does the low-carbon city pilot policy achieve the synergistic effect of pollution and carbon reduction? Energy Environ. 35
(2), 569–596.
Wang, J., Dong, X., Dong, K., 2021a. How renewable energy reduces CO2 emissions? Decoupling and decomposition analysis for 25 countries along the Belt and Road.
Appl. Econ. 53, 4597–4613.
Wang, J., Wang, J., Wang, Y., 2022b. How does digital finance affect the carbon intensity of the manufacturing industry. China Popul. Environ. 32 (7), 1–11.
Wang, W., Sun, X., Zhang, M., 2021b. Does the central environmental inspection effectively improve air pollution?-An empirical study of 290 prefecture-level cities in
China. J. Environ. Manag. 286, 112274.
Wen, Z., Fang, J., Xie, J., Ouyang, J., 2022. Methodological research on mediation effects in China’s mainland. Adv. Psychol. Sci. 30, 1692.
Xue, L., Zhang, X., 2022. Can digital financial inclusion promote green innovation in heavily polluting companies? Int. J. Environ. Res. Public Health 19, 7323.
Yang, X., Wang, S., Zhang, W., Li, J., Zou, Y., 2016. Impacts of energy consumption, energy structure, and treatment technology on SO2 emissions: a multi-scale LMDI
decomposition analysis in China. Appl. Energy 184, 714–726.
Yao, L., Yang, X., 2022. Can digital finance boost SME innovation by easing financing constraints?: evidence from Chinese GEM-listed companies. PLoS One 17,
e0264647.
Yin, X., Xu, Z., 2022. An empirical analysis of the coupling and coordinative development of China’s green finance and economic growth. Resour. Policy. 75, 102476.
Zeng, Q.H., He, L.Y., 2023. Study on the synergistic effect of air pollution prevention and carbon emission reduction in the context of" dual carbon": evidence from
China’s transport sector. Energy Policy 173, 113370.
Zhang, R., Wu, K., Cao, Y., Sun, H., 2023. Digital inclusive finance and consumption-based embodied carbon emissions: a dual perspective of consumption and
industry upgrading. J. Environ. Manag. 325, 116632.
Zhang, X., Yang, T., Wang, C., Wan, G.H., 2020. Digital Financial Development and Household Consumption Growth: Theory and China’s Practice. Manag. World 36,
48–62.
Zhao, J., Jiang, Q., Dong, X., Dong, K., 2021. Assessing energy poverty and its effect on CO2 emissions: the case of China. Energy Econ. 97, 105191.
Zhou, K., Wang, Q., Tao, Y., Li, X., 2024. Information infrastructure construction and firm export performance in China. Res. Int. Bus. Finance 70, 102311.
Zhu, B., Zhang, M., Zhou, Y., Wang, P., Sheng, J., He, K., Wei, Y.M., Xie, R., 2019. Exploring the effect of industrial structure adjustment on interprovincial green
development efficiency in China: a novel integrated approach. Energy Policy 134, 110946.

1130